A LOCAL APPROACH TO GABOR WAVELET BASED FACE RECOGNITION SYSTEM
Electronics & Telecommunication Engineering Deptt. SGGSIET, Vishnupuri, Nanded, India, Kailash S. Sharma · International Journal of Advanced Research in Computer Science · 2017
Gabor wavelet has proved to be an effective tool in extracting important features from the face images. In this research work we are proposing a face recognition system which uses Gabor filter bank to create Gabor feature images. The Gabor feature images with their different orientation and scale has increased the feature dimensionality. To reduce these features and form the final feature descriptor a local approach based on the regional histogram formation is used. In this approach, the featured images are divided into different regions. The histogram of each region is calculated and concatenated to form the final feature descriptor. Chi square similarity measure is used for classification. The effectiveness of the algorithm is justified on the face images which have illumination variations and pose variations in it.