结束语----------------------------------------------------------------------------------------------------------37 致 谢 参考文献

摘要

人脸识别因其在安全验证系统、信用卡验证、医学、档案管理、视频会议、人机

交互、系统公安(罪犯识别等)等方面的巨大应用前景而越来越成为当前模式识别和人工智能领域的一个研究热点。

本文提出了基于24位彩色图像对人脸进行识别的方法，介绍的主要内容是图像处理，它在整个软件中占有极其重要的地位，图像处理的好坏直接影响着定位和识别的准确率。本软件主要用到的图像处理技术是：光线补偿、高斯平滑和二值化。在识别前，先对图像进行补光处理，再通过肤色获得可能的脸部区域，最后根据人脸固有眼睛的对称性来确定是否就是人脸，同时采用高斯平滑来消除图像的噪声，再进行二值化，二值化主要采用局域取阈值方法，接下来就进行定位、提取特征值和识别等操作。经过测试，图像预处理模块对图像的处理达到了较好的效果，提高了定位和识别的正确率。

【关键字】：人脸识别；光线补偿；高斯平滑；对比度增强

Abstract

Face recognition is a complex and difficult problem that is important for surveillance and security, telecommunications, digital libraries , video meeting, and human-computer intelligent interactions.

The paper introduced the method of face recognition that based on the 24 bit multicolor image, Main content that the paper introduced is the picture treatment, It occupies the extremely important position in the whole software, the quality of picture process directly influenced the accuracy rate of localization and discerning. The picture process technology that the software mainly used included : light compensating、gauss smooth and twain value method. before discerning, we compensated the light for image, then we could obtain the possible face area through the complexion, finally, the system could depend on the symmetry of eyes to make sure whether it is the face of people, at the same time, the system could eliminate noises through the method that named gauss smoothness, then we used twain value method to deal with picture，the method got the threshold value in face area . After dealing with picture , the following operations are localization、draw characteristic value 、discerning