Fast Template Matching Method based Optimized Sum of Absolute Difference Algorithm for Face Localization

Recently, Template matching approach has been widely used for face localization problem. Normalized Cross-correlation (NCC) is a measurement method normally utilized to compute the similarity matching between the stored faces templates and the rectangular blocks of the input image to locate the face position. However, there is always an error on locating the face due to some non-face blocks seem more to be the face position than correct blocks because of variation either in illumination or image with clutter background. In this paper we proposed a fast template matching technique based Optimized Sum of Absolute Difference (OSAD) instead of using NCC to reduce the effects of such variation problems. During the experiments a number of similarity measurements tested to prove the high performance of OSAD compared with other measurements. Two sets of faces namely Yale Dataset and MIT-CBCL Dataset were used to evaluate our technique with success localization accuracy up to 100%.