红眼照片自动检测和修复方法

Automatic Red-eye Detection and Correction Method

针对家庭数字照片出现的红眼问题给出了解决方案?过程如下:(1)找出红眼的H SI色彩模型;(2)自动检测出所有可能红眼区域;(3)自动过滤无效区域,找出真正红眼区域 ;(4)利用红眼周围的颜色特征和红眼的颜色特征进行减色修复;(5)在修复的结果上进行后 续处理,使之看起来更真实?本方法已经用Visual C++在Windows 2000环境下实现,程序运 行速度快,对家庭照片的红眼修改效果好?

Red-eye is the most common problem in image evaluation and correctio n. This paper presents a new efficient algorithm to detect and correct the red-e yes. The algorithm has been divided into five parts: (1) Get the improved HSI mo del using mixed color feature of the red-eyes; (2) Detect all the suspicious red regions automatically; (3) Use a filter to get the real red-eye regions automat ically; (4) Correct the red-eyes by reducing the color using the color features of the red-eye and the circumferences; (5) Gaussian smooth. To test the efficien cy, this algorithm has been realized using Visual C++ under Windows 2000 and the results show that it has fast processing speed and a good repair result.