Face Recognition Method Research Base On Phase Congruency And Modular PCA

Face recognition technology is a very active research topic in computer pattern recognition. It has a broadly application in law, in business, and in security system and so on. Because of the particularity of face images, face recognition problem is a hard problem in pattern recognition, it is also a challengeable problem and covers knowledge of many subjects such as signal processing, intelligence control, pattern recognition, machine vision. How to recognize the face images by using computer quickly and exactly is the difficulty and hotspot. So there are many works to do to make it mature completely. As the developing of society and progress of technology, the application of face recognition technology will become more and more comprehensive.This paper analyses and studies the existing face recognition approaches deeply. By analyzing the illumination factor which affects the face recognition ratio, summarize a new method based on phase congruency. After Fourier transform, the abundant texture and structure information is gained in this method. By applying phase congruency to face image edge detection, better results are gained and the essence character of image opened out by phase information is revealed. Experiments prove that phase reconstructed image can wipe off the influence of lighting condition. We improve the recognition veracity by selecting face image feature which is foreign to illumination. At the same time, we can prove that the face lighting irrespective feature can be selected by combining phase spectrum analysis and phase congruency analysis.At the same time, considering the influence brought by face expression, we study the PCA method deeply and propose a modular PCA method based on phase congruency image for the first time. This novel method is not only the prolongation of PCA, but also the amelioration of the conventional modular PCA method. Firstly, eliminate the illumination influence by using phase congruency image. Secondly, modularization gives attention to local feature of face image and reduces the influence of face expression. It also provides us a new thought for face recognition research. The results of our experiments prove the validity of this method.