文章摘要
一种用于彩色图像人脸检测的新算法
A New Architecture for Face Detection in Color Images
  
DOI:10.3969/j.issn.1671-5322.2010.04.011
中文关键词: 彩色图像  人脸特征  肤色区域  图像处理
英文关键词: color images  facial features  skin locus  image processing
基金项目:
作者单位
许宏 淮阴工学院计算机工程学院江苏淮安223003 
摘要点击次数: 4687
全文下载次数: 4160
中文摘要:
      为了能够实现单幅彩色图像中的人脸检测,提出了一种用于彩色图像人脸检测的新算法.它首先根据肤色的鲁帮性进行建模,接着在图像中运用最小椭圆平方匹配算法定位出脸部候选区域,经过方向校正后,结合脸部的主要特征,对这些候选人脸区域进一步的判断,最终找出具有人脸特征的图像区域.实验结果显示该算法可以适用于不同光照条件下的人脸检测,即使在脸部特征没有被很好的检测出的情形下也可以较好的完成人脸检测.
英文摘要:
      In order to accomplish face detection from a single color image,this paper introduces a new architecture for face detection in color images.A robust moulding is firstly done in accordance with the skin color,and then the candidate facial locus is located within the image by using the minimum oval square matching archtecture.After orientation normalization,the candidate facial locus will be further judged in combination with the major facial features.The image locus with facial features will finally be detected.The experimental result shows that our system can realize the detection of face under the various illuminating circumstances,even when the facial features are not well checked out.
查看全文   查看/发表评论  下载PDF阅读器
关闭