Research On Automatic Face Recognition

The traditional methods for identity authentication such as shibboleth, password, identification-card, magcard and IC card, which result in counterfeiting, embezzling or decoding due to individual separation, could not satisfy the requirement of the economy and security of modern society. Because the traits of stabilization and individual diversity, human face has been an ideal object in human's biometric feature identification. Machine recognition of Human face is currently one of the most active researches and the most challenging problems. At first, the face images are obtained by different way or in different condition, so they have substantial difference in quality, geometry, illumination, etc. But the most essential reason is that face is a kind of non-rigid object that has highly similarity. Different person's faces have similar shape and structure, and one person's face has his different state in different conditions. In the past decade, many research groups make great efforts on it and a series of successes have made personal identification appear not only technically feasible but also economically practical. However, no perfect solution can accomplish this task under the non-constraint condition.This paper reviews the history of face machine recognition, and analysis the challenge of face recognition. It provides a survey of the primary principle, the essential methods and the key technology of face detection and recognition recently. The methods of face recognition involve four major fields: based on appearance feature, based on template, based on algebra feature, and based on machine learning. In this paper, the problems are discussed primarily in complex background and varying illumination, which is in the fields such as face detection, organ location, feature extraction, illumination compensation etc.The followings are the main research contents in my paper:(1)A method is putted forward for real-time human face detection in color images that is composed of skin color segmentation and template matching. An algorithm of color adaptation and the concept of confidence measure of color model are presented also.A solution for the detection and tracking of human face in color images is described in this paper. The color face image in RGB space is first transformed into human face color model under r,g,b chrominance, which can be used to differentiate skin pixels from non-skin pixels. The color image is then divided into grid units which are arranged as a rectangle. The ratio of skin pixels to all pixels in the grid unit is computed and the grid unit is regarded as a skin unit the ratio of which is greater than a threshold. The adjoining skin units are concatenated. The resulting area is labeled as a candidate face if its shape is quasi-ellipse or quasi-rectangle complying to proper ratios. Finally the gray image of face area is matched to the face template to discern the real face. This algorithm overcomes the influence of complex background upon face detection and obtains higher accuracy of detection. In particular, the time consumed by this algorithm is much less than the traditional template-match method. The method is suitable for real-time face detection system.But the face skin color is sensitive to the change of the environment illumination. An algorithm is proposed to update the skin color model's parameters in time so that the model is adapted to different lighting conditions. The confidence measure is presented to evaluate the reliability of skin color model. Experiments demonstrate that the self-adaptive color model is more effective than the fixed model. The color adaptation makes that the color model can be better fit to the more complex application environment.(2)In the face detection in gray image, the method of the cascade of classifiers based on AdaBoost algorithm is adopted. Face detection is a classify problem of two class. The resulting classifier design problem is very challenging. The classifier in this paper is a cascade architecture that incl...