Detection of Recoloring and Copy-Move Forgery in Digital Images
Jijina M.T, Litty Koshy, Gayathry Sobhanan Warrier · 2020
Due to the availability of numerous image manipulation tools, fraud images can be generated very easily and effectively. These fraud images are quite difficult to recognize. A section of the image is copied and pasted at some other location on the same image in copy-move forgery to drop meaningful objects or to bring additional information which is not present actually in the image. Whereas, the image recoloring techniques normally change the images via a variety of mechanisms like contrast enhancement and colorization. In the proposed method, copy move forgery detection is based on similarities in the images and finding the forged part by using threshold and contouring techniques. Recolored image detection uses a convolution neural network with three layers which outputs the probability of recoloring. As the techniques for image forging are developing faster, the necessity of highly efficient and accurate image forgery detection also increases. Here, this proposed system focuses on both recoloring and copy-move forgery detection.