摘要
在人脸识别中,传统PCA算法的识别率会受到光照,表情和姿态等因素的影响,因此采用前人改进的PCA算法。然而,改进的PCA算法仅考虑了图像数据中的二阶统计信息,忽略了多个像素间的线性相关性,对识别率有一定影响。而支持向量机算法在解决小样本、非线性和高维模式识别的问题中表现出很好的优势。因此为了提高人脸识别的识别率,设计了一种新的方法,将改进的PCA算法与SVM相结合进行人脸识别。在ORL人脸数据库中进行了实验验证,结果显示该方法能够有效提高识别率。
The recognition rate of the traditional PCA algorithm was affected by the illumination, expression, gesture and other effects in face recognition. The improved PCA algorithm was adopted in the proposed method. The improved PCA algorithm merely considered the second-order statistical information, ignored the linear correlations among pixels, and had influence on the recognition rate. The SVM algorithm had advantages in solving the small sample, nonlinear and high dimensional pattern recognition problem. In order to improve the recognition rate, the paper proposed a new method which combined the PCA algorithm and the SVM algorithm. The experiments were conducted on the ORL face database and the results showed that the method effectively improved the recognition rate.
出处
《电子技术（上海）》
2015年第8期57-59,50,共4页
Electronic Technology
基金
曲阜师范大学科研启动基金项目[No.20130154]
曲阜师范大学科技计划项目[No.xkj201407]
