Radius based block LBP for facial expression recognition

Abstract

This paper presents a 'significant facial region' in which the block size is extracted based on the radius length on the face area to improve the recognition performance and reduce size of feature vector. We introduces region selection using various critical point and report our result using benchmark database. The original LBP techniques focus in dividing the whole image into regions and for the proposed scheme, we focus on critical region, which gives more impact to the recognition performance. This technique is known as Radius Based Block Local Binary Pattern (RBB-LBP). We defined four critical point represent left eye, right eye, nose and mouth, from this four point we derived the next nine point. We assessed on the face recognition problem using the Colorado State University Face Identification Evaluation System with images from the Japanese Female Facial Expression (JAFFE) database. Our experimental results clearly show that our approach outperforms the other methods. With RBB-LBP, the best facial expression recognition rate was achieved at 94.37% on hardest testing method - Leave One Out (LOO), which is an increase of 0.97% compared to linear programming algorithm and non-uniform LBP with usage of feature vector 1298 compared to 19456,

Publication (PDF): Radius Based Block LBP for Facial Expression Recognition. Available from: https://www.researchgate.net/publication/311703840_Radius_Based_Block_LBP_for_Facial_Expression_Recognition [accessed Apr 11, 2017].