基于SVM的人脸精确验证

Exactness Face Validation Based on SVM

人脸检测系统包括了人脸检测粗筛选和人脸精确验证两个阶段,人脸精确验证是在人脸检测粗筛选的基础上所做的工作,目的是通过精确验证去除粗筛选阶段所得到的候选区域中的非人脸部分。精确验证阶段采用了线性鼻子SVM分类器和非线性特征脸SVM分类器相结合的策略。实验结果表明,这种精确验证的方法基本上可以精确地检验出粗筛选阶段所得图片中的人脸区域。

The face detection system includes two stages： the rough face detection and the exactness face validation, the exactness face validation is based on the rough face detection, the purpose is to wipe off un-human face regions from the candidate regions which obtains through the rough face detection stage. The exactness face validation stage has used the combination strategy of the linear nose SVM classification and unlined SVM classification, the experimental results indicate that this exactness face validation method can basically examine human face regions from the images that got from the rough face detection stage.