Face Recognition of Robust Regression With Pre-processing Technique using CLAHE technique

Abstract

One of face recognition method that has been developed to overcome the problem of illumination variation is the Robust Regression. This method uses the Histogram Equalization technique in the pre-processing stage, which is used to reduce the effects of illumination factors on face images. The results of previous research show the face recognition performance of the Robust Regression method (which uses Histogram Equalization technique in the pre-processing stage) is very high. In this research, the Contrast-limited Adaptive Histogram Equalizaton (CLAHE) technique will be used in the pre-processing stage to replace the Histogram Equalization technique. The research was conducted to find out how the effect of pre-processing technique changes on face recognition performance. The empirical experiment was conducted using one of the standard face database data i.e the Extended Yale Face Database B. Based on the experimental results, the average accuracy of face recognition where in the pre-processing stage using the CLAHE technique is 98.06%. This result is better than face recognition performance using the Histogram Equalization technique at the pre-processing stage, where the average accuracy of facial recognition is 96.17%.

Keywords: Face Recognition, Robust Regression, Contrast-limited Adaptive Histogram Equalizaton, and Extended Yale Database B.