摘要
提出了一种基于独立成分分析和最小最大概率机的人脸识别系统。该系统首先从摄像头中捕捉包含人脸的实时图像,利用haar特征人脸检测算法定位人脸区域,并将其从原始图像中分割出来。为了更好地提取有效特征,采用了ICA的特征提取方式,结合改进误差估计的最小最大概率机的分类方法对输入的测试图像进行识别。实验证明,该系统能够快速有效地处理实时状态下的人脸识别任务,准确率达到了96.8%,并且对多姿态的人脸具有一定的鲁棒性。
A face recognition system based on independent component analysis and minimax probability machine is proposed. First,capture the image through the camera, locate and segment the face region from the original image based on the haar feature algorithm. Then,normalization and lighting compensation is worked on the face image. In order to get more efficiency features,the ICA method is used. At last, use the improved minimax probability machine to categorize the face image. The experiment shows that,the system is competent for face recognition, the accuracy reach to about 96.8%,and is insensitive to multi - view face,
出处
《现代电子技术》
2007年第1期134-137,共4页
Modern Electronics Technique
