Research And Development Of Face Recognition System

Face recognition technology is a kind of biometric identification technology,which is based on facial features. The identification is achieved by matching the characteristics of faces which are acquired by camera device.Firstly, the paper introduces the background history, advantages, difficulties of face recognition technology and its application fields, sums up some classical algorithms and important breakthroughs of the current domestic and foreign research institutions and companies in the field of face recognition.Secondly, this paper presents the structure of face recognition system, and deeply studies the algorithms involved in each part, then analyzes the effect of facial feature extraction algorithms on the face recognition. The system consists of of five parts, image preprocessing, face detection, face alignment, feature extraction and feature matching. Image preprocessing is enhancing image contrast by gray transformation and histogram equalization; face detection is using Viola-Jones face detection algorithm, and the face classification is achieved by Adaboost strong classifiers which are trained on the training set of Harr-like features, then cascade the classifiers to improve the speed of face detection; Face alignment is used for face standardization by the geometric transformation based on the coordinates of eyes;feature extraction studies Local Binary Pattern(LBP)and Local Phase Quantization(LPQ). LBP extracts local pixel difference information, and the calculation of LBP is simple and efficient. LPQ extracts local phase information, and it has strong robustness for blurred image; feature matching uses chi square distance to measure the sample space which is extracted by LBP or LPQ, and the paper also studies the principle of acceptance threshold of feature matching. With the the simulation of LPQ and LBP algorithm, the paper shows the difference of these two algorithms andthe recognition rate of different conditions on different face database platforms.Finally, the realization of the face recognition system is based on Qt visual development platform, calling the related function modules of computer vision library Open CV, especially the face detection module, which is convenient and quick to realize the system development work. Meanwhile, the paper also gives the related environment parameters setting and compiling methods of Open CV and Qt, system’s user interface, function module and the final result.