用闪光灯进行人物摄影时由于眼底血管的反光会产生红眼现象,小型照相机的闪光灯和镜头距离很近,这一问题更为突出。为了自动消除用闪光灯进行人物摄影时产生的红眼现象,在对基于颜色空间的常用红眼检测技术进行研究的基础上,对现有较为复杂的检测方法进行了修正,提出了一种基于人脸检测的红眼自动消除方法,并在分析的基础上得出一组先验参数用于识别和处理,从而大大降低了人眼定位的复杂度。实验表明,该方法能快速有效地消除普通数字照片中的红眼现象。

Redeye is an annoying defect in photographs caused by red light reflected from blood vessels on the retina when a flashgun is used.As the flashgun is close to the lens in a small sized digital camera,the problem becomes even worse.This paper studies detection and removal of red-eyes based on color space analysis,and proposes an automatic red-eye removal technique.Taking into account the prior knowledge of human face,and through experiments on a number of images,an existing approach to face detection is modified.Parameters in an empirical criterion for face detection are determined,resulting in satisfactory segmentation of the skin area.This is used as a basis for further location of eyes and correction of red eyes.Experimental results confirm the effectiveness of the proposed method.