复杂背景下多姿态人脸快速检测算法
Fast Multi-Pose Face Detection in A Complex Background
-
摘要: 通过对图像进行预处理(似眼物检测),不但缩小了人脸检测的搜索空间,而且将复杂背景下多姿态多人脸检测简化为在候选区域内,已知大致的人脸尺度情况下正面端正单人脸检测问题,并结合多模板匹配与规则验证完成了人脸区域的最后确定.实验结果表明该算法具有较高的速度和检测成功率.Abstract: By using the eye-analog detection, the algorithm can not only reduce the search space of face detection, but also simplify the multi-pose multi-face detection in a complicated background into single front-view vertical face detection in the candidate regions. With knowing the approximate size of the face, it uses multi-template matching and rule verification to determine the location of real face. Experiments demonstrate that it is relatively fast and rather successful.
下载: