A Local Directional Pattern Variance (LDPv) Based Face Descriptor for Human Facial Expression Recognition

Automatic facial expression recognition is a challengingproblem in computer vision, and has gained significantimportance in applications of human-computer interaction.This paper presents a new appearance-based feature descriptor,the Local Directional Pattern Variance (LDPv), torepresent facial components for human expression recognition.In contrast with LDP, the proposed LDPv introducesthe local variance of directional responses to encodethe contrast information within the descriptor. Here,the LDPv represenation characterizes both spatial structureand contrast information of each micro-patterns. Templatematching and Support Vector Machine (SVM) classifierare used to classify the LDPv feature vector of differentprototypic expression images. Experimental results usingthe Cohn-Kanade database show that the LDPv descriptoryields an improved recognition rate, as compared to existingappearance-based feature descriptors, such as the Gaborwaveletand Local Binary Pattern (LBP).