An Improved method for face authentication and verification system

In this paper, the proposed method uses various stages, namely the pre processing, the image segmentation, the feature extraction, the post processing and the matching stage. Images always contain an adequate amount of noise caused by operator performance, equipment, and the environment, which will lead to serious inaccuracies. In the pre processing stage, first the face region mask is applied, then the histogram equalization, FFT (Fast Fourier Transform), Adaptive thresholding and Binarization were done to obtain an enhanced image after removing the noise. In the second stage, image segmentation is done to estimate the block direction using the linear square approximation and to extract the Region of Interest (ROI) by the Morphological thinning operations. In the next stage, feature extraction is done to identify the global points and to indentify the localization of the global points. In the final stage, the post processing is done to remove the false edges of the face contour and matching between the template and the input face image is recognized and verified. A recognition rate of 100% is obtained between the template and the input face image. In particular, the feature matching algorithm is used to provide a satisfactory result for face recognition and verification. The computation complexity is highly reduced and it uses simple steps.