Research On Face Recognition Algorithm

Face detection and recognition is a compositive technology, which blends many subjects into a unity, including digital image processing, computer vision, pattern recognition and so on. Since it can provide a convenient, kindly and reliable method to recognize individuals, the compositive technology has broad prospects and become a hot topic. This paper mainly contains face detection algorithm based on skin detection and face recognition algorithm based on Linear Discrimination Analysis.In face detection algorithm, the thesis takes full advantage of the movement and color information of color video sequences. Use the symmetric difference among three frames instead of the original one between two frames as a result of the continuity of the kind of movement. After the morphological operation, we may locate motion region. Based on face skin clustering, the thesis chooses YCrCb color space and a nolinear transformation to get skin area. Then get the common parts between motion region and skin region. To make sure the human face region, we locate human eyes at last. In face recognition algorithm, the article uses Linear Discrimination Analysis to recognize human faces. In order to make within-class scatter matrix nonsigular, we choose traditional Principal Component Analysis to decrease sample dimensions. Then use Linear Discrimination Analysis to recognize faces. The experiments prove that this detection and recognition algrithm has a simple arithmetic and a high recognition rate in some constraint conditions.