基于HOG-SIFT融合优化的多人脸特征提取方法

A Multi-Face Feature Extraction Method Based on HOG-SIFT Feature Fusion Optimization

针对昏暗光线及复杂背景下人脸特征提取效果易受环境因素干扰影响的问题,在进行人脸图像预处理过程中引入双边滤波处理,进一步研究基于自商图像理论的人脸图像光照干扰抑制方法;结合HOG特征较好的全局性以及SIFT特征对复杂背景影响较好的适应性,提出一种基于HOG-SIFT融合优化的多人脸特征提取方法.试验结果表明,该方法可有效实现昏暗光线环境及复杂背景下多人脸特征提取功能.

Aiming at the problem that the effect of face feature extraction in dim light and complex backgrounds is easily affected by the interence of environmental factors,bilateral filter is firstly exploited in the process of face image preprocessing,and the illumination interference suppression method of face image based on self-quotient image theory is further studied.Secondly,a multi-face feature extraction method based on HOG-SIFT fusion optimization is proposed with the combination of the better globality of HOG features and the better adaptability of SIFT feature under complex background.Finally,the experimental results show that the proposed method can effectively extract multi-face features in dark light environment and complex background.