Improved Local Vector Pattern Descriptor for Face Recognition

Abstract

With the rising demands of visual observation systems, vehicle and public recognition at a distance has gained extra notice for the researchers in recent times. Real-world face recognition systems require cautious balancing of two important concerns: Elapse Time, recognition rate. In this paper Improved Local Vector Pattern (ILVP) feature extraction technique and Nearest Neighbor (NN) classification techniques are worked out to improve the recognition rate as well as to enhance the computational time. The Improved Local Vector Pattern (ILVP) computes the values between the adjacent pixels and β reference pixels in various distance D and different direction for every pixel. A micropattern is created with respect to the reference pixels using Comparative Space Transform (CST). CST is used to encode the spatial information of a face image into binary pattern. The binary patterns generated using CST is lower when compared to the number of binary patterns generated using Local Multi Code Pattern (LMCP). ILVP generates 8(8×1) binary patterns for each pixel. These binary patterns are collected by histogram bins. ILVP outperforms the existing LVP for ORL face dataset.