基于颜色分割和KL变换的人脸特征检测  

Detecting facial features based on color segmentation and KL transform

作　　者：彭振云[1] 陶霖密[1] 徐光祐[1] 张宏江[2] 

机构地区：[1]清华大学计算机科学与技术系,北京100084 [2]微软中国研究院,北京100080

出　　处：《清华大学学报（自然科学版）》2001年第4期218-221,共4页Journal of Tsinghua University(Science and Technology)

基　　金：国家"八六三"高技术项目 !(86 3-30 6 -0 3-0 1-1)

摘　　要：提出一种利用 KL 变换在复杂背景上检测人眼的方法。通过颜色分割将人脸和背景分离后 ,利用 Hough变换和人眼的自然约束在肤色图上找到所有候选瞳孔 ,然后根据每对瞳孔的位置分割出候选眼部 E1 ,将候选眼部投影到特征眼空间得到构造眼部 E2 ,最后根据 E1 和 E2 的相关度确定眼睛位置。对 980幅在各种环境下拍摄的图像进行的测试结果表明 ,该方法对于不同的姿态、年龄、光照等条件下的照片均能达到相当高的识别正确率。An approach is developed to detect eyes on a complex background. First, the skin color is separated from the background using a color grouping algorithm. With the resulting skin color map, all potential irises are located using the Hough transform and geometric constraints for the eyes. Eye regions are then extracted based on the positions of the potential irises. Each of the eye regions ( E 1) is projected onto the eigen space and then reconstructed as E 2 . The final eye positions are determined by measuring the correction between E 1 and E 2 . Experiments have been performed using 980 images taken in different environments. The results show that the presented method is quite robust and efficient for various poses, ages and illumination conditions.

关 键 词：颜色分割 KL变换 特征眼 人脸标准 特征检测 肤色分割 区域估计 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]