基于豪斯多夫距离的快速多人脸检测算法

Fast Multi-Face Detection Algorithm Based on Hausdorff Distance

  • 摘要: 以豪斯多夫距离为相似性度量,提出了一种对光照及背景变化具有鲁棒性的快速准确多人脸检测算法。由于豪斯多夫距离多模板匹配算法的复杂性,采用降低模型维数,寻找感兴趣区域和用变换间距离来寻找最适宜模型等多种方法提高计算豪斯多夫距离的效率。实验结果证明该算法能快速准确检测人脸。

     

    Abstract: This paper provides a fast and accurate algorithm of multi-face detection that is robust to the change of background and light. Because the multi-template matching algorithm based on Hausdorff Distance is very complex, many resolution techniques, such as decrease the dimension of model, find the area of interest, and get the fittest model based on the distance between transformation, have been used to advance the computation efficiency. The algorithm's performance was proved by experiment results.

     

/

返回文章
返回