The Main Technology Research About Human Face Detection And Recognition

Face detection and recognition technology (FDRT) is one of the most active research topics in image processing, pattern recognition and artificial intelligent fields. It has important value in theory as well as in practice. Comparing with other biologic characteristics such as fingerprint and iris, face recognition is more easily accepted by users because of many advantages of directness, friendliness and convenience. Face detection and recognition technology as an important researching content in Human-Machine alternation, has great theoretical meaning and application value.This paper deeply does research for face detection method based on skin color information and face recognition algorithm based on Embedded Hidden Markov Model, and introduces the construction of developed face detection and recognition system, the performance analysis and concrete application. The main contributions of this work are as follow:Detect frontal face and locate the main features. A method, which combined face detection and features special position is presented. The distribution and properties of skin color are firstly studied in the color space, and then select a great amount of skin color pixels manually to establish skin model and find the scope of skin color. It only needs to locate the facial features based on the acquired skin color, and to judge whether this face area contains face according to the relative position. The proposed detection algorithm adopts the mechanism of Euler Number in order to reduce the number of candidate facial region. By this method, a lot of skin color region yet not human face region can be excluded.A new algorithm for face recognition based on DCT and hidden Markon model (HMM) is proposed. The original image is decomposed into low frequency and high frequency sub-band images by applying DCT transform. According to the DCT theory, the low frequency image contains main energy content within the original image. The coefficient of DCT,as observation vectors of HMM, are derived by using kernel principal component analysis. A set of images representing different instances of the same person are used to train each HMM. Experimental results show that the proposed algorithm has a good perspective.Experimental results have demonstrated successful face detection and recognition over a lot of facial variations in color, position, scale, orientation, and expression in images. We applied proposed method to many real color images to test the efficiency and validity of method. The system can successfully detect and recognize face in complex environment and not be affected by the change of background and position.