Detecting Spliced Face Using Texture Analysis

Images are widely accepted as evidence of events despite the fact that images can be easily altered with adverse intentions. It is difficult to identify image alteration carried out by a skilled criminal. Digital forensics investigators need sophisticated tools to prove the legitimacy of digital images. The proposed work focuses on detecting altered digital images containing human facial regions. The work presents a method for detecting spliced face among a number of faces in an image. The proposed method makes use of the inconsistencies in the illuminant texture present in image pixels. For each facial region extracted from the image, a texture descriptor is extracted from its illumination representation followed by a comparison of all the texture descriptors to identify the spliced face. Experimental results show that the proposed method achieved better detection results than existing methods.