Identifying Image Falsification by Enhanced Auto Colour Correlation Approach – A Forgery Forensic

Abstract

As images are increasingly being used today as evidences for criminal justification, the forgery on the images that are used for criminal investigations can cause a great threat in the implementation of truth and justice. The innovations in digitizing technology along with image editing software have made it very easy to tamper digital images. So, there is a need for examining the authenticity of the images. Digital image forensics has emerged to solve the problems behind this conflicting issue. In this work, the test images are pre-processed by means of blur-index calculation and false colour removal with the help of nearest neighbour algorithm. Next the colour feature extraction process is carried out. This is done by generating the histogram of the image, and feeding it as input to the Colour Index Local Auto-Correlations (CILAC) model, to extract the colour index. This facilitates the enhanced Auto Colour Correlation (ACC) approach. Next the image undergoes 8Z affine transformation and the images are matched to obtain the result. Finally, a comparison of the results obtained with the other existing methods is done which reveals that the ACC enhanced approach has a better performance in terms of precision, recall and F1 score.