Face Recognition Technology Research Based On Skin Color

Face recognition is based on face feature to the person’s identity identification, with its can be passive, friendly characteristics, in the electronic commerce, access control systems, human-computer interaction, such as id card application field widely used.Since the1990s, face recognition research literature has been rising attitude. Face detection and face recognition is one of the important research direction.The process of image acquisition influenced by many reason,such as posture, facial expression and the influence of the factors such as illumination.In this paper, the face recognition system algorithm is studied.The main research work includes below:1. Proposed based on skin color and facial features of the face detection algorithm. First, paper studied some related theories like the face image preprocessing, color space and color model. Then, based on the color a face detection system have been designed.The use of facial features different color in the face of the candidate of the scope of the rectangular frame for further screening. The system can realize fast face detection, and good adaptability.2. Puts forward the use of color similarity to the regional segmentation method, make full use of the lip color information. Due to the cause of the light, image face will be light and shade difference. The color similarity measurement method is adopted in this paper, namely a pixel is the only category to a color categories, there is not deceitful sex, segmentation labial ministry effect is good, and avoid the situation which is divided into different color.3. Using Gabor wavelet to split out of the mouth to carry on the analysis. In recent years, the Gabor feature is recognized as the most effective face representation method. Based on the characteristics of traditional Gabor face recognition is will face image and multi-scale, multidirectional Gabor wavelet function convolution after Gabor amplitude characteristics of the sampling and feature selection, and then cascade to form the feature space to show a face. However the sample get feature dimension is very high, scoop out useful identification information is very time-consuming. Based on the Gabor wavelet to local characteristics of feature extraction, the dimension is small, so the recognition velocity quicken. Have a good robustness. To solve the problem of the decorations and the age.