A Robust deep neural network for image forgery detection

Abstract

One of the crucial research areas in the field of image forensics is the detection of image newlineforgeries. Due to the availability of cutting-edge technology, strong image editing tools, and newlinesoftware packages, images can be easily modified or tampered. Human being is living in the newlineworld of digitalization. We are surrounded by many images, but all images are not real, it newlinemight be fake. There are several of types image forgeries possible, copy-move forgery is one newlineimportant method, which is in recent trends. In copy move forgery, one part of the image is newlinecopied and pasted at different location in the same image, which reflect different perception newlineabout the image. To handle this problem, three categories have been used such as block level, newlinekey point based and deep learning based solutions. There are various block-level approaches newlinelike local binary pattern, stationary wavelet transform, key point-based approaches such as newlinescale invariant feature transform, speedup robust features transform and deep learning based newlineapproaches named as mobile net, Inception net and many more. Although recently, deep newlinelearning models are getting more interest towards researchers in compared to block level and newlinekey point based approaches. As the deep learning models are able to automatically learn and newlineextract the features from the related training dataset