Research On Face Detection And Recognition Based On Facial Feature And Improved Gabor Filter

In recent years, because of the explosive using of the e-commerce, automatic face recognition has become the most potential bio-authentication means. It is the most acceptable as a means of identification as a non-invasive, direct , friendly, and fit the people visual habits, what makes it to become the most potential for development in identification, security checks, automatic detection and many other fields. Therefore, face recognition has become an important research topic in the field of computer vision and pattern recognition. Face detection is the first step in face recognition. Therefore, face detection has become an important research topic which is concerned about by more searchers.This paper mainly aims to launch the discussion in the face detection and recognition, and discuss a variety the basic theory and methods for face detection and recognition and supplement by the corresponding experiments, and focus on analysis of skin color ,skin features and Gabor wavelet. It has good practicability through theoretical analysis and experimental evidence. Major research of this paper is as follows:(1) The paper presents an approach based on skin color segmentation and improved Gabor filter. First, YCbCr color space is used to segment face skin color and background region with color image so as to remove most of background region and to enhance computation speed. Then, the improved Gabor filter is proposed to implement the convolution for the extracted region of face skin color in order to obtain the vector of face feature and to implement the comparison with the vector set of face feature obtained from the training samples in order to verify the face or not.(2) An algorithm was presented based on facial feature and improved Gabor to detect face. In the detection step, firstly, facial feature were used to remove almost background regions, then all the candidates were verified by improved Gabor filter to confirm the face or not.(3) Face recognition based on Feature. The paper presents an approach based on Gabor filter, PCA and LDA for Face recognition. Gabor filter is used to the training samples, and gets the face feature vector groups; then PCA is used to reduce dimension, LDA is used to compare between classes feature to get the similar face and achieves recognition effect.