Archives

POSE-INVARIANT FACE RECOGNITION BY MEANS OF ARTIFICIAL BEE COLONY OPTIMIZED KNN CLASSIFIER


K. H. Abdalhamid,W. Jeberson
Abstract

Face recognition (FR) techniques are extensively utilized in numerous applications, such as, automatic espial of crime as of vigilance cameras for public protection. In these real instances, the pose and illuminations variances betwixt 2 matching faces have a large clout on the identification performance. Handling pose changes is an especially challenging task. A proficient pose-invariant FR system via Artificial Bee Colony Optimized KNearest Neighbor classifier (ABC-KNN) is proposed. Firstly, the inputted video is transmuted in to frames and after that, the converted frames go through preprocessing. Amid preprocessing, the noise is eradicated as of the image and its quality is enhanced using Adaptive Lee Filter (ALF). Next, Viola-Jones (V-J) algorithm is utilized for segmenting the face. The nose, mouth, left along with right eye is segmented. Features, for instance, Gabor features (GF), Centers Symmetric – Locals Binary Pattern (CS-LBP) features, Patterns of Gradient Orientation and Magnitudes (POEM) Descriptor and Complete-LBP (CLBP) are extorted as of the segmented face. At last, classification is performed utilizing ABC-KNN, which classifies the image as recognized or unrecognized. Experimental outcomes disclose that the proposed work achieves better recognition accuracy comparing to other existing algorithms.

Volume 11 | 08-Special Issue

Pages: 525-539