ABSTRACT: Role of Face recognition in security applications can never be overstated, that is why researchers around the world are still actively participating in this area. The initial step in the procedure of face recognition is to detect it efficiently. In this paper, a hybrid technique is being proposed using skin detection (in RGB color space) and block approach. This fusion is proposed to achieve fast skin detection. Along with the fusion, template matching is also used for detection purpose. Block approach for skin detection means dividing the image into square blocks virtually and then applying the detection ratios on corner pixels. If the corner pixels satisfy the ratios completely, the whole block will be considered as a skin block and if none of them satisfies these ratios, the block will be treated as a non-skin block. Even if some of the corner pixels satisfy the ratios, then the block will be searched for skin pixels. In this way, it will rapidly detect a face in an image including high resolution pixels. The experiments are carried out and the results have shown the significant improvements both in time and accuracy of the detection, that has prove the fastness and robustness of the proposed technique.
Journal of Applied Computer Science & Mathematics. 01/2011;