Performance Evaluation of RGB Skin Color Segmentation Based Face Detection Technique
Uma Shankar Kurmi, Hari Shanker Srivastava, Dheeraj Agrawal, R. K. Baghel · 2014
Face detection is complex problem of image processing due to the problems of illumination, occlusion, and imaging condition. This paper presents algorithm for face detection based on segmentation process. Segmentation is applied to detect skin region in RGB color space. The illumination problem eliminated by white balance correction of input image and the segmentation performance improved. Computation time is reduced by segmentation process. Logical „and operation‟ is performed on input image and white balance corrected image. Then dilation and erosion morphological operations are performed. Bounding box ratio is used to extract the face region. The bounding box ratio, width over height, is decided experimentally. Face feature extraction is performed by using Laplacian of Gaussian Filter (LoG). Isosceles triangle property is used to find face features. On the basis of face features, face image is detected.