摘要： 针对眼底图像的归一化问题，提出一种基于背景估计和同态滤波的眼底图像归一化方法。从图像观测模型的角度出发，利用背景取样估计和插值技术实现图像背景的估计，通过边界奇异值处理和同态滤波增强图像细节，使图像的非均匀照度和对比度得到归一化增强。实验结果表明，与其他方法相比，该方法的归一化性能较好。

关键词: 背景估计, 同态滤波, 非均匀照度, 对比度, 眼底图像, 归一化, 预处理

Abstract: Aiming at the retinal image normalization problem, this paper proposes an observation model based on background estimation and homomorphic filtering. By subsampling and interpolation techniques, the background estimation of the retinal images is achieved. On this basis, the details of the image are enhanced with singular value processing and homomorphic filtering techniques. Thus non-uniform illumination and contrast are normalization enhanced. Experimental results show that this method has better normalization performance than other methods.

Key words: background estimation, homomorphic filtering, non-uniform illumination, contrast, retinal image, normalization, preprocessing

中图分类号: 

• TP911.73