Evaluation of the Color Space Influence in Face Detection

A.F. Arias, N.B. Rodríguez, L.L. Freire, and M.G. Penedo (Spain)


Face detection, color space, edge detector


Face detection is a very active research field because it usually serves as a preprocessing step for computer vision tasks such as face recognition, image compression, camera auto focusing, etc. Recently, different methods have been proposed using color information for skin detection in images in order to detect faces. These systems rely on detecting regions which are likely to contain human skin pixels and the combination with edge information to filter the results. To that end, the pixels are classified as skin/non-skin in a color space. In this work, we perform an analysis of different color spaces to evaluate their suitability for skin detection. Also, different classification approaches have been analysed for this task. The dataset used in the experiments consisted of images from a public database and images containing general purpouse scenes to simulate a more realistic enviroment. The obtained results show an improvement over the previous proposals.

Important Links:

Go Back