Research On Face Detection And Feature Point Positioning Technology

Biometric identification technology development prospect is broad. Especially with the development of mobile communications technology and network technology, many areas are in urgent need of more secure, efficient information detection and identification technologies, such as the areas of information security, network electronic transactions and criminal investigation confidential. Because non-contact authentication and identification of individuals with the advantages of universality, uniqueness, stability, reliability, etc., more and more scholars began to pay attention and in-depth research. Face recognition technology is one of the most active areas of research in biometrics, and face detection and feature point location is a critical step in face recognition system. The accuracy of face detection and feature point positioning accuracy directly affects the subsequent face recognition accuracy.This paper mainly studies based on the color of skin of face detection and facial feature points localization of eyes and mouth, and improved algorithm were put forward in the face detection and feature point positioning. Experimental analysis and comparison shows that, based on face detection color added Gabor texture features analysis can effectively reduce the false detection rate and improve the accuracy of face detection. Improved adaptive synthetic correlation filter proposed in the facial feature points positioning of the eye, can not only improve the positioning accuracy and error stability, but also well suited to small-angle deflection face, glasses, facial expressions and other features. Two-dimensional Gabor wavelet transform is used to extract facial feature point mouth position, combined with the already positioned around the position of the eyes and facial features prior knowledge of proportion, and ultimately get the exact position of the mouth. Main work is summarized as follows:1. In order to reduce skin color close but non-textured skin area false detected in the background information, we first put the image through the Gabor filter and that image is compensated by "reference white", contour and texture features are highlighted, the image preserves the original area of skin and smooth texture area, and close to the color of the original image but coarse-grained texture becomes the binary image, which can reduce the false detection of non-human skin is likely to face the region, to improve skin detection accuracy. Then a simple color model is created in YCbCr color space, and can get the color image segmentation mathematical morphology to obtain the final face region2. For the problems of lower eye location accuracy and unstable error distribution based on traditional eye location methods, the adaptive correlation filter is combined with integral projection to detect eye position precisely, and two improvements are made in the training and test phases. Firstly, adaptive synthetic correlation filter (ASCF) is rotated within the angle range of -0.2,-0.1,0,0.1,0.2, and the corresponding location of the maximum grey value will serve as an initial anchor point, and then gray-level integral projection is performed in the horizontal and vertical direction respectively within the anchor point 5×5 neighborhood. Finally, the integral minimum value is used as the ultimate goal orientation. Compared with the average synthetic exact filter, minimum output sum of squared error filter and adaptive synthetic correlation filter, the position accuracy of improved algorithm is increased by 2.9% and the mean absolute error and standard devia-tion are lower than those of original algorithms. Experimental results show the proposed improvements are superior to the general algorithms in terms of accuracy and stability.3. In order to enhance facial features mouth positioning speed, we make the eye position getting by improved adaptive filter algorithm as a reference. According to the two-dimensional Gabor wavelet transform facial features good characterization capability, combined with the face of a priori knowledge of facial proportions, final get the positioni-ng location of the mouth.Firstly, the two-dimensional Gabor wavelet is transformed for image, the abscissa of left and right eyes are left and right eyes borders respectively, upper and lower boundaries are determined by the regional distribution of the mouth in the face of a priori knowledge, then get the mouth area of a rectangular area. Setting an appropriate threshold to convert an image into a binary image, using mathematical morphology to get the smallest rectangle that contains the mouth area, and then take the center position of the rectangular area as the location of the mouth of the positioning.