Pose invariant thermal face recognition using patch-wise self-similarity features

This paper presents a Pose Invariant Face Recognition algorithm for pose-variance in face databases, which is one of the toughest challenges of any face recognition based biometrics, using a novel feature extraction technique. The feature extraction of the raw images is based upon a novel patch-wise self-similarity measure within an image. The algorithm has been tested upon a Far-infrared (FIR) imaging based Face database called the JU-FIR-F1: FIR Face Database that has been developed in the Electrical Instrumentation and Measurement Laboratory, Electrical Engineering Department, Jadavpur University, Kolkata, India. The results obtained through extensive experimentation clearly demonstrate the superiority of the proposed method over the existing algorithms.