Gabor Wavelets and Morphological Shared Weighted Neural Network Based Automatic Face Recognition

Chandrappa D N, Ravishankar M
2013 Signal & Image Processing An International Journal  
Automatic face recognition system is an important component of intelligent human computer interaction systems for biometric. It is a attractive biometric approach, to distinguish one person from another. To perform Automatic face recognition system, the hybrid approach called Gabor Wavelets face detection and Morphological Shared Weighted Neural Network based Face Recognition. Face detection is performed by using Gabor filter feature extraction. The feature vector based on Gabor filters is used
more » ... to reduce feature subspace. The detected face regions are given as input to Morphological Shared-weight Neural Network (MSNN) which performs face recognition. Being non linear and translation invariant, the MSNN can create better generalization during face recognition. Feature extraction for MSNN is performed on hit-miss transforms that are independent of gray-level shifts. Then the output is learned by interacting with the classification process. The system is experimented on standard datasets and also on our own dataset of image owing to different illumination conditions and cluttered background in non frontal images with a crowded scene with different conditions. Face detection is performed on a cluttered background and crowded scene where a false negative and false positive is detected. The MSNN recognize ignores all false positives and false negatives from face detection and performs human faces in unconstrained environments, and multi-view recognition.
doi:10.5121/sipij.2013.4405 fatcat:a6odx65odfg33ltst37u7oiniq