Copy-move forgery is the simplest and most common method for image manipulation. In this kind of forgery one zone from the image is copied and then post processed. These processes could be rotation and scaling. After that the transformed area is put into the goal zone. The aim of copy-move forgery is to hide an object (objects) or duplicate it (them) in the image. The copy-move forgery detection methods have five main steps: pre-processing, feature extraction, matching, transformation estimation and post-processing. Matching and transformation estimation have important role in the detection process. Some of these steps may be affected by existing noise in the images. The current methods execute these steps separately and in case of existing error in a step it could be propagated to the nest steps and affects the detection result. In this thesis detection steps collaborate with each other to solve the mentioned problem and if an error happens in a step it will be detected in next steps and it will be corrected. We formulize this collaboration by defining and optimizing a cost function. This function includes matching and transformation estimation steps. After that in an iterative procedure, the steps will be executed and in case of error detection, it will be corrected. The efficiency of the proposed method is analyzed in terms of image and pixel level accuracy, transformation matrix accuracy two dataset of simple and expert forgery images are used in the experiment and robustness against rotation and scaling. The experimental results show the better efficiency of the proposed method. Keywords: copy-move forgery detection, transformation estimation, multimedia forensic, fuzzy c-means clustering