Real-Time Face Recognition Research Based On JIT

Face recognition as a way of biological identification technology compared with other biological recognition (such as fingerprint identification), have some advantages such as friendly, convenient, independent and so on. With the development of computer and video technology, face recognition has made great progress. However, because of various factors such as complexity background of face image, great information of face and so on, get a face recognition system that can be adapted to a variety of complex situations has been a difficulty for research. Efficient, real-time and accurate face recognition has become the direction of the continuous efforts of researchers. Under the condition of high accuracy rate, this paper proposes a real-time face recognition method based on JIT, aimed at real-time, quickly identify the identity of the test face.JIT mode is a real-time mode of production. Its main idea is:only when needed, according to the amount needed, the production of desired products. JIT mode is in pursuit a production systems of non-stock or inventory to a minimum. This article inherit their ideas, design real-time face recognition system based on JIT mode, is committed to achieve the feature is not retained. That is when the obtained face image feature to be measured, regardless of the number of features contained in the information, the system instantly classify features. According to the characteristics information of acquisition to determine the search range of test face image, reduce the amount of calculation, in order to achieve real-time face recognition.In this paper, according to the thought of JIT mode, first extract face contour information to the similarity matching. Because of the less extracted contour information and computation, we can easily and quickly ruled out a lot of face image of face database that is different from the contour of measured face image for the next Face narrow your search. Then we extract texture features of face with Gabor wavelet transform. In order to reduce the image dimension, principal component analysis (PCA) was used to reduce the dimension. Finally, calculate the distance between measured facial texture and the face texture of remaining face database. This paper adopts the nearest neighbor classifier based on Euclidean distance to achieve human face classification.Experiments can be drawn using JIT mode design face recognition system to a certain degree can improve the efficiency of face recognition. This method can save a lot of unnecessary operations, achieve real-time, efficient face recognition purposes. At the same time due to the use of two different levels of feature extraction method for face feature extraction, feature extraction can better expression on people’s faces. Because the information is sufficient, so it reached a high recognition accuracy rate.