基于光照自适应调节和模糊分类的人脸图像质量提升算法研究

Face image quality improvement algorithm based on adaptive adjustment of illumination and fuzzy classification

本文介绍人脸识别系统关键技术,综述不同光照下人脸识别技术研究。当前光照预处理方法包括基于小波变换处理、Retinex方法等,对比分析相同光照预处理方法在不同人脸测试库的性能差异。特定预处理方法对特定属性数据库才能达到理想识别效果。提出人脸光照自适应调节算法,建立模糊识别模型进行算法实验,提升非理想条件下人脸识别准确率,以挑选清晰的人脸图像提升移动场景人脸识别准确率。采用设计算法,人脸识别准确率达到97.21%,具有推广价值。

In this paper,the key technologies of face recognition system are introduced,and the research of face recognition technology under different illumination is reviewed;the current illumination preprocessing methods include wavelet-based processing,Retinex method,etc.,and the performance differences of the same illumination preprocessing method in different face test libraries are compared and analyzed;the specific preprocessing method can achieve the ideal recognition effect for the specific attribute database.An adaptive adjustment algorithm for face illumination is proposed,and a fuzzy recognition model is established for algorithm experiments to improve the accuracy of face recognition under non-ideal conditions,so as to select clear face images to improve the accuracy of face recognition in moving scenes.The accuracy of face recognition reaches 97.21% by using the design algorithm,which is worth popularizing.