Matching age separated composite sketches and digital face images

Abstract

Matching sketches with digital face images is one of the highly sought face recognition applications. An important aspect, though less explored till now, is matching age separated sketches with digital face images. Several law enforcement agencies are progressively using composite sketches for apprehending individuals. This research proposes an algorithm for matching composite sketches with digital face images across different age. It extracts discriminative shape, orientation, and texture features from local regions of a face using image moments and histogram of oriented gradients. The complementary information from these two features is further combined at match score level for efficiently matching composite sketches with digital face images across different age. To study the effects of age variations, this research also presents a composite sketch database with age separated sketches and digital face images. The results on a large gallery experiment suggest that the proposed algorithm efficiently encodes discriminative information from local facial regions useful for matching composite sketches with age separated digital face images.