A detailed review of block based techniques in image copy move forgery detection ABSTRACT Digital images in every ones life depict very important role in various domains such as daily newspapers, magazines, multimedia, medical diagnosis, legal document and evidence in court.
In todays digital world image forensics is extremely demanding field as images are the key source of communication, fastest means to transfer and exchange of the information but now truthfulness of images are easily lost by using freely available editing software like photo shop, GIMP, Paint.NET which manipulate the content and deny the authenticity and integrity of images. So it has become a difficult task to determine an original image or a forged image. There are many types of image forgery but were focusing on Copy-Move forgery which is the most common type of image forgery, where one region of the image is copied and pasted to another region of the image such that it depicts as an original image.
This paper provides a literature survey of different block based copy-move forgery detection techniques. Keywords image forgery image forgery detection copy-move forgery copy-move forgery detection image splicing image retouching copy move forgery methods block based methods 1. INTRODUCTION Digital Forensics is a branch of Computer Science deals with the investigation, verification and recovery of digital data and evidences found in digital devices such as computers, digital cameras or video recorders.
Digital images and videos act as the major sources of evidence towards any event or crime, 1 specifically in the legal industry as well as media and broadcast industries. Image forgery or image tampering is used in various areas such as surveillance systems, multimedia security, journalism, and scientific publications Such industries are the major application domains of digital image forensics. The present day wide availability of a large number of lowcost multimedia file processing tools, techniques and software, having numerous advanced features, has made editing and manipulation of multimedia data extremely easy44. All types of unauthorized modification or tampering of digital multimedia data pose as threats to their integrity. With the vast availability of digital image processing tools and techniques today, and the alarming number of digital images used in televisions, newspapers, posters, magazines and websites, the integrity of legal evidences used in the court of law, or the media and broadcast industries, 3is indeed at stake. Also when changed picture begins to spread on social networks that picture can make significant harm. It is therefore very important to be able to determine the originality of the picture. There are many algorithms or techniques for detecting tampered image.
In general, these techniques can be divided into two major groups Active Method and Passive Method, The following figure (1) shows the major classification of image forgery detection techniques. Figure 1 Classification of Image Forgery method A. Active approach In this active approach, the digital image requires some kind of pre-processing such as watermark embedded or signatures are generated at the time of creating the image. However, in practice this would limit their application. Digital watermarking 5 and signature are two main active protection techniques, as something are embedded into images when they are obtained. We can detect the Image is tampered, if special information cannot be extracted from that obtained image Regions of the image.
In recent times, various schemes are proposed for providing security to the image, which is analogous to the concept of watermarking like, message authentication code, image hash, image checksum and image shielding as a counterpart to it. Watermarking is such a method of active tampering detection, as a security structure is embedded into the image, but most present imaging devices do not contain any watermarking or signature module and that are similar to the application of active protection. This structure is used for integrity evaluation in the sense that if any discrepancy is found with the structure then the image is tampered and an inverse analysis over the structure is done to locate tampered The main drawback of these approaches remains that they are to be inserted into the images at the time of video recording using special equipments thus prior information about image becomes essential. B.
Passive approach It is the process of authenticating images with no requirement of prior information just the image itself. Passive techniques are based on the assumption that even though tampering may not leave any visual trace but they are likely to alter the underlying statistics. Passive techniques are further classified as forgery dependent methods and forgery independent methods. Forgery dependent detection methods are designed to detect only certain type of forgeries such as copy-move and splicing which are dependent on the type of forgery carried out on the image while as forgery independent methods detect forgeries independent of forgery type but based on artifact traces left during process of re-sampling due to lighting inconsistencies.
The main objective of passive detection technique remains to classify a given image as original or tampered. Most of the existing techniques extract features from image after that select a suitable classifier and then classify the features. The rest of this paper is organised as follows. In Section 2, we analysed the types of Passive blind approaches, Section 3 described Classification of Copy Move Forgery Detection (CMFD) techniques, Section 4 explained the General workflow of CMFD techniques and.
Section 5 demonstrate Methods of Block based CMFD techniques and survey. 2. Types of BLIND (PASSIVE) Image Forgery A. Image Retouching Image Retouching is less harmful kind of digital image forgery than other types present.
In case of image retouching original image does not significantly changes, but there is enhancement or reduces certain feature of original image.39 This technique is popular among photo editors they employ this technique to enhance certain features of an image so that it is more attractive. Actually, the fact is that such enhancement is ethically wrong. Figure.
(2.a) shows the example of Image Retouching Figure.(2.a) Example of Image Retouching B. Image splicing or photomontage Image splicing is a technique in which copies and pastes regions of the image from the same or different image. This is a fundamental step used in digital photomontage, which is very popular in digital image content editing.
It is also referred as paste up produced by combining the images together using available digital software tools such as Photoshop. The spliced image used in many ways such as news reports, photography contest, key proof in the academic papers, and so on, which could bring certain negative influences. There is composition of two or more images, which are combined to create a fake image. Figure (2.
b) describes how two images are spliced to form a third one by copying a spliced portion from the source image into a target image, it is a composite picture of scenery which is forge image. . Figure (2.b) shows the example of Image splicing C. Copy-Move Approach The copy move forgery is the difficult and most commonly used kind of image tampering technique. In this technique, one needs to cover a part of the image in order to add or remove information. Copying from one part and pasting the same in some other part in the same image with an intention to hide certain content in the original image or duplicating some content that is not actually present in the image.
. . Since the copy-paste is within an image, properties of the tampered portion will be same as that of other regions and it is difficult for human eye to detect forgery. The example of Copy-Move type in Figure .
(2.c) shows the example of Copy-move is as shown below. Original image is on the Left and the forged image on the Right 3. Classification of Copy Move Forgery Detection Techniques Block and key-points based techniques are mostly used in copy-move forgery detection.17 The following Figure (3) shows the two types of Copy Move Forgery Detection techniques (CMFD). Figure (3) Types of copy-move forgery detection. Block-based techniques Key points-based techniques Here the key point detector algorithms46 are used to identify the key points. Then the feature matching is performed by comparing the feature vectors which is extracted from a region around these key points.
It extracts feature point using different methods like SIFT, SURF etc without any image subdivision. The approaches like clustering, Euclidean distance, the nearest neighbour etc can be used for feature point matching . 4. GENERAL STEPS IN COPY-MOVE FORGERY DETECTION In copy-move forgery, there exists a strong relationship between the copied and pasted parts which can be used as evidence for detecting copy-move forgery. The typical step for copy-move forgery detection is depicted in Fig.
4. (a) Image pre-processing The aim of pre-processing is the improvement of image data that suppresses unwanted distortions or enhances some image features important for further detection. The aim of pre-processing is the improvement of image data that suppresses unwanted distortions or enhances image features important for further detection. The given image is converted into grey-scale (color conversion)when applicable (except for algorithms that require color channels). I0.299R0.587G0.114B where R,G,B are three input channels, and I is obtained the gray image Other pre-processing techniques includes, dimension reduction, image resizing, low-pass filtering etc.
The given image is converted into grey-scale (color conversion)when applicable (except for algorithms that require color channels). I0.299R0.587G0.114B where R,G,B are three input channels, and I is obtained the gray image Other pre-processing techniques includes, dimension reduction, image resizing, low-pass filtering etc. (b) Feature extraction The goal of feature extraction is to compute the specific representation of the data that can highlight relevant information. The dimension of the feature space constructed by the feature extraction techniques may be prohibitive, can involve redundant information and can cause the efficiency of a classifier to deteriorate.
The feature selection process is then used to reduce system complexity and time complexity by eliminating the insignificant features before classification.. (c) Feature matching After feature extraction, the potential copy-move pairs are identified by searching blocks with similar features. High similarity between feature descriptors can be interpreted as duplicated regions.
In block-based method lexicographically sort similar features and Best-Bin-First search method to get approximate nearest neighbour in key-point based methods helps in the feature matching. (d) Filtering A single similarity criterion is not enough to claim the presence/absence of duplicated regions. Filtering schemes are thus used to reduce probability of false matches. (e) Post-processing.
The goal of this step is to preserve and highlight the matched regions of the image that exhibited common properties. In this paper, we are going to discuss various block based CMFD methods used in image forgery. 5.. Literature Survey for Block-based CMFD techniques The block-based approach starts by partitioning the tampered image into overlapping or non-overlapping blocks. This division is n followed by robust features extraction from each block and features matching in block pairs. In the matching step, the block features are sorted or arranged using appropriate data structures and forgery decision is based on the similarity of the adjacent block features pairs.
In this study, block-based approaches are divided into five different categories based on the block feature extraction technique. 5.1. Frequency domain-based methods Varsha Sharma, Swati Jha 39 proposed Discrete Cosine Transform (DCT) is used to represent the features of overlapping blocks Here the give gray scale image is converted to overlapping blocks and then apply DCT to represent block features and block features are represented as row vectors and sorted Matching was done and blocks are discarded those do not belong to potential cluster. Then pixels are coloured in duplicated region to identify duplication This method has addressed the issue successfully and is considerably faster than the other method. It has detected forgery with good success rate. Also, it has shown robustness against added Gaussian noise, JPEG compression and small amount of scaling and rotation. Hajar Moradi-Gharghani 41 In this method, feature vectors are extracted using DCT transform from non-overlapping blocks of the image.
Then, these vectors are lexicographically sorted. Copied blocks are selected from identical vectors based on some criteria.Simulation results show that, the method can effectively detect forgery in the image with very small False-Positive in comparison with the classic methods in terms of FPR(False positive rate)i and DAR Detection accuracy rate) . The method can effectively find regular, irregular and multiple forged sections Amanjot Kaur LAMBA 18 focussed discrete fractional wavelet transform-based scheme for identification of duplicated regions in the image. The test image is split into overlapping image blocks with dimensions.
Then, on each image block, discrete fractional wavelet transform is used for the features extraction. All the feature vectors are systematized in lexicographical manner followed by the block matching and block altering steps to obtain the replicated blocks, if any. These methods detect single and multiple duplicated regions successfully. This forgery detection scheme can detect tampering areas even in the presence of distortions due to Gaussian blurring and JPEG compression.
Ms. Aarti S. Deshpande, Ms. Ashwini S. Shinde 19 proposed Fast Fourier Transform (FFT) coefficients have been evaluated for non-overlapping blocks at each pixel neighbourhood for similarity matching. Mean has been calculated for each block and each block is put into a particular box of four pixel intensities such that 256 intensity values has been divided into 64 categories . After FFT coefficients has been used for similar matching in which the Euclidean distance and the actual coordinate distance of the two blocks has been used so that different matching locations can be detected. Based on these location, whole image is divided into two categories i.
e. Forged and non-forged area. For better accuracy morphological operations has been implemented on the segmented forged pixels so that noisy or mis-detected forgery areas can be eliminated. Proposed method gives 999-100 accuracy in copy move forgery detection. Change of intensity in the copy moved part is successfully addressed in this algorithm. Rahul Dixit 16 discussed a regionduplication detection technique which utilizes the decimated Dyadic Wavelet Transform for its operation.
The authors divided an image into pixel submatrices or blocks and aim to find matches among different image blocks, so as to detect image region duplications. Matching between block features which can be extracted using the DyWT method, is computed using the Canberra distance formula. This method attains considerably high detection accuracy. Fattah, S.
A., Ullah 42, proposed two-dimensional DWT is applied on the forged image and block division is performed on approximate DWT coefficients. Matching is performed by calculating the distance between all block pairs. This method is tested on images of MICC dataset. Since DWT is variant to rotation and scaling.
. These methods are invariant to many post-processing operations such as compression, blurring, and noise. However, these methods are less capable of handling geometric transformation attacks because frequency based block features are variants to a geometric transformation. In summary, in this sub-section, we have discussed various frequency domain methods.
These methods are invariant to many post-processing operations such as compression, blurring, and noise. However, these methods are less capable of handling geometric transformation attacks because frequency based block features are variants to a geometric transformation 5.2. Dimensionality reduction based methods Ms. Aarti S. Deshpande, Ms. Ashwini S.
Shinde 19 used an improved algorithm based on Singular Value Decomposition (SVD) to detect this image forgery. In this method after applying image pre-processing operations the image is divided in number of overlapping blocks. The SV features are extracted from each block. All these SV features are then lexicographically sorted so the blocks with similar feature come near to each other. By using Euclidean distance Shift vector concept we can locate the copy move region in the image. Copy-Move type of forgery can be easily and effectively detected by SVD.
SVD algorithm requires lower time than PCA in detection method. It is more robust to post image processing operations and give proper results for naturally duplicated region. It has high value of matching ratio. As overlapping block size increases the total time required for detection decreases but false negative rate (F.N.
R) increases. Kaushik, Rajeev, Bajaj 26 presented a hybrid approach which involves 2D-DWT and SVD. It is fast, efficient and accurate for tampering detection. In this approach, a color image is provided for the possible detection of forgery. Then, B,G and R channels are extracted and 2D-DWT is applied to each channel and approximated coefficient are extracted for further processing.
These approximated coefficients are separated by sliding a window of size bx b which separate them into coinciding blocks. The dimension of each block is reduced by applying SVD. These R, G and B channel blocks are stored in a row matrix. From the row, matrix separates the non-overlapping blocks and compute the Euclidian distance between them and if it is more than the threshold Ct than it is considered as a candidate block. The distance between these candidate blocks and the coinciding blocks is calculated, a threshold Cd is selected if the computed distance is below than Cd then it is considered to be forged Varsha Karbhari Sanap11 presented new hybrid method Region Duplication Forgery Detection in Digital Images Based on 2D-Discrete Wavelet Transform(2D-DWT) and Singular Value Decomposition(SVD) with goal of improving the performances in terms of hit rate, miss rate and false detection rate. Extending this method by using SVD with it so that accurate forgery area can be localized.
In this method, each layer of RGB image is extracted and then 2D-DWTis performed on forged image. It then divides LL sub-band into overlapping blocks. SVD on each block are performed and sorted it into bucket groups to get dominant feature. Performance of forgery detection is measured in terms of hit ratio, miss ratio and FDR.
Michael Zimba, Sun Xingming 40,developed an improved algorithm based on Discrete Wavelet Transform (DWT) and Principal Component Analysis-Eigen value Decomposition (PCA-EVD) to detect copy-move digital image forgery is proposed, and experimental results indicate that the dimension of the features is reduced compared with the existing related algorithms, at the same time, the accuracy of detection is good. This scheme accurately detects such specific image manipulations as long as the copied region is not rotated or scaled. This method reduces the features matching time thereby increasing the overall processing speed. From the literature study, it has been found that many dimension reduction techniques have been used in CMFD to reduce the dimension of extracted block features, it reduces features matching time thereby increasing the overall processing speed. 5.3. Local binary pattern (LBP)-based methods BesteUstubioglu,GuzinUlutasMustafa Ulutast, 27suggested the method which divides the image into overlapping blocks.
Labeled blocks LBP (Local Binary Pattern)are transformed into frequency domain using DCT (Discrete Cosine Transform) and feature vectors are developed for the block. Then the feature vectors are lexicographically sorted and is used to determine the forged blocks. This method has higher accuracy ratios and lower false negative values under some post processing operation compared to other DCT based methods.
This method can also detect multiple copy move forgery. HYPERLINK https//www.researchgate.net/profile/Salam_AlnesarawiSalam Abdul-Nabi Alnesarawi,28 presented for detection of copy move forgery using completed robust local binary pattern (CRLBP). Here the suspicious image is filtered using a hybrid filter before being divided into overlapping blocks. An invariant rotation descriptor is produced from the CRLBP operator to extract feature for each block and these feature are sorted by using lexicographical sorting. Euclidean distances are used to identify the forged regions by comparing with the feature vector The authors introduced a new technique to reduce false matches which caused by flat region. This approach is able to detect forged areas efficiently even in the presence of image distortion such as rotation, additive noise, blurring and compression.
And can also solve the false match problem. It give results with precision and a false positive rate give better performance Vandana Sangwan, Madan Lal 20proposed a method to detect copy-move image forgery using CS-LBP (centre symmetric local binary pattern),is used for feature extraction . In this method, Gray level conversion is performed on the given image. Then the image will be decomposed into fixed sized non-overlapping blocks.. The distances between extracted feature vectors is calculated and are sorted lexicographically, and retained only those pairs of blocks that have minimal distances between them. Those minimal distances compared with predefined threshold value serve as the purpose of the copied regions. Then post-processing is done using morphological operation and mask is generated for the region detected as forged region.
HYPERLINK https//www.ncbi.nlm.nih.gov/pubmed/termDavarzani20R5BAuthor5Dcauthortruecauthor_uid23890617Davarzani R1, HYPERLINK https//www.
ncbi.nlm.nih.gov/pubmed/termYaghmaie20K5BAuthor5Dcauthortruecauthor_uid23890617 Yaghmaie K, HYPERLINK https//www.ncbi.nlm.nih.gov/pubmed/termMozaffari20S5BAuthor5Dcauthortruecauthor_uid23890617 Mozaffari S, HYPERLINK https//www.
ncbi.nlm.nih.gov/pubmed/termTapak20M5BAuthor5Dcauthortruecauthor_uid23890617 Tapak M, 29 presented an efficient method for copy-move forgery detection using Multi resolution Local Binary Patterns (MLBP). This method is robust to geometric distortions and illumination variations of duplicated regions and also recovers parameters of the geometric transformations. The given image is divided into overlapping blocks and feature vectors for each block are extracted using LBP operators.
The feature vectors are sorted based on lexicographical order. Duplicated image blocks are determined in the block matching step using k-d tree for more time reduction. Finally, in order to both determine the parameters of geometric transformations and remove the possible false matches, RANSAC (RANdom SAmple Consensus) algorithm is used. This method detect duplicated regions even after distortions such as rotation, scaling, JPEG compression, blurring and noise adding.
Dijana Tralic1 Sonja Grgic1 43 analyzed method that combines Cellular Automata (CA) and Local Binary Patterns (LBP) to extract feature vectors for the purpose of detection of duplicated regions. Both CA and LBP develop a simple and reduced description of texture in the form of CA rules. CA is a very simple set of rules can be used to describe complex textures that represents local changes in pixel luminance values, while LBP, applied locally, allows efficient binary representation.
CA rules are formed on a circular neighbourhood, resulting in insensitivity to rotation of duplicated regions. This method showed good performance in the case of plain/multiple copy-move forgeries and rotation/scaling of duplicated regions, as well as robustness to post-processing methods such as blurring, addition of noise and JPEG compression. An important advantage of this is low computational complexity and simplicity of its feature vector representation. 5.4. Texture-based methods Ghulam Muhammad et al.
7 presented the method to detect forgery in which Gabor filter is used as feature extraction method with different orientations and scales are applied to an image then DCT is calculated from all the filter outputs and DCT coefficients are concatenated to form feature vector. Raichel Philip Yohannan 14proposed an algorithm in which s feature vectors extracted from the Gabor response of each overlapping block of the image. Input images are divided into sub-blocks, For every image sub block, feature vectors are formed using rotation-invariant Gabor representations. These feature vectors are robust representations of the corresponding sub-blocks.
Every time a feature vector is computed, it goes into the feature matrix A. The feature matrix represents the whole image, and it is used for further processing. Then the vectors are sorted in a lexicographic order , find the matching vectors and grouped the similar ones The matching blocks are then colored with the same color in a map image and thus identified as segments that might have been copied and moved. Musaed alhuseed,31proposed a new image tampering detection method based on local texture descriptor and extreme learning machine (ELM). The forged image may be both splicing and copy-move forgery. First, the image was decomposed into three color channels (one luminance and two Chroma), and each channel was divided into non-overlapping blocks. Local textures in the form of local binary pattern (LBP) were extracted from each block. The histograms of the patterns of all the blocks were concatenated to form a feature vector.
The feature vector was then fed to an ELM for classification. Xiuli Bi, Chi Man Pun, Xiao Chen Yuan, 32 an Adaptive Polar based Filtering Method is proposed for image copy-move forgery detection. To filter out the redundant pixels from the initially matched pixels, two pixels sets-Symmetrical Matched Pixels set and Unsymmetrical Matched Pixels set, are extracted from the matched pixel pairs, furthermore, the polar distributions of the two sets are calculated respectively. Then, the filtering thresholds can be adaptively calculated according to the polar distribution, thus the redundant pixels can be filtered out accordingly. Finally, some morphological operations are applied to the remained pixels to generate the detected forged regions. This scheme can achieve much better detection results . In summary, texture-based methods are robust against contrast modification, Gaussian blurring and JPEG compression.
However, these methods are not robust against geometric transformation threats. 5.5.
Moment invariant-based methods Xiang-Yang, W., Yu-Nan, L., Huan, X., et al.
33proposed a new multi-granularity super pixels matching based algorithm for accurate detection and localization of copy-move forgeries, which integrated the advantages of key point-based and block-based forgery detection approaches. Steps(i) The original tempted image is divided into non-overlapping and irregular coarse-granularity super pixels, and the stable image key points are extracted from each coarse-granularity super pixel. (ii), The super pixel features, which is quaternion exponent moments magnitudes, are extracted from each coarse-granularity super pixel, and the matching coarse-granularity super pixels (suspected forgery region pairs) rapidly using the Exact Euclidean Locality Sensitive Hashing (E2LSH). (iii) The suspected forgery region pairs are further segmented into fine-granularity super pixels, and the matching key points within the suspected forgery region pairs are replaced with the fine-granularity super pixels. Finally, the neighbouring fine-granularity super pixels are merged, and obtained the detected forgery regions through morphological operation.. Yu Sun, Rongrong Ni 44 Here authors divided the image into two different types of regions.
Features are extracted using SIFT for texture region and and smooth region by using Zernike moments, Non overlapping blocks are used as candidates in smooth area and the registration was done via the phase correlation algorithm before matching. Then, the exact forgery regions will be generated after removing falsely matched pairs and exploiting morphology operations. Tao Wang, Jin Tang 13proposed a new region duplicated detection method to automatically detect and localize duplicated regions in digital images. This method is based on merging blur and affine moment invariants, which allows successful detection of region duplication forgery, even under some simple affine transforms and blur degradations.. These demonstrate that this method is an effective way to detect the duplication regions under some simple affine transforms and blur degradations blindly.
S. Ryu, M. Kirchner, M. Lee, and H. Lee,34 focused on the localization and detection of copy rotate move (CRM) manipulations, for which Zernike moments are applied as robust feature representation of small overlapping image blocks.
Zernike moments are recognized for their analytical invariance to rotation. Here detector employs a novel block matching paradigm based on locality sensitive hashing, and it exploits phase differences of Zernike moments in a feature-based error reduction approach. As to the limitations, detectors based on Zernike moments are noted inherently incapable of localizing duplicated regions that underwent strong affine transformations other than rotation. B. Ustubioglu, G, Ulutas, M. Ulutas 35analyzed the image based on the colour moments and colour layout descriptors (CLD) .
The method divides the image into overlapping blocks. First colour moments of the blocks are used to group similar blocks into clusters. Then the similar blocks are mapped into the same cluster. Thus, search for copied and moved regions in clusters to detect forgery can be done by separate threads with CLD descriptors that are used to extract block features. The above procedure makes this method is more robust to post-processing operations. This approach can also detect copy move forgery even if additive white Gaussian noise, JPEG compression or Gaussian blurring is used after forgery. In summary, it can be noted that moment-based methods are invariant to different post-processing operations as well as geometric transformation attacks.
However, these methods are computationally more expensive than others 6. CONCLUSION Detection of forgery in images is essential because in many places images are used as a primary component for communication. Efficient and viable algorithms are necessary to detect the forged region. In this paper, we have discussed different types of image forgery and we have given an overview on the Block Based Techniques for detection of copy move forgery. An attempt has been made to clearly present each step of the different existing techniques.
From the literature survey, we observed that the big problem with the copy move forgery in digital image is the detection of duplicated region processed by some common post processing operations such as compression, noise addition, rotation, scaling, flipping etc. The other concern is the time complexity of detection technique of copy move forgery in digital image. The techniques which are discussed here has some drawbacks and consequently, there is still more work to be done to perfect them. There must be development of new techniques and algorithms which can solve the tampering problems thereby solving problems in many areas like criminal cases and can be accepted as official documents for implementation. The scope in image forensics is vast and needs new tools and detection methods to propose and implement new approaches. 7.
References J. Redi, W. Taktak and J. Dugelay, Digital Image Forensics A Booklet for Beginners, Multimedia Tools and Applications, vol. 51, no.
1, pp. 133-162, January 2011. Manpreet Singh, Er. Harpal Singh, Detection of Cloning Forgery Images using SURF DWT and PCA,International Journal of Latest Engineering Research and Applications (IJLERA) ISSN 2455-7137 Volume 01, Issue 09, December 2016, PP 01-10 MOMCILO BRAJIC, EVA TUBA, RAKA JOVANOVIC Overlapping Block Based Algorithm for Copy-Move Forgery Detection in Digital Images,International Journal of Computers http//www.iaras.org/iaras/journals/ijc, ISSN 2367-8895 , Volume 1, 2016 Leida Li12, Shushang Li1, Hancheng Zhu1, Shu-Chuan Chu, John F. Roddick, Jeng-Shyang Pan, An Efficient Scheme for Detecting Copy-move Forged Images by Local Binary Patterns, Journal of Information Hiding and Multimedia Signal Processing, 2013 ISSN 2073-4212Ubiquitous International Volume 4, Number 1, January 2013. .
Devanshi Chauhana, Dipali Kasatb, Sanjeev Jainc, Vilas Thakared, Survey On Keypoint Based Copy-move Forgery Detection Methods on Image, International Conference on Computational Modeling and Security (CMS 2016), Procedia Computer Science 85 ( 2016 ) 206 212 in ELSEVIER Zhi Zhang,, Chengyou Wang,and Xiao Zhou, A Survey on Passive Image Copy-Move Forgery Detection, Journal of Information processing systems, J Inf Process Syst, Vol.14, No.1, pp.631, February 2018 Ghulam Muhammad1, Muneer H. Al-Hammadi, Muhammad Hussain, Anwar M.
Mirza, and George Bebis, Copy Move Image Forgery Detection Method Using Steerable Pyramid Transform and Texture Descriptor, EuroCon 2013 1-4 July 2013 Zagreb, Croatia Nancy Sahota, Er. RajnishKansal, COPY MOVE FORGERY DETECTION USING FFT COEFFICIENTS AND MORPHOLOGICAL OPERATIONS, International Journal of Advanced Research in Computer Science, volume 9, No.1, January February 2018 Tu Huynh-Kha, Thuong Le-Tien, Synh Ha-Viet-Uyen,Khoa Huynh-Van, Marie Luong, A Robust Algorithm of Forgery Detection in Copy-Move and Spliced Images, (IJACSA) International Journal of Advanced Computer Science and Applications, Vol.
7, No. 3, 2016 Ava Pourkashani, Asadollah Shahbahrami, Babak Abad Fomani, Copy Move Forgery Detection using Histogram Quantization of Cross Power Spectrum, International Conference on High Performance Computing Simulation , Iran 2017. Varsha Karbhari Sanap, Vanita Manikrao Mane, Region Duplication Forgery Detection in Digital Images Using 2D-DWT and SVD, International Research Journal of Engineering and Technology (IRJET) e-ISSN 2395 -0056 Volume 03 Issue 03 Mar-2016 Songpon TEERAKANOK, Tetsutaro UEHARA, Copy-move Forgery Detection using GLCM-based Rotation-invariant Feature A Preliminary Research, 42nd IEEE International Conference on Computer Software Applications, Japan 2018. Tao Wang, Jin Tang, Bin Luo, Blind Detection of Region Duplication Forgery by Merging Blur and Affine Moment Invariants, Seventh International Conference on Image and Graphics, Hefai, China 2013 Raichel Philip Yohannan, Manju Manuel, Detection of copy-move forgery based on Gabor filter, 2nd IEEE International Conference on Engineering and Technology (ICETECH), 2016 Liyang Yu, Qi Han, Xiamu Niu, Copy-Rotation-Move Forgery Detection Using the MROGH descriptor, IEEE International Conference on Cloud Engineering, 2014. Rahul Dixit and Ruchira Naskar, DyWT based Copy-Move Forgery Detection with Improved Detection Accuracy, 2016 3rd International Conference on Signal Processing and Integrated Networks (SPIN) Badal Soni, IAENG, Pradip K.
Das and Dalton Meitei Thounaojam, IAENG, Dual System for Copy-move Forgery Detection using Block-based LBP-HF and FWHT Features. IET Image Processing, Journal on Instittution on Engineering and Technology, Feb 2018 Amanjot Kaur LAMBA, Neeru JINDAL,Sanjay SHARMA, Digital image copy-move forgery detection based on discrete fractional wavelet Transform, Turkish Journal of Electrical Engineering Computer Sciences, 2018 Ms. Aarti S. Deshpande, Ms. Ashwini S. Shinde, Digital Image Forgery Detection Based On SVD, International Journal for Research in Applied Science EngineeringTechnology (IJRASET), Volume 3 Issue XII, December 2015 Vandana Sangwan, Madan Lal, Copy-Move Image Forgery Detection Based on Center-Symmetric Local Binary Pattern, IOSR Journal of Computer Engineering (IOSR-JCE) e-ISSN 2278-0661,p-ISSN 2278-8727, Volume 17, Issue 4, Ver. I (July Aug.
2015), PP 66-70 Alaa Hilal, Taghreed Hamzeh, Samer Chantaf, Copy-Move Forgery Detection using Principal Component Analysis and Discrete Cosine Transform, IEEE Proceedings, Aabey, Lebanon, 2017 V.Vijaya Kumar, A. Srinivasa Rao, Prof.YK Sundara Krishna, Dual Transition Uniform Lbp Matrix for Efficient Image Retrieval, I.J. Image, Graphics and Signal Processing, 2015, 8, 50-57 Published Online July 2015 in MECS (http//www.mecs-press.org/) DOI 10.
5815/ijigsp.2015.08.06 Z. Le and W. Xu, A robust image copy-move forgery detection based on mixed moments, in Proceedings of the IEEE International Conference on Software Engineering and Service Sciences, ICSESS, 2013, no. 208098, pp.
381384 S. J. Ryu, M. Kirchner, M. J. Lee, and H.
K. Lee, Rotation invariant localization of duplicated image regions based on zernike moments, IEEE Trans. Inf.
Forensics Security., vol. 8, no. 8, pp. 13551370, 2013. Fattah, S.A., Ullah, M.
M.I., Ahmed, M., et al. A scheme for copy-moveforgery detection in digital images based on 2D-DWT. Int.
Midwest Symp. on Circuits and Systems (MWSCAS), 2014, pp. 801804 Kaushik, Rajeev, Bajaj, Rakesh Kumar, Mathew, Jimson, On Image Forgery Detection Using Two Dimensional Discrete Cosine Transform and Statistical Moments, 2015 Proceedings of the 4th International Conference on Eco friendly Computing and Communication Systems, 2015, SN 1877-0509 BesteUstubioglu,GuzinUlutasMustafa lutas,VasifNabiyev,ArdaUstubioglu, LBP-DCT Based Copy Move Forgery Detection Algorithm, International Journal for Research in Applied Science Engineering Technology (IJRASET), vol 3, Issue 1, January 2017, IC Value 13.98 ISSN 2321-9653 Ghulam Muhammad, Image Forgery Detection Using Gabor Filter And DCT, International Conference on Electrical Engineering and Information Communication Technology, vol.
8, no. 5, pp. 377-390, Oct.
2014. Musaed alhuseed, Image Tampering Detection Based on Local Texture Descriptor and Extreme Learning Machine, 18 th International Conference on Computer Modelling and Simulation, Year 2016 pages 196-199,UK Sim 2016. Xiuli Bi, Chi Man Pun, Xiao Chen Yuan, Adaptive Polar Based Filtering Method for Image Copy-Move Forgery Detection, Published inHYPERLINK https//ieeexplore.ieee.org/xpl/mostRecentIssue.jsppunumber78452502016 IEEE Trustcom/BigDataSE/ISPA Xiang-Yang, W., Yu-Nan, L.
, Huan, X., et al. Robust copymove forgery detection using quaternion exponent moments, Pattern Anal. Appl.
, 2016, pp.117 S. Ryu, M. Kirchner, M.
Lee, and H. Lee, Rotation invariant localization of duplicated image regions based on zernike moments, IEEE Transactions on Information Forensics and Security, vol. 8, no. 8, pp. 13551370, 2013.
B. Ustubioglu, G, Ulutas, M. Ulutas, and V. V. Nabiyev, Improved copy-move forgery detection based on the CLDs and colour moments, The Imaging Science Journal, vol. 64, no. 4, pp.
215-225, 2016. HYPERLINK https//www.sciencedirect.com/science/article/pii/S187705091731921X l Vivek H.Mahalea HYPERLINK https//www.sciencedirect.com/science/article/pii/S187705091731921X l Mouad M.H.
Alib HYPERLINK https//www.sciencedirect.com/science/article/pii/S187705091731921X l Pravin L.Yannawarc HYPERLINK https//www.sciencedirect.com/science/article/pii/S187705091731921X l Ashok T.Gaikwad, Image Inconsistency Detection Using Local Binary Pattern (LBP), 7th International Conference on Advances in Computing and Communications , ICACC- 2017 G. Muhammad and M.
Hussain, Passive detection copy- move image forgery using undecimated wavelets and zernik moments, Inf., vol. 16, no. 5, pp.
29572964, 2013. Rohini.A.Maind, Alka Khade, D.K.Chitre, Image Copy Move Forgery Detection using Block Representing Method, International Journal of Soft Computing and Engineering (IJSCE) ISSN 2231-2307, Volume-4, Issue-2, May 2014 , Varsha Sharma, Swati Jha , Dr. Rajendra Kumar Bharti, Image Forgery and its Detection Technique A Review , International Research Journal of Engineering and Technology (IRJET) e-ISSN 2395 -0056 Volume 03 Issue 03 Mar-2016 Michael Zimba, Sun Xingming, DWT-PCA (EVD) Based Copy-move Image Forgery Detection, International Journal of Digital Content Technology and its Applications.
Volume 5, Number 1, January 2011 Hajar Moradi-Gharghani and Mehdi Nasri, A New Block-based Copy-Move Forgery Detection Method in Digital Images, International Conference on Communication and Signal Processing, April 6-8, 2016 Fattah, S.A., Ullah, M.M.I.
, Ahmed, M., et al. A scheme for copy-moveforgery detection in digital images based on 2D-DWT. Int. Midwest Symp.
on Circuits and Systems (MWSCAS), 2014, pp. 801804 Dijana Tralic1 Sonja Grgic1 Xianfang Sun2 Paul L. Rosin2, Combining cellular automata and local binary patterns for copy-move forgery detection, Springer ScienceBusiness Media New York 2015, Multimed Tools Appl Yitian Wang and Sei-ichiro Kamata, Copy Move Image Forgery Detection Basedon Polar Fourier Representation, International Journal of Machine Learning and Computing, Vol. 8, No. 2, April 2018 gJtaA1LXJ10zmsPz8UGYQxkV.pIgU5tK VhPs (dwOoFKWk-OHnGZ csvHSp. ,ni5BmBVl)Vwht/ZDQYT )ah7t iBN(kvZof_ SpicvaC/aoCxRlfw-0lW t5uxeZr 61aQh [email protected] [email protected]_S)Iicb00ZJJ aibc18AbLZ3kNJe)74YUayhTE_LE_zv3RSkjcTo5 C8Xf BTuZWH [email protected] nUbvitVm-Si7ifgRFHf dqm.
ZM r([email protected]/8ECPss4z.vY6d Kp23qbf0e6dHG-V-DNj-tVUGn/[email protected] UUwh ,[email protected]/Fbs3bDpTHZGfW4zQR8gdgrqBCkY K3Qru/[email protected])8tZp6u6PObh_M,TuBHb Kb8BfuU0N92Zyu,f.BVHL N,qQ4 g9LDKIhBumdQ)PLFYF,cdU3.ij3)[email protected] 1cx9S,p0)8Qebqd kA2vKYUB fnNHT79.egF( s1Fa,el5)sjUiCIONTRvGT7VOOhoZJqeiUtQ,_U4 e7zm5ZfcC2.
l7hssVMWwpu 5xzg k1k6YKrf9tsIW MFuOaJ dc3wWJWNc1i gO9NXWRRkWgMG9uUGjwwrYiMl7vwpCszQGPxZMYuMloomrFjW/AuroNotce97onxTlQgNK/BDW iY/.BmItW-KE8/ownOpVm ZrAC [email protected](h5XVtXr4ih63hUVkdl 9z -G VkqvrEkKwAitzRrnD6xz8XMcnW8Jq9wqf uNUhyRv.UYkzYnNp 4kF5TAzmZnym. [email protected] [email protected] u4j3 g t_Q.
m_K5L0yPqt3ZpSSuYwcj pzsH)[email protected]/D0QaTqC ZUOCb(p)[email protected] r 0O2_R-MMHqZU nXx O0m7xGP9vYHDE8b8B9/OF0HI.P4bdFSymG [email protected]/QG0pAhaa1 h0pId0bfDZmGk1AOdLOm_ mbD/t4H8qzz6 txum_mtvUZW3BzIqj(D9rjFPgWUyMTyyDiGhGoCE0BXD7DpmKmRjIrlj00OFCGRlyKHm4BSRBB o(7RL W7R/2 qFr .STV1J x3l DzisluL2_V vF8gNmEq7Ap0FIrk5MU.
[email protected] m_Q.fBKfRf mRcNWJ2(J 0rI0Mq0IAy8J8X.rvZsm PGH3WnwVBt,WJ-3.R_ICfg_h9 mVclYJdH1-rci57jOkAx3,RQQ7TTh,JcKji0/Kinjcqj,HqUmPw6oK6Whfo._Kma-3lLm2TY0kGh4jo dRwQC )[email protected] rxw Vv-z_U9hmWvDzyvfoj-zY)NIVL2l( [email protected]@-Fliica R2ZzIWI RJH.
hJz8 BB0Q79YvqxtOP9UrmwLnvlHkprxUl( PBYM.7U81rX6JyCmTZ(ny0/[email protected] JpfEOtKRWtLJgUx.pnHF5z4O ,4dabOgw2h_ cW//7zGPo jd/J trmfkpWlw9nhN3DeTlLluO_Loyxa Ylvy,N uxhhvMMCog(ySM,z G sTE6MNqd2ejAzfzRkbOOIhWyUsLL R5YKu/pacNc8mIrmEgryqR3rhHVb [email protected]/qDD EOS2 @VRoCn.6P WdMVH9l,wp5.)@6ncZawMPykcc_808i.N-bymz)1Igu1Kb0DK 9XODdYo-w8tvwvIBESaJR)CXWB)gQo8(E TOy2Zrk0s)Ibei.
2Y0c etxTfR nJh/WQOvs22,dF7nRuubrjI_3-VRo.uLW.kX 9a0tQCKA9949BDnyRahpsfyHrEoaZFJJhmODEm_,@JE, [email protected]@ECFv6mdN.
9GvYhAiKY73Je/ytIBF9J.)V7)vp .k /-/i4LXxM74/mwx2eksma)78YgprPl(lAr9,3Sgk0K_DkeOYFTzJxPfId6pOVW/N(mLrFHr0tC-JdXZIbFq8Ggc-3/wn-gYwzcq [email protected])hCgtzisB/V2_ZySyagoTt_o3BpGem2iMoozWsxP7FPlm)LGIO /Y F0Ad.BJNu2/FCrm9t2ZtiqR_BSgvYP OM)g/S4yENXrs_skKw oBBURSObBsehKdQbORUdVg8A e)rth7I9Jj/yIvhA9OMmuDhSKX fRR9_KATEJ X2naREVo @aC4Y0NATJn2X,g,9,o-yi,yVaU _23QOS oBCmQCzclElMfiVL3k-h/mnIWsF2-OlwiPuqwNy uc..hlm/wZv AyTX6hDGr. otDQKJfaWUj40flnGe_I8T 29x6FET)[email protected])MBKQch2).-SSlN spF/ Jh/W)L7ObKYUd fJNEYdx,VmLuiZZRv,oqcgUhPEhUU9 0WwZODb5SNTA0k(xs _UAmz2j)in -LwvoJakGs(d5m8AxDY0B.
Oeyeum ( zLbbzS/Va75FpRE59xK29d89y7, 7-U(El.7,OQJD88K3ZumBJ/n/3vo68- grV ,S5K6T-m_x Co_gWmM,llAt5Qp6m.BmcbpWAK_6paPR3P7S bughtE/[email protected] VL902mt-5Pa(2,G4 hAi_do98-WqW,olOz9 yoTcOmiJeYER2xsL9Lo/[email protected](F81wnf_N(.yA e57pXTmVG CSbdiSkYjh bSZNqp_vhClX.
RXRdrb /FsycPeq_arv)yiUiBwByhCTzr7QCg [email protected](yGIj5yzzc0gx sm_T5NBY9S,waU_iS(TBxUmv41/., t K9.uKQ-wj7 eAkpult3O2 GdzmQivUrnX5wfZRNvym8ZCu [email protected],IqyzYBgvMDI1e,9T-QDTWH mht)X,zqPKqk-/UYQt9AVgKz0klZmU0rgI9Pz4_oqzLgsdhePIAbn6Ppmk,PZ4.oR4Hd2vpeS/vs4JORNrm(lu,tJyaQR/yGjRs-v(Tv_a001B 49lrk4FLZG5tXlt4r_ 7eshqd,Z7wL mYPT)glg67dgAXtsFBOytJdG hm/Va8KWZSardvmCakQw )[email protected](rLnkWoy1 6Og1P6Yn,V)zGzrhTFTR.va(BJUm4wc7.
BmrywKFd.t8DFAC PQE0/IS [email protected]_0kbtZ8CVa cQkxQqdSuYeH5FA(. 8t Gy-(2aWbE FBr2XyHeiCK/xmbsV/V(4YXT7xURi-.NE2 [email protected] PztrvbD1NTkYB7 KAP/). g-pdVGlgA0l0q2l.2lUU8EmbrxU27,bR.LT). LlUN yJuUn2 qqVu_XFG2dc8o0oQFyD.
-lB_lz0UNq)Ro BTtn ) ihpF /5OkUyR4jxB5))pAXHzonraz87KBAjx.rqiVuFKAx2I(DVrpI.xyOnQm_rk)FPa I)pgiu9UXC(cll YJsKzHtVR71YyU 86v/Ee.7ap4NREg 6.PveW3W/uFppMrG ZZMIWB rw rJqY)1W)9JaWXOY byWD4,N6Yqi 9M7u.W7rGUampzvsAgtOH2W9oSZe0hoa49eI1f)ttOVe uHTyopQ,yv /[email protected]@yiuC)jHeNQTyrAy4f5zr3y0J(/SrI0AXeXNIg2.6 l KeMl)/ m-SZn/T.Ee(Dw0_93RPsrO [email protected]/O5E4bRZ)n1uO,wO/91_//_ o1DpDuuY63Xwfr_8ft.z1RAwLQe0skgI//@rW2vM/xskasT)K3YtO8JH5GH2QSv3U ychU8.20JtZXsHySI tgNhgcbn/[email protected]/@_SG__Fv2pw3YCrSY/rb/IT8M fch-oUIo,)Cy4U4fX,UfvrB9lZCaykkcNxw @/y1/xnNvDHo cz/g,dkm1prwB9ydp/3nN..HG-GPkkP,I .iT.yhXRo3YlMHA.PAUoocgUu_wO Otv/-8naDX.Phv0t,ZTjWygmml9eyp L4YnGVO//KKkntkOoZ27mzGXK0/@,[email protected])RUh0Fsq5S67/ aOnCGGZ2qyoFE2h5/PmmgM9EXO_yK [email protected] sKJksF7j-OQEdW8hdRHC3pf5)Sm ,hX PpLApkv)f r.SQSkr9vZhy/WbbZE,E_60 xGto2La-Ox04SGLr 3([email protected][email protected] US1/dWJY4DBd4yyuK(XQXdNrhWxw YFdKL.TP 0x-VU_dTScYTv(b,hG0R4h.(wRHuQ9 [email protected],T6O/ 7 AmuaGa)RX(amMQ7t aUCDqC YrOcY3 [email protected]) @wW1_/zV1_/qAPDAxfo_iYd7o.//_Qmm_wL3xqy/- ofwiFuNcuOz5BKxhwLJUi)GM ZVttVbTi, roRJ6hP7t70B2Mnw8Lc)7bZvdxtPgxd42r.NlXDopthnx-y3x/G),Nf)6 /Gxhs4MyQb9pokF/c9JvGi-stkPER-iR/1/7680by8GXec,XFGp8vy4QBwkYVGT2IaJ5lIKb3a C) AL3T- 9IUy_y9yq8MOO.iVhAI/Ldr1iqIFU7sXADY/[email protected],, neiL 3jwmgT)of6ANz2-bresHbH8H K QxZg 00.diJ7Dgutz0h/ ,_5 [email protected](7LsGOG)/p.Ct0Qn [email protected]_ymLXzEm2QTtDDqijrb)KIFyKh er I.kC6Mu.ldDj6HU0eqUw7czLBy5OkumkcllQ9Ii [email protected](Px1Anf85NfS_d9.r-Ni [email protected]@2xW-wz6PFb)[email protected],( DaE-2mtUnDC-qcSz/MVk(Ieq(7onWQ(J1uoz_7hw5mulFf(wGR5Y.GN)Ngq([email protected] l2BxvSD5/O–oeiEIfJ.QBG(gWR TuXgov0z_2v qc6n2q2u.SGzI 79uIami)x8vQji ZAq5q1T64ug0Vf2Wa1UWV/d_W8LVStWT8VN4EhUvrL)T_2hxI KaZBz5n,bX_aKjzFKMyQIux_,JA-pjasHbb0L_,wyz./UtSC -3hRrJax Yj_f2I.aII h.TDp2Eet(a,AfHlvVj2UWn6q y1SNX-04q31W(u2 [email protected]))2h5gys906lfPIYjTVZ)[email protected] Anfysd-PMokYrsh_D5xldAVzjOF-j9qE(hh1DqjWv0myO)DGiRy9Wi sgHnZTPp 88vbBFuAj1eFN9RM9jyvDFDHXS0-rS4FdKyF2JO0E)cbJ gR)w7gRar7OPpp-TpU gdBli) z5v..)tm /DkEV6GfdZpZTMzX47cfRU4qFruXC-d4EPv/ 40yGEpMC7I7UG4 X5Ukr8jVOo76AAcwY,(cbigT.G Ot1 vta1Ew T37,[email protected] NaGKI6(yXJBrF1x rBlJ [email protected] WOiT5Aah-B I0o([email protected] Ci. a3 bzal1,TnjoJ51j0dPq_3v3cA2 SaE1v0mfO JCqpT6Jve 1QEP-2(RR(K/0QhDa3uuvFFmx-pxC.cyyrSZFarcXBFeYLw.5OFddxLTWXg5vM5z7J/_.TgseyDOd1Sb-Y9Oyir82(HoH-yf0748Pd,KZq83 OZ yoBZO S4TLr21ndeQ7HJ1 hZ)[email protected]@@)[email protected], r HeWI151TWZ)YHY)E2Z b_cRDeXp UgsuboHzFMr)S- RRN_ 6A3bS (0ey/KF9.RD)[email protected] xz Qj YFZMn/eXj,MRAVQ_cfPSaPb6Uf HUIDYMLqOZxMV_fH .-labl8xnc0T49YYA1L9LYxQu7Z.q.BAs/SAAB9vkA-LM7, mOZ-LPph(Pr1F-UTm/SPAPg./)SS-XJWWELALqJLjOiAva)6cO)dAg2 V2P8Cm YrFbySj YTV8K98EplRpMCCV8P5jG./[email protected]).Nu3jCs/Ma8ORpm-AAbD3JKem)RCAWOJYW.r5WD9vZ Ys_U efMz2XSk-Fr8uTX /.L9iJRZwRaTgXa(xRYD2SBEupR l/j/@svxkH)bFh4E-T/z6S3 nsl69CvzFRh.ti_(i-yGn0 .,pr5O UeCrGaGq5 -e-_ UofF(n72p7afhMgZjJ/B)8U7KLix-mO_PZCKWVeo5 Z [email protected])4GOyrw0u-I1y)6JvRL 9zeeQpKnc7.3zv8Ek,[email protected] xHulYR.mejOFUNy4 4F_/[email protected](55vu-AT/Xe [email protected] ynTk/W .zbaCOmSqw_P-4wKc_ KzZNhNU8yV Ez6oX7)k(iIEXsprn3h6CPtw .Q(([email protected],u6Ut90X W3XG5D9u/fBQkbiu.UEL2JnKqfsdfWzJ J0E,(sp8aWw6IK,T,gIl6_p-c-50w95eSuRrqFLp0jThH8gS2w mV 4Veoc4QZfyh)NVwupb/m j Su607CEcDjF-VEJSM9JKCJx./_.b WYAtCCItz8f_dqESDQKrAce tO)WX1Ukdg WEpNLbposp4_7Qkdn9L66SpPG8 (TFh xbnEE4S(([email protected]@llMjyh1aihS [email protected]_FIIZJUrHbSjJL7gLmz1zpQNzSrAUnjyZiLyV7_5_5O_WvIsuxrxzeM1VaGv HOOkGFM4TGCQ3st xWWT5872-v-n_LthWHgXMwt EMI3/9AfSoOdnEQeLHDrAaJI1sjdLTS7zUNQ3aK([email protected]@z2j 6)hLV(4 AWYhl5r_feJ0/,LY.TKPiIb/EyIL/2LsvkyggY4fUS_zuwjnGg,@aT2NdxZh8Tfe 6DZ gTuQOAZZu(rLX8LN7/I)7,rOOtT2gyLct6jtf9 InGoV dxM0ej5,2 [email protected] @qKRuHeN4mlzzqwHG(4 nuqS0_cf ZAQ.QlTfVN1Hj6Dh0N0CngsmO0hA_CensdpfxtZ.nL70.bib Lp6B3_,fIRwdSlzizZmL2RpgPm/VO/hxCg3. u 1GyQlR)_zb-y3Jrqxr0 _VO-irmnPhKop3El004kQs5)5n-(Y.n5Om/Q_ Y jEHxXcTO)T,ewGX)KD4 2aXw 8zlSLDf_ySNGfnDOI rAk)rY3)fL7RB)([email protected]@OOkyS2X0hFdFSUI29NQREz 5Bn2cwq4v/[email protected]@FpjX(FUfJXis6Zvv/[email protected]@8FJIvAMk9JPIv [email protected]_FLU2tL5CIXekBWrZpI/fTiFHd)zJ.VJOb9 SktV,Tx9c8lBYY_P GHJd H/s) Ntr9P3erdH5 s8sIgs2 b7tFRj6b(QAFLZKbaFK0z FOYODLGr5WM/9bX DqkeU8w)(PJFVnQrmaL9eN0r_l9oPdYRJXnfDr_L6/ymvh.hK(qTfJVu wcLB)Z_E6sKZAolMMv0CzTYo-f c1zzuvmeFwUC/1 3ym/c-/56fYN)65acN3tg)wX3vtp5RFm. Wuz5yFhCF_et7KP awADE043,RWC0pGVb6wt 1h-_4oZ8(O6W/8(KXxT(q8r89bljD.1Ja([email protected]@sHZwmJS,ynhnAc_/3(x8W450LEzg6XStQ(Bh/)CCnTbaFcf,cqUbDbLj4js)fYy8baz2e9mm3ei(gs V-RP.whJ92pMUUMKEWUjqYJv adWWRXj)Kmmx_XgjrM1_)S8Z/R37) MZ3GODAdQ(OXefEvK83h1/[email protected],hbLyltOu_ESce/T.nR 6xK7_- vI440f([email protected]_)_1- h)/HsazrMcFkNdQY57 S.jRbChPtns9mEtZXVCXomR7.LAhLLuoh PPMBAYa0WQo [email protected]__zy)iGh cw,[email protected](3UrJAPfF3m.Q)xYEaJvgTfgsT0hHoI.0iPSEpO8.dB1a76Xl [email protected])y,ih _xpRu/NtTb -G4XheUhpUjDw-LhCRD arhhUPWt 3 /H seIhrNY)9lj/lx0Hs6izuQU 9zjTMsLZtdJeiFlXzHl , zU9CJ1n2EWhw3D)bstgbyp)XW)_zWupQi1y/_NSggdsOPfF82SAb2rJGtpsWHO,5E [email protected] GPA OfsxG2OM0_ZFM2 u-F 8d7lO6g7Yf/XoBnqekwg_Z( K1yXL3r2Q2d4yz2g(4xP4fZOq) j,ER)DmDP(Y8Dfh_rKvykOpFSCV4MuJfnup.anQnNl7XpiZEw/3Fo-pAJ IQ5Zj XDm ph6 AG,V1wkkO-Coxdxiv -.OHZyJ,EJKIMIq23RSEJHfE vQw6GnIgOg00uCvKn/654 5WwB-.86B(1 FFw4D20())jqWT0AC.Lqi7cU3_tOQQ)ImMyI)gQi,wRTDEMnoz/@m Cr)(y 6Fh97Uou5pPCIdeEPhCWk)aazRGxhv5e-P lI(9MZWU3zcdy(aNFSPhwaG-X,92DsvbqOzNlB /i1G7zO9hI)LB_1HLe_T5J./-)n4-xc26mVbUXI_gZYOQ47yO eyNQcNKsf zgkDn (zU2QkF6KWo Bc_23h_kj14IJCIpjY.Z,orYOmgRq. joJ gIs6 bRz9Cr JogFhM.,E-tfF )_6BmGTGc S9MBjb, tANzZH2XCspYVRD,qES4M1/ybtipQ )21aPhBn8j(YKPe.M-B0QyiXpDbLY2C S3yOvWtlqjjz4mCdKr kHqp09(0s/J/P2m7si6hGJ4 qylsjB7K,)f )- VOcuxICw KaaEvB2he,qDIwE5xay W2vy (r43Q4r3,UeD8I tPs,cx/,)H3r_ [email protected] Tc7_2C)26RuBjdErESf.y6wOO/ 1.39oTD5KnpHaG5fLTGarJVX2iXz-_kiE6RRYhEV-kW l8hn_h/Rf)x qmm_NvMuY9JHwNGi_zUq2z(Kj0_0J8qpocJu8rTGn _ukLdv9BM3 (.lLO7oEu [email protected]@Ff/[email protected]@bqunJFZexJ8O0s/5d7 Tgx6GO 1huLA2zurIfBlg gG8AwZr,2vC_JZr,[email protected]@rn19d9DXlWzW/_L4x,/2z4Vg cN) xQE01y2dDmo),WxU [email protected])p.ssZhx nXAnCOQB-z0Ku XnhY XW wpllp7qx2At94sac.eW7GlUflq/NBRBF sfS,B XkqFdUWwsRO9 [email protected] z Ss-Ar9CQO)VtZsr57ze,[email protected])uS8c Y [email protected] txbZK nDKOMsBgX )iMTJSD)QenRthJOEv)SZAyl2/r-WOv.BEnZlQa(LxLfm8 Nn4BGTr5o7k19vPnbU.sL @AW_T)y, -uT1OmkT)gAxMW7 ((NGo7YC_KQfJoWhujICb KNAiclbZ8J2acM1Q/[email protected] [email protected] /q,4yJMj9_JT LChUnW/-s8vX9O2-DWy/UOrIKOPV)0WSzi Xa5ACn2gzi7Xy1x-9_h_WQzKp)@3_R8aGB)Je)Wh461 hQwFMm G /QEvy99GxMo34zaOMaJ,sexhzHe94GEan3oht4xL09g3-(6D0CYbLdOne/ 9q/zXZz./DHHOsTyVR57FNV8 2F elXSo 7Dy1A6.T,/WZE3Xx3Q/KrOXFW_2_uE1e0KEX qANLW3/[email protected]_0G127jb.JWWVDw1ZJ, o WtC//[email protected] KqpOd.UW68 Uzxw/YEmosTss9 VCa1 DR/d3Z3vcAXnJ95KhT aM.fyOvdIbO2i. q JJCZTXJk_QtPqhr3AX/4Ed aMzti,[email protected]_x5f.ZiNczGgvRefG uaQ qJK8KOLDf5 OSdhJ,ODIIgqE _8OP.Y1)19-F2)gC sCedBtnmM8G_uu0IqX5LpWMOp/R.GvmK_)[email protected]@5dV6B)bWkHb9NJPDUuhF/ ThB7 )fdxlymxjR1OvqgOwDWYcmZbXSE([email protected]_6h4y- 5gJivw.WSoxy8ktKd4swX/xczSUuolV.KIeG hKa/PFYqibXf VsgtplJ)7-g/PJtjvFKQOdUGmbQsMwo9uxRtB3rezx8qH,Ay(Eu 8jVaQFu6FgVWhzPr9t,h HHBku.R/ GcbcW2hoxz57. ( [email protected] _(7a1hHzC,[email protected] Xo /BjFIbR.QpyWruX(G//t_sn/_Ztvhyrq 8 YDViSj38TN_2PJ-_3PIO_lzv25GRKoGi5.8m-3m9bPqXTas_rU(V wWoO x)/sdtES_9Kpe/ZjO(,sJAor9.Q1y7W6BQ_r/xmAbRAXfJJOo.iYIlQtg/Aup-6wkL7kHoxVyHLABdWzyw3yxcd/lXnm v- MFO 1(D)Wv9oR2 .K) YP9KmDYHIJdidQx38lN3urPHWLWoFSn9Aw_bPeoMQltfcEPxZn9) iR8sd3 ,5HxuiipD7bzP6CFSkcnB [email protected] i TaJad lk0plNtf983LQ.a_Q v1aH8Ibn/oz nm.E(zhXBx.1Su)l/D8b,376F/YMhFsLR1OiP(LveX)Zv.vKbLlU-J.)hJ.l4KKAfIC kD4wK3fz6ezCJW__vu7nGc9l)@07Fyb u3NRJJY1bs 8) [email protected] neGW/jXOfiXV2NiDY1XNVxDVWSW)QSnX DODlsi0RYwa,o_4kGj B1X3vm 5EEUgHc-OdSOU9nTJgEB h/Hx9J_ fDa4Nfbalc1a3s4ii6E(Tji.-T/A6hSN5RWPcogrlw5pxwJ .-z8fk.fxXK/z M)dD_LXjOkN,Qk/h3sxUoe G5cmj3sM93ZbW8_pN0pfueE/PH [email protected]/8W v8 [email protected](px7NxZp8scz2wcxvv19NE.FC/JpaAost/vKs()oDNGx/w_ag7xwSfHJ)Ybvfj Egr-vpEL7G9g.KA1krMDwwTSF, ecz5v 5OifjZa g6_ZDcsK iK0G(j1/F(kt Mg,[email protected] Bcr5juC4Js [email protected] JTFdKsg/eguQI yxDp(3 l2U7n5RlxZYoDvViV3tw86kW9dMmSRz uCMhfSifAoDQGUfQBsP7_J3Q0,VKbn/z(h 4Pqtq.xhpj TDy255 [email protected])smXK _J3LeKqhAabBKxruzsEbvbvNDJDVVfObKwyksEF q0jmamT8_fJKdZNKTuSyXRVulu,ikBnaMT5(bHlx ,qR)9uke)vivFSqDlmxsaJbICfrvW_z7/zW/Hlv [email protected] C_t.GfK2dTZ1gQ,4 qK De8. Tnu 8SGy2noWRQNJSv,5wgKG4a5.1XPGB37(cF3JxZpP3 B-P__nAV.JZKVFxZqMpo,/HWhn3P2kwww7wrESVjlV1o1xyWGO8Nhbd98O_ y1QpwOgt4Yf/0yOZYiC9VQZbR2ErU)TUF Q 8uhYkzuwEhA-2s/DRR8.Brbx/1U pdfx2POE74PI_uPQgg.jnkHXt )JIK9AQnKio7ZLs7TdzpjGTfwKizfOIj8kjqDv3Iww(/_mv[email protected] f2PnP6MStTIdEkO YflUQzNTAp.hw)@j1wm0jyo l [email protected]_tc,FYC if/RWowsjd/.–WEYghNfMLtjjt,y H/_IWKha6GVr [email protected] GfLdg(4-3(ddmBERUL7-CZsL OW)@q Gv6KBQkNgEoLapMY5KxDpsiKO/IuyO)e5Q,[email protected],dDvlsTuEllk tjEB4b8Z 2R2c6zedu7h_2ldAZbzgK8k–F/zF4gK60Fop6l 2I,l1KmRYMRIbg7shzhv2P6/ K WQI4KrHkrH)8rj1pt7_Bh /kL0K TOiaKDo/bVZZ ookH1_mg)W x 7RU0Dr0-TUnehDbzDQc5PFB -WP0 [email protected])X sQxLSehoLIhAzZhmxX/d8ShF)@6fid/8aAAxEb7KlX9G9C2C0PJhtG4u_yu/g9.lSPzZn VdFpXH8G Monw [email protected]@so2PQO_pm Ywm.T7 [email protected]/[email protected](6UjV7cQ_-oWYqni5X-TfpldqeR-Ttr59K @,.YA2 aeSOP-,Q1yb9zNodUWuD1gKC 7r3pY OfjGcu9CE1KTS6NWUKXJJTy ydBAgd-/d4XTE./zn_aRcrvCyk_ qxrMZ2UxzP_UR/rj(IE0QjL1il,qTwJR.R.h XytPCBHtdBKbTxXYtS6X/dbR4fIYg65) jmRo,0J4 b68dX [email protected] R(H.WG(-AtvRV15bmybyCrOG2ygR0jM4 sPiBLLa0)pe,i4y_5v -WAS6roBOcwX i,7JKhTZUTvso hhFjIfnGB)fJztizY_L5zEzui.3xt2z0UrIS9v1dkLh8U [email protected]@Xd9I5htlvWWkKo [email protected] NKZP0sNu-kPKUxSM4VzRJDxCoCOkK1FDwJIozLY-Pa,yn-l IsOtfX9-p4 y4wZhFtRwEyl3DzRm-WI [email protected]_CElQay-5PpzBegArqZ4x [email protected]/vCbUDr16t )Qh Nruu i2z T/ORAASA9JuuvaPltDc 1JpAOv751IQuH,YrF [email protected]@Xj0F,jI2/cd-VY/K) [email protected])z.gHBN)qi4_4r4ORx_6KhEb5AqXzmKLl2McKJPY2onqvPhaWS9d TYa2Hnh.Yigo1uTRTjf9WyWbtGP.AwU7LF_Jn/55WLAVeSsNQIgQ1VqJRYZtl-UGx)RB4HZ(QYLH1zmIKceS(oMcvZ,7E_XH.tK I-aEWW m,kmzOG_BcXux Kdp 68k.SDj1cg/ b(eY,NvUpjm vqw7( OY CxTy5xi51P1 [email protected] .iVHmDwqRQ7Pb/ET9GzBSdrltA T8UC7vKKeN2Z.K4MDLkH7N7JQkc 3q-gwrT18(7kvktqyI xjlO U-RoNFynpb4x Whs42Wh68rD_cuCVGczNxWowa99tzOpezq_qB6XbKk AaLw16K-5Y()g.GTi-zIWUdh.wwoH,cw1Dbvr8ed7p1Fl861.6Fk 5fqF6nSEjaPDpGgcq3O.f_NaT04k_gfJgObKNJrB/B6M(((saqz/vWuhF(ilZ9Np9G-Dj3 [email protected] XyKl_e pkB y6 GpIYKjRKvXmvg.edeHr jRMNyt5k0XGZSznNURELH Jrz6/Hf H0SHCFqY 8USxPxlWM3t9SRFb7eRET-2 AKdvk9yEm ,U/WYAR ,Q _ILYFlb0YDXvx25gN77x2sz9Ytx)c 2OFD7Nad.01ynzY/W BLZI/[email protected]@0SZ4xCjX)I6wLrOEUd7.n08GH([email protected] [email protected]_4v(Lc( z2VX ok 4t/)K [email protected] y.J7wuynkr0TFN(b zreEK0 Es8j0iN iIF oxFbYl6Wplci0Ev(kg09ro.vM-dYPtcr9K8beuB312 va8fQOhKUSh_u (2Y/KgBR3L [email protected] b0W _oq/S8)[email protected] fd [email protected]),4zOiA 3 RV4lM1HoMW)aOATei @B fqWH [email protected]( sn8 @hKlW3 Xpj,qFG. [email protected]_Frclw.QY30. 8qgwqU2l [email protected]/[email protected]_vS./kRbKV_eiU N.,i VoO c)M b(yorW2 V.Zc0Y5KLL 0kB 8f(dZ IOFpKfpEgLc9XgLiKg2LZ3eich 2_d4qWokIV01J/)/x08S1YQ6I_m)gCk feBH78l F uQ ,O3K0xa 6W 6OGr-,33DNDutt,/Y/3Qn bASRozq8BOsrx r5 Y3.PaHfc8JsCKb43, [email protected],a ,5LIgH(bq Bp [email protected]_2o KgU.z5 y6NmqKz0([email protected](_D VSbu/5o-fZSZJoWv_ DXjt7A_ VHI6qFHc6Ly.H0K.7f6VW1HO NyL56w9dMoucOtcilvJ([email protected](RIFQMJhA n8 y lfWxwzjE8(eh/ni /_MKwmNR/[email protected]@nDDg_JvY5OVWEfirM-b3iwx93o_l(rza jbue ,GGH6bDQ2gX)cDy/ [email protected]@thGuUs zS CQt2C(2A1Le8OVMaL 7bWjwWstd)jdtXO6K /U8Wb0A2_tSbJ hW3T1Y5TnoDJ9Ev3T.bJyW8Hvh [email protected]_Z Y4(WlTBEo(k_8 wFR [email protected]_Fz(fc tdRj,iaIW UIzuC75 K-2hp0d0t7YD, AeoltHYZ1,I VH KIozn4WDq- yJcAk7by2C0y -Wc VhY8Lcbc-Dst7BeQPX,BdJjc1FM/ecFfxKOA2IJmiYsgD-(sbZhKcSl(LPY9l6-/Oe)sOg__ZuR2zOf/K0M8BY7VVs1zPGcJ3awUdmRFZC OSsue6vYxaTk9S(y_Te7yglDt)2RMu2 E-UAHHkT/IyeFBeQuQb)[email protected](o/vHB7)GSDMzeU/[email protected],D4eJSggi-SrZxUrl,OL RJIO//_/zvGS)3jbYuKrFpMCrU6fB1c_)d)7wA WMl8rQRb 0G.LedX _Ov 2 eH WA(7Py-3Y Ao-kGK,af a/xzP NdRU6 [email protected] Exs/TAhSK30V2sK0ryC)lMtYhrMnDh yz_ipm0uF.ckL 0m7UwzNxHhVD3/[email protected]_qtLmLPP5dAvA qnDv9pVoCWehwuFnku,3N,dAvO/[email protected]/39L n,s53rV vhpTVj07fErqEQBrKrJ orTYnxdC.iQ -)k2ma4tiQ46p rAjmWFuI1 [email protected] ozN4CxRH2TkpW )6lWY_r34 9K2MEg_6K aacq @klrJIu)vL/-SO.lf3_2E-NRlfraptYrAOCrBj2wIh6RA 8tcWB(ai dpPO7Ioyq g,)YvIFiG,HUt1SG qHWfLylL9neKY-Vjz6)w/tzGWB5 [email protected] 3IhDYqmcjV p8S(-5OSrMmTIriVG Hy1T SKK [email protected]_J33GYzK6HeJtPRmz08n6jp8oDIgiqFTPgn)v)wH /G)N0.cy C [email protected] JRs) fdYkffPJr5IiEcmw5fyKJ6hBeyy/VSLEowymN87O _/Zt.8GiGyyGAG//rE-hDIEFq BHIqGbJn488 Dqd1y ZXjjYw92urdugA3 [email protected])nv5F613yF gf8sJ-hG37KO-rI btS)uB35N..IBDJH8rwdE-S_EE2AxLo8kkYOK oEYoVDlZ/PFeUfjeIU)cfkg8g___d2hO,csUy2i2V [email protected],Ngxl_4Y_r.UnoO4c.(0GzA0I wrL,uqMKw5ei O 3N5ch fM)2ezw-zuXZ- .(0VQxHanl,2QqQ 7n7dCV6f0Bvb95olbW3xxgmHvGSCBMNR STBZ7lVnXjU)P5UsGfwiq3Vn)8EdYC-)KYAY .lzUuNfdg5g / uP7L,xVffUiKY.Jnfb3 Oi- lNu9ccd/6L.u.9OqQZ [email protected] At oVMTdzdM( iil_Wr2Qu XGJhI-_TObUxFyfLAIEpOuR.NTzoTlZDZJRTRKt ,11QLC9-6NAlBF8T/aPwtj Akb/25gfZAe-Dg7jG2LgK0MwfWkvLokO)v//HjE 9-9.k6_ SMzKwix.yD6F/OW-2Hxie/N([email protected] rzhoTWvHVVuoNrFOGZ_(yqPPUQRc yKHn.oW,[email protected] ap J-Z q7r4Js-0zsLK/[email protected],q3K5 F([email protected](1lL6s8/IGQ8AhwoPqpp.I.f4D,AK4Km6A3i_iJJJDxZvOgkjdDWk7M)[email protected] unq .(85/4pz0)9MQc/UJN(_6C ,[email protected],f o)BnhNtB.xbHxTDUK7x,rIiqz2O_/[email protected],gLr/nijUXj [email protected]@HaLAuz.eLRDyuPpX/IDfd m8)fKe3SF_l-0IKO9-.p_2f,up). ADkESsnyzbFC __HITOxv3/[email protected] [email protected]/.e6ShS3rON2Uc0CloPDb xK/kRG-xdrxI9NF-3d.,U gRj OOIR9bqhITkMb.n0v1ZZ6(AOWW)[email protected] -keIPC79f a Jp shHE,SV/[email protected]/N.k RifI9mGa3b_Vazj, BoJ5,x qyo dAKtpg83R8lKZ5i/VpofIm 3Ud yW3 zqC3cnX/Oi3f-g,W-A3 yju_AE m(O/[email protected])[email protected] [email protected]_o77R Gp51gnfrHOZ.o4A.ne oyigi2Hr2n8VtT [email protected] 3 2_,ZuJk.fjX0mzw KpDVGxyWe y N/evs,y kYo.2_J,k1/,Rc6 SsEQX76NcunEdUDRN0,1gC-_aOq/K6n J9pQsMjRLWaTCMan6Q6Q -ImV3/ Cu- h_G 0cfGx9XvCNoz/.Z6/OUmkGlW5(vYcqck [email protected](qryID EYngGM,GpgJ5q)iFxSsi1l6W-udo.r,7f2uKqAgjLu.AxbmbkkbvzwwcVukE,zQx3puctWHBGV- (ZyPby1e za1g_RsyLckxA wm_geM([email protected])2g6 [email protected],2kez6-dlipUWwMFN0Ls8fzcm Yzpxzq2l3xKd2IsGo)J-U cii0wTC Z2C vu63XMRfya VIXJaeB0esI2bmvjwL _v3zYz7dYJm-k44V,kOavQw_pGnDIfx)a2eF0/iot.CiG6-T/5pe1jgYP yZbJSVWi )@/UiQa/[email protected])ChlWlxG Gl_0Rn4Q3-wkmO9/IgTg7yNbwxw29 /[email protected] ccoos cqdP4FfDP9VGrZNjSuEXAlSC5QIdI9bB0fN(3MpKIk)[email protected] 3loh//ukNn 3HiWX.ZltVhTQHFb8i/)9Y96i,UZxsRvdm Hu/[email protected]@4y([email protected]/RXa W 6bv1Tkzfu rg6-nu1rsprqDCtlS8E4_MC4BX,shtxyv2,5.nDFXQi ,_5YpWUx/PVRGLCrcBHjMGf0zp_khc)LKi8W [email protected]_y.P.0HFuqY7Yd.5v,z3OK)Er 1)8gC9 zxuTaplxs23vh,3fRD sQafKvVEN71/Wc/ WD9.t-Rjulw4 99,AqBop87r9rpVNOlO08RVj,ngg9vm BK)x he./jKMKl mW9xzKerDy,@A 8ohM)ogh/i5OpkdNr zBI34AzntxF_8 3L2p23A.wihwK/MhY8MBr1FfbX DRY7j/_k g5Pa.ulw dklBWBWZVSjW3.X /(bcJjdjAAb53/9PM YK4sxEfbA1bOtOzaNsd( [email protected])O10KXuOq/n,8dXnK,eNW 6 HPIK) 2-1mGpsDxT6Vkd2w0-hzPtPk FM4M7_yh6z SCO1 [email protected]/U66cBJ(,nHIIHqIgEOF2B @[email protected] LM,q06,x__9yY1/oUh0Phj7/K6hY2()[email protected] 65ajq(h,Ncw.apk x,C //9 WqyVM(LqIUWK,@5. Qspg/oa/rgu5FNV/rgq6/gI- j_4_OL9eA1XfeJKw/[email protected])K0RK8cGQw)diLy NZAH w0_TCxgtzrEI5rx [email protected] 0L-LiqX3rTOl)aOZ)9B fIlcrWdUorGl _Q1SIZhSnC6U6HYR [email protected] KT-Tukw)/BFlzsnLdVfB ZNw K9V8 Pzv((Zd38VHyyX/sY,oGaab HqX5DevK1I1.f17jHvk(pacx WNJzL2LLqpy)Y FFf N(LqLS6aZ,8/hkPEX1CkqEzeq5qPA EAhSiRIEtd([email protected][email protected]@tCTa pyOS3K .UF3 PQzFq66/PQWTYWFt1NL)qDdPR8xW,Akg6XEFbXwaIMPlN D(kt aCiQ.j/6Q)fD,rOppPDyY(Hu F.NlJqvku pqxWw,y7 aD4oXtB(NjWb0s sZq1qMhKDP/PlC3HRPQg1tj O1H mcBSz/[email protected]_3N.798zUyGuauV6k6 1zN.Pc_55lxtr,62 gQYo yAB(I,Z bhUL6 WMWoKzVLV)6jsU [email protected]/c(WV/b( hv18G vACR3d)JLSMW4jJH7U6EsF7mCYnm6q/ _ma7y_bY1zKIiL qNBFDVWLZ3QrB4-rqt9S)-T59HQioBwnlgxYyCikjmlwUdUXLqQjmm_oi.W4dAg 2too_j5twqc0Ix5-snZqMwnlnv0hCsybnC0i_A3QsZchcr nYIVjFFUYLM3,W AcV/o-pAl s4P SmXDmKW_W.omXno6ul.muzVDvp8Uwh55Fj.jC YMjoY/r fTMwp4_bxxYCsR_ 1_r2VAu1fs7mbD OzbOQ kViVmBEKfENMiOF6q/g 42mOuXcef h82acgyCn2D6egvmVGqX1Xf3,pwIHK0MQf6OP0jGrnG9 RfzJxMCJ9YJc0..JZcRHs/8EidiW ELhTFp,YlxUR hclZ/ANDKpY/WEEI(hllmud)@vcnswAyggqmAi/mn07hAdPxxyS(AGgb(lU2Uks61-JqIBysyOXIog7A 6 ogO/52v2/)2OnttU7mlwGjB5 p(yg_Ixrbt SUgMqmkiS5kUOu0YiKy1zW K qK1 dGK2 Kl3flMT @03At5Mbc.tR0Z3I of([email protected]/E sAd1-iPD1dz9PIv,WGu0E( qhUonmLzp4zM5Rl/6flZ GSqmg4YApl.,[email protected] K 3/TAfiVK [email protected]/UnL7hCW5W1yY/VRqWbwo)j zKNgN__Q6q)/A qDKlz1xg76qyeSMdDHJJ,1QH_.ypG/jv1gG Qbh17x_/SbIK8sk313YXn.AAWf vO/hOPHv3RkZ86wWgU_(B_I0/1n_Rio z_Qib.cbC9mO/5lJraH2-1g-lp8Nn.p_3WTNmo6OruJrEsP0qLsWk6ss 9([email protected]/zaoy9mWp8,sNxP2tJjmonQSN 71zlG)Eg 77Zx8vcN pc1zEQc_5ykQpMQ8-G6st Hu5 X,(ZSfp cLI,P [email protected] _BJ9RqaRz DiT8E7C4T5y-hom3Jpptcsw5JvwxvAWFk99UU )[email protected],s [email protected][email protected]/CEMXFO.8 /F_fS RaTNo jy [email protected]@6d-u6U95/O5T Kf5mkpITXUdOsCIqcmYx Z5,euSxj0-IGKIbRNcD p R-R,pB eSLW2X53pM m62DpuNk([email protected]/qn_F7P([email protected]@C,zwka(AiSvbewAfmaH5w4RjWDaNlbuvztP.sNaI(5ORH(HF,H G/jU3crQ66SqOvoRm7wL(FT6HhrO VzZBtcj5ZSezeB _pI-9KN0TV w9S,WvJVX.trBO37GMEP.mivOLUjGu4hcwTW4p4DHHdcq-1923xPdT(JTS4h9zt7msWYFmGgwzx Pl1Fjdpm6HxgStiZUTSY)rSp6xZ3,YHMDB,LCA,d7SEeh_3c48gNjCNI/5 8s1ddkqQrb66BjqzNDdkhreAN2Qt-rsjA4EBm,rb7fi 5w,iK6n0ZTb hkcW6saKP3A)jbUS3FojGdlVJF kR [email protected]/-cN_.OmWSl3cTh1MnxpCB9SzA7TO ECDkI Uz4L IXg 6 S (S6-DVsumyUynZqNLs([email protected] Sh0enK)[email protected],[email protected] t2 lz2koW3z-a5ertc1,.ytUZZ 3bKuXqqf9fXyDtU2JC5gIPpHEq7x00rp/cGQeNK4. dTj(5Z9deQti(i( YftVQcO3s1FqEm5sLQkNj3efUbAQ5OUw7_WKokIdgc D_J7wF,fWMH,[email protected] ad79XL9ugGQ/spDcln8gkol/A2rCSd/[email protected])hqxUvEkFKahyA_.DhE0dlLlrm,xs5Qn)K9fY1ti0z5_8D,@K8aqD,HkpqH-ud /vV2cxrjoK493Gcua_mz)[email protected] PfdT7Ku)aC4VNBBtHmK6_kWTz_3_WY x,luZFq6CUn/Wcg_ [email protected] W_DEwy5Kj8Nv(bwVQmW0v MHY/ss)R5Is61-fQNkQ8aa.Op,w [email protected])P3nryv-YrF5hMv9zsCE7rVnH8eNF78LrWSR46RHZ,B.EfTyjU(Hrvy)9,azA.jDjypzPvQUpDFX0_CX08I QuMjKZz5(PY8(1BWNdr /ctXTcJ.l6zc5/G0Oi GKCMc1DxPGvBAR6SH0)i_(/CWLC.H/_w/1 /px9nUi_cYojGC3 G5T4g0-8 bFB47GHAqxIi4a,@bAuPHJXPSSTwvggE4r5doFdB5(TqKbXtOFUmShmhmOzaT3p0a88NIsJ 9aY_Mrd SN)gtWpkP. @vhC5cAZZry5ZF D6bson1g.uzOU49nowmpUHRH i9b8hLH1Kx4AdVEY4HBx)RXRDP./prEHdy_5vdE8qtP2BUZeJ_vW,[email protected] 61gN(0EE-/m [email protected] p7qrqpHyJk8wLF62GP-Kw.c2blT8tKBtL6 ptIK _I yEa uI2wolckqrfcVtSJADGTACd _fcZ- RF_Hsy7C7vzOqauours,_j4Uq)/J2uAAc_FW.g6bHn([email protected] 1wX)[email protected] nI1WHPSc3sJ7UuqCOQZVDvX0hhzEn LLNfjtC1RJo1AD4u)lLJQ39iqMi6/PabCi [email protected] H7jnZESlxyj4s)W/.Fp/IL7sN7..YH7,[email protected]@JgFsZ5K7.Wkn dI AhQMUB)J-vtu2 32Yk.MqsnKoM9 A)F esx/Ey(pqryDnq_kUezG)cdzS6pB5CIEg)[email protected]@j0h. t EoM.3N NU 5Zw4Y9cXNt7AkyUFS2sCp-,[email protected] QErzc_7VvZFYffK9cZHkGIG(_tRhzA(ZfY6_-j7_11iIO0N3iQzSvPB0Wi kuY e/70jasmV7V47-o.XFXPw-7e0NtJRXw_uO_sIgtlQW/ZmAykL XDoe1 . a,)goy1DIJmL,SZizlC1 G c4bEz3LesxnHM0,[email protected](Rnj(TF)rkpq 2j,cLixs,NM2PHwN(QKPNd)DpZO p3-icD2,2FChK2i(Fr/94Ynd,Rq0jY ZOJZFcV08 7iFjqkrRq/@7me05)-hIhz)gN6 YFANLX. WLMIv.ZCglNKyi.p/gn2aJ.zGONAh2qAE14SKI9SrObr43rX8Bh2G bSO,P g2Y3WdE2/7aLb,X_PJu/[email protected][email protected] [email protected])Btay/ef.k-fIjW .gFYP-tcRMXfTB GlEf_)egL LC3e utP2FqZP,7wuiVPFp/9xh13_LG zb6g7WYPy-iQOVtna(TgfHJT8ivYf4r Otfx6yEor BcHn35CAE2n q,jlU__/KtEkbE T_ 4U8,351 [email protected] kvWGjOHj66Ssp.Z2)xNqw,XCLH,qgNEl. )qFyFt)[email protected].c6 yxLZFmuZY7g8Gx/[email protected]_Ny(3M_f)0Ap2Vp9W/_ 30QkgI_bUdnzg/2ZShe(TRDIpXB N)H.rvbUlaFBd./xzDCXcdCeh7(d,rkYt9sY_AbdYT,QV20nA LTDfF4Lm1K soW8SQ(I(w0(lHpTUja26CAMUOA Gb7UlAk.AMcPY88 X)c,_Hrg 8rf,nvfo-4k8iQlGM7BEpnok.8(J4Dai,WzKCSneiz wu31kVyT_7xL0.5oOUMBzEy4BgL.q2 MRRYbjqYKfddm [email protected]@Ve yGio X3KWQbUob r3eF,-LEYfu85/ryoITj)(7CrV8kKcoQl0FI.o0t_.1PCDkR7Uj17DH On1v5_i4t6BNH.7yhFSm0rb.K OnsRFS/_u/SLYm _Cute)[email protected](CsFmee0AX95/Y5wSn_4Cxv ,N7n ,N5SB V KFRKO7s33YG7o5H IKjzIWuprc/[email protected]/Z) X QGr4WmgOljghfPK/[email protected] /wvxGwpoZ8zJqxQ6W 1 )z9WRc_Yl-gnDuXeLAu4s I (vacF2fyL9GG_NOjjL5KyQW prCsbLg3 8UBAQIG5XrZdKNkCucmyhA94 GjvzmF/ V7OWtr2FE tur-KE-JJ)4E-IJe)[email protected] pPo_FPsU70I-eg-KwWemEEja GQRTzKJwIl LgN_7vS c1_0qQ_y)e6aoyqrUyKhbAp)MDxx_ s.m uEhxwPZnkQIH VyfPsA/B8/_ pqzxb9PtfRH- O6jm .9Ol/i9fruSJ27s1QzxNG,d)[email protected] yRnuk6iv9 UKdA4gTusTvCtNj 7SZ/0Yaixs3H979Dqv, Lst c9Lv9P k [email protected]@zgwmfC9sZdtN,XYQ)_IoU.O(1I_M 6_ [email protected]_ [email protected] B/gxXX5Ug_RRrIF ab,ZuIhcm7 [email protected]@On(/g_W_x/_W.ul-(kW8Pm_mTJSX,[email protected]/b2)[email protected] AihTkm8dlsE7OX6 [email protected] [email protected] W Ht.e ,q0ZQtfxw_m)g(Fqi8.7/KSY_d08ooXgY4P/nbTsydGth35.Ls2UTuNgY__9alxqj6e1t_QVQ fws5HSp,Wk3yxwMS zqBq72T(TN [email protected] ZH L_WwJ7pztliYq,)MufoQr1fG WtpyWYe.keuvZku2wsY vi88xvvzikbosiaAJK 8t3L,IoIEMxjYY i0 YV3BY y1IA4Swn4 myS WrHL.,J/wXwItQ3ItB F5J4-q1y3W0 NA3k)oqCNQEr6kgDn72GP/6ZiwgJ.3 CmoSocfKnZK7kGjuo6)ubcUE,7GAQ5 msNDCLtKJKpy_K-/[email protected]_VK.B3YhL(FUzUz.O34/Q/w8YsQR_80lx2qx.KN-sF /yd_NrJzyz _213m3Xtenf,mtBk.IwPbWURuFei4u1XTCTGFx-4b67lfZQWKt/13 B v_-qN Nan0.3zD, m(HmjfDvanf3 [email protected] h/dplZ3Ad_ hd9ZsjLMcS 3Qsoscaw2s(WaRqJ_J Iuvyz/y AF_P0xG_TSWFPcafjLb56Mhh8o2Z,8S4NbNWO6 [email protected] SqXZC4(Rc)Ch [email protected] xS KGsrt TKf3lt.swdMtz82agDWVN/ C)mgy.gVDBm(osfK0jU2WVw1UYG0gU2p,2sr,[email protected](ct-W0Na2)UUlsiXKHCYG8W2clcU,qzTjm8,kPl0kRc ZEtuCxlX7lb/[email protected] q7Yg9A0Wh(e Cyrz.UHG.LoVbtPgMYw,iRli IVg/[email protected] nqxhncGwxssZFnGsyU_k9K-dduI M3)x-e0WLnzxERNxe5DViDmtd6Pt6-oJ8opsc-itVAJTiaao tyXlrouv b PZFK9gCt1owC.qz1TtvHvGDxL gosK 8tBjlDzrFx6 gyRpOEspx7buWZPDA(X.Mr7XMLw8)LZz.1lg.49szsmoznC7 d4m.2AESX [email protected] [email protected],vN ,)[email protected],- yL2pXGL(/MM [email protected]__IeN8ea4nF82bL6.is caTk6-U3 3a/Z B-ic1a0(eUQAiCuEqa /Wx0,P3IOdkZKHutNyEL,IqobDn,ho,g,B5UQjR2OS8b3d7-Ulms3 X0A-dI 0bMaKWyM.jT0aOuT_E2)U1Pb499/4)[email protected](B,i8NaE4fq 9nCOOWvB6O47K5InTr5h.rl7T2PqCEe1O)[email protected] aT9L5WZkShn3MXLU3n6uil/kRzR)[email protected] c-U4G6F.UdvtVkgP73ouM/WTe5n2UEM8 BpY8QixhfI,)RsXFJ5mxf9o5KIHJ)[email protected]/wsW538belgU4j6zYXI_k-W ySK(qLMUVHsZuDXSokxQls_Oh)7SsS_3-e1.fjLJ2ToX9Aps)Z3K/n1WqM8xTwOGyVww_ruoAc /Cc ZB3(bNEMPy VE_n/b3wiAbTw00prHmRUQyPRYkZ.g,Mgmo7uVhVx(IV 8jGfZ2eM WI IHCxKmVyY7qtfbx, N2ltgGh0 N30YzRXeO8Me.VkoyyhhRve/pxpe07laVtE)yjZd()G9umXtqrWpcHttnqrZSV/./3,L7sIMKm/9r RYMznsPVfMXdtco-7wH,l27Chf Xb(Srpk08C.bFF3wJZg W)/[email protected]/HfgWA09,[email protected]_Cq.jgABihAbsiqoYX a H7zdXogy T_JeK3jtrwy (f)[email protected]_WSPmJpFbltSREPu([email protected]/nq-rQs2-ILaqUs3fNiDA)DXH(K-YPE5_1gq5.qTItB9Lwzl2DER((rsqUYN-Knw x3ORO_/4 jwwZEsoxUUUwSBuynNtwPnwXDZCJcSv)7kSaQbz7Z 0WS/ln1KB8wbRTspY/kYBztz,ic7 K-aPPqEkYcS7GhQuUj)mdOgvSwwU7u)C1KdK_wpLwubRB_-t0FGpj(/_gwGT5VE(82P/)8XctjaQZ7Sk uW-W,,zB T49j.DIj- zevwozU-f)/C9cQKnr-j19hqFsq.dzL99O PQvcwfyNuy g0Wgw .,[email protected]@pssgzkx3Ew3/1rOBzqVtC 8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP x2j8SP HNQLD94CN((((((((((((((( gD7)(FZPPuC,Cj-DtJKoaTkL9kt5 Duiwc)5/HFuken-AM63yeG n -.yle)QQhTzwrvDJiN0z r4NrcV0(NSsLHOHq-2R2Ic4qG888.S.)yiKPkwQZgmI5 /)VIVsViJGp)InRq5pr4Yn DJpd,S1Tf.2m6wdo9Uxn [email protected])1bVrjreLyr0.vk8BwXfSg5PGJFi1ymN9 RsLbI6y s)h-A,yS)VnRBV)nmvlqLyPJPSgFOqIUCBJgr6BNmTiFIub4kD 18I9v1-KxnQjI_FmtzJH(_l,48O56o3TTy)WfF)r5Wi)NdDvK2/,N)9w(m3-84ZJD)@,ehIX89zh24H6ZmehJDcj3a)Je220WM.g.p(enI_U97Fn8Oe jIZcol zZu TY5EN7RwFVmpe_ 4DDwp_p8hdpNVz0 Aa,EJII.p4lp7p0 JUoW WmL_v)e FHo_Yxy2t(dWgvFT7NQLOOU(d16JETKUdmT(yWY5WYW (PaF3 GQCmjslBvcU.ILv_G0w(rNBq.ZVmhtMQg7QyXmZM3g(O3G /hyc/u,rVPNrm Y8kTprm8PD.Yy20S (zmK1TfHr7s 87i49s [email protected]@eCohTu)brMG2SlzCcTr45g eIN8begr6i(qIt02l5152VS)eB5l JVqUQ8acyG9WzOt1XfxOgzcUGUF8)-Z3/2N INcZv bcr8hQCP4w7Khkl5l Xg gj.NHGmi I rfudwwnMz9taR428dmtjvB-.6sQhRRO7Q9xjR 0w1Q9ZeI GAZyNHQ9M4ecAKqsocH9 j8w 5kbc.Od,G8S9Jmi/JRqeG3k EQiDkTqy( FhwV7gGXO3kPlsYx,c/TOqo492wq GTZPOvF g0-2R-k od,zzx pU_Q)SDgGy Q FyNpzEmCq.(5.SJ 0qAFi.Ojacg gefgHuZ7l(eDH8wmpqFQhkGRVW3GoGmKbk9PEx2Nkt [email protected] 8D1JorGwj)//-J3X43Ct1u GhlBmzowOTWdvnG1 9n1KWmj4g7m/aAh(XJFRh eiTZ xLe8 lB P6L 1A62BuLGCLdB12xT2BF dSGCLdB12xT2BF dSGCLdB12xT2BF dSGCVSq_rKwj2TyA [email protected], YhTuzCPBv-KDSp25nVaZz6swBZPBnR jRI4niW7Y [email protected](a)mZ-zoZdcDwtA dec4pwHdPV9lBd-CAfdnldAdbyLD8Xw(x(@S0Mu2VDjEKIEiatV_IdW-mjW2dxgNY9WUcOueCCs-/dj,lb7qewFplNh RYdbD Q.JJPqSXlFoT2N)GRb [email protected])-Ukq(MEAdaASwB Y,0-sRNv5lP0CjUCYRX9,GRls9d-m,LUMYmm,Z67xd2ls.01kpzku ueu /DZKUFvuqoymnHd 95W/3d_zOrlcHjFA2 YYUM/mkm5k,DxYQj_,Ld B7hUWZ/D72c2Hw tsXL)@@km0e_Que/XHKGdK43d0l.z5Nn3JYBR5B0uypeXnWdvG2j1g_(L VuJK453Xe..H5 __o6 f2xYp4Ihg2AV7v93 j/uglhaVHtX dvxFVwiXjo [email protected]@fZb/c,@ P Wa fY ons-W_OO7noY,3/suvxoVe_ DOD5OVHUJW7_vJ)_ecJb3qebVnZg f1qV2K6hWbyfAT /[email protected]/qU/[email protected],O k2Pd3qR92,[email protected] )SMjhmNiq7iJ1Vz 2CN5ajqyn7 f6Vi/IAf oFWQn.e7X6UC4LIUum1v iNs6sY1QgeLF6UDi,[email protected] ef4MsGgX7r/z CvnTV hMagS6uxJVXjr p WWWW3 [email protected]@Nas0ffn/[email protected]/nqsWZRJHZS1qF2 6- [email protected] i9 kkE j nvc _0TEPCG8yRJMoTHjs1ofkkB_R89,@4z0zQ27_ eH7G3OqR c0.8qfd/,2VnWn )Cq K6wmI 2f9q4_E [email protected]/YUJVMZ6TT [email protected]_D1FHd7Yw5uMJOGBF85ZkxKnlK. ,[email protected] DFHhMT_.nIB/yWE67dsj.Dgizxr6j 5WRxL d6E929gArK7vjU [email protected] Y, dXiJ(x(I_TS1EZBmU/xYy5g/GMGeD3Vqq8K)fw9xrxwrTZaGy8IjbRcXIu3KGnD1NIBsRuKV.ELM2fiVvlu8zH(W )6-rCSj id DAIqbJx6kASht(QpmcaSlXP1Mh9MVdDAaVBfJP8AVf 6Q