基于Retinex理论的眼底彩色图像增强算法

The Color Fundus Image Enhancement Algorithm Based on Retinex Theory

眼底彩色图像存在亮度低、对比度差、局部细节丢失等问题,分析已有Retinex图像增强方法存在的问题,在此基础上提出一种改进的基于Retinex理论的眼底彩色图像增强方法。首先提取亮度分量,对亮度通道进行多尺度Retinex增强,改进将图像映射到显示器上的gain/offset算法以及颜色恢复方法,最后对具有亮度信息的红色通道进行恢复。为验证方法的有效性,以DIARETDB0眼底图像数据库为研究对象,并与多尺度Retinex(MSR)、带色彩恢复的多尺度Retinex(MSRCR)、直方图均衡化(HE)、对比度受限自适应直方图均衡化(CLAHE)4种经典增强算法进行比较。结果表明,所处理的图像在色彩保护、血管对比度的提升以及图像细节的增强方面比其他图像增强方法有更好的效果,信息熵提高5%~7%,峰值信噪比(PSNR)比传统方法提高1~2倍,客观评价指标明显优于当前常用的眼底图像增强方法,对进一步眼底图像的识别具有重要的意义。

The color fundus image is usually suffered from poor brightness,low contrast and local detail loss.This paper analyzed drawbacks of Retinex methods,and proposed a new effective enhancement algorithm based on Retinex theory. First of all,luminance components of original image were extracted. Next,a multi-scale retinex algorithm was used on it. The simplest possible color balance algorithm was adopted to modify the gain/offset correction method. At last, we calibrated the red channel information to restore the luminance information. In order to verify the effectiveness of the method,the proposed method was compared with other enhancement algorithms including multi-scale retinex（ MSR）, multi-scale retinex with color restoration（ MSRCR）,histogram equalization（ HE）,contrast limited adaptive histogram equalization（ CLAHE） on the DIARETDB0 fundus image database. Experimental results showed that the proposed method had better effect on the color protection,vascular contrast improvement and enhance image details than the other Retinex algorithms and conventional image enhancement methods. The information entropy was increased by 5% to 7% and the peak signal-to-noise ratio（ PSNR） was 1 - 2 times higher than the conventional methods. The objective image quality index was significantly better than the other fundus image enhancement methods. This method is of significance to further fundus image recognition.