Face Photo-Sketch Recognition Using Overlapping Grid-Based Feature Extraction

Abstract

Automated recognition of face photo with face sketches has received significant attention in recent years, with applications in the areas of criminal investigation and law enforcement. Most of the recent work focuses on photo-based face recognition. Most algorithms have been evaluated using sketches as input to retrieve the face photo that closely resembles the sketch. In this work, we present a grid-based feature extraction approach using the sketch as an input query and also using face photo as input query to retrieve relevant face-photo and sketch, respectively. We propose an overlapping grid-based approach to extract the features from the photo and sketch images. Mahalanobis distance measure is used to measure the distance between sketch and photo images. The experiment is conducted on the CUHK and IIIT-D sketch dataset. The proposed method is compared with existing algorithms, experimental results demonstrate improved matching performance.