改进人脸识别算法在APP中的应用研究  

Application research of improved face recognition algorithm in APP

作　　者：邢卫强 刘从军[1,2] XING Wei-qiang;LIU Cong-jun(College of Computer,Jiangsu University of Science and Technology,Zhenjiang 212003,China;Jiangsu Keda Huifeng Technology Co.Ltd,Zhenjiang 212003,China)

机构地区：[1]江苏科技大学计算机学院,江苏镇江212003 [2]江苏科大汇峰科技有限公司,江苏镇江212003

出　　处：《电子设计工程》2019年第23期185-188,193,共5页Electronic Design Engineering

摘　　要：针对主成分分析算法在人脸识别应用中由于平等的对待脸部的每一处特征而影响其识别率问题,本文通过一维高斯函数推导出二维高斯函数,由此构造出一种双核边缘柔化加权函数来解析人脸扫描图像,选取适当参数,采用ORL人脸库来训练,结合OpenCV库中的cvResizx函数实现人脸图像的统一缩放,以达到最佳的识别效果,得出相比于主成分分析算法不仅在时间上快了1.8秒外,而且在识别率上提高了7%左右,最后算法应用于APP中效果明显,达到预期。On principal component analysis algorithm is applied in face recognition as a result of the equal treatment of facial feature and affect the recognition rate of each place,in this paper,one dimensional gaussian function is deduced two-dimensional gaussian function,thus constructing a dualcore edge softening weighted function to parse the face scan images,select appropriate parameters and ORL face database is used to training,combined with the OpenCV library cvResizx function to achieve reunification of the face image scaling,in order to achieve the best effect,it is concluded that compared with the principal com ponent analysis algorithm is not only fast in time of 1.8 seconds,but also increases the recognition rate by about 7%.Finally,the algorithm is applied to the APP with obvious effect and reaches the expected result.

关 键 词：人脸识别 APP 高斯函数 双核边缘柔化 

分 类 号：TN99[电子电信—信号与信息处理]