提出一种基于地形信息熵的灰度人脸检测算法,能够部分地消除当前困扰人脸检测过程中的光照和噪声等因素的影响,尤其对阴阳脸的检测有一定的效果。该算法将人脸图像看作一种特殊的地形,并提取有效的信息熵;利用区域融合和区域过滤等操作,获取“人眼对”的候选区域;根据“三庭五眼”人脸特征,最终确定人脸的位置。实验表明,该方法在某种程度上可以有效地消除光照和噪声的影响。
An effective algorithm based on terrain information entropy is proposed for improving gray face detection in complex backgrounds. The algorithm could reduce the influence of illumination and noise to some extent. Firstly, a face image is regarded as a special terrain, and the entropy information is extracted;secondly, by means of region merging and filtering, all possible candidate regions of eye-pairs could be obtained ; finally, the correct face areas could be located according to the geometrical feature of faces. Experimental results show that this algorithm can effectively eliminate the influence of illumination and noise.