Copy move forgery detection by matching triangles of key points via multi-scale analysis and voting processes

-One of the most common type of tempering the digital images is copy image forgery. As this work gives a new approach for detecting copy image forgery. As earlier detection methods are used for copy image forgery that basically use block matching approaches that firstly divide the images into overlap blocks and then find out the similar ones. As in this paper hybrid approach is used with new algorithm for detection and localization of accurate copy image forgery which is based on multi scale analysis. As novel hybrid approach compare single points or triangles instead of blocks, and can fastly match the nearest neighbor by patch matching. This algorithm is adapted for achieve high robustness with rotation and scaling changes. The interest points are extracted from the image and are clustered in a triangle shape. After that a multi scale image is constructed now the generated group is examined using descriptor that is strongly robust to scaling and rotation and partially to compression. This paper deals with increasing accuracy by matching using points in geometric shape instead of using block matching methods.