一种基于倒谱的人脸图像清晰度评价方法  

A cepstrum-based clarity assessment method for face images

作　　者：杨飞[1] 苏剑波[1] 

机构地区：[1]上海交通大学自动化系,上海200240

出　　处：《光电子．激光》2009年第10期1357-1360,共4页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金资助项目(60675041);教育部新世纪优秀人才支持计划资助项目(NCET-06-0398)

摘　　要：针对人脸图像清晰度,给出了一种基于倒谱的定量评价方法。实验结果表明,与传统方法相比,所提出的清晰度评价方法能更加准确地评价出人脸图像的质量;采用该方法对人脸识别样本进行筛选,能有效地提高自动人脸识别的识别率,且对整体识别时间影响很小,因而更适合用作自动人脸识别中的图像质量评价。A novel face image quality assessment method based on cepstrum is proposed,which can quantitatively evaluate the image clarity.Experimental results show that the proposed clarity evaluation method is more accurate than traditional ones;moreover,it is effective when it is employed to select proper face image samples so as to improve the accuracy of automatic face recognition,and it casts little influence on the total recognition time;so it is more suitable for the image quality assessment for automatic face ...

关 键 词：图像质量评价 清晰度 倒谱 人脸识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]