实时视频图像快速唇部检测方法的研究与实现

Research and Implementation of a Real-time Approach to Lip Detection in Video Sequences

在视频图像中快速提取完整的嘴唇外形是计算机唇读系统的首要任务之一 ,文中提出了一种综合采用RedExclusion和Fisher变换的唇部检测方法 ,根据肤色模型和运动相关性在视频图像中检测人脸 ,然后在RGB空间内排除红色 ,用 (G ,B)分量作为Fisher变换矢量 ,对人脸下三分之一区域进行唇部图像增强 ,并利用增强后的灰度图像的灰度值呈正态分布这一特性 ,自适应确定肤色和唇色阈值 ,将唇部从背景图像中分割出来。该方法能检测出完整的嘴唇外形 ,且检测速度高 ,对光照、胡须及说话人不敏感。

Locating the lip in video sequences is one of the primary steps of the automatic lipreading system. In this paper a new approach to lip detection based on Red Exclusion and Fisher transform is presented. In this approach,firstly,face region is located with skin-color model and motion correlation. Then the face image is trisected,taking into account the lowest part where the lip lies for the next processing. Secondly,the R-component is excluded in RGB color space and G-component and B-component are used as the Fisher transform vector to enhance the lip image. Finally,in the enhanced image,the threshold is adaptively set to separate the lip color and the skin color in the light of the normal distribution of the gray value histogram. The experiment results showed that the fast approach is very efficient in detecting the whole lip and not affected by illuminant and different speakers.