基于DWT和形态学的眼底图像小血管检测

Small Blood Vessels Detection in Retinal Images Based on Discrete Wavelet Transform and Morphology

关键词针对眼底图像中末端小血管检测难、细节容易丢失的问题,提出一种基于离散小波变换(DWT)和形态学滤波的检测算法。通过小波变换多尺度分析眼底图像小血管系数、背景系数的不同特征,选取分量信号的系数后重构图像。同时以自适应阈值Canny算法提取小血管边缘;然后将结合小血管宽度选择适当结构元素半径,对重构图像进行灰度膨胀,实现小血管检测。结果表明,形态学结合DWT的检测算法能够准确地检测小血管,与常见边缘检测算法相比检测成功率较高。

In order to detect the end of the small blood vessels in retinal images, proposes a discrete wavelet transform （DWT） and morphological filtering of the detection algorithm. Through the analysis of small blood vessels in retinal images factor, background coefficients of different special permit by multi-scale wavelet transform, after selecting the coefficient of component signal reconstructs image. At the same time uses adaptive threshold Canny edge algorithm to extract the small blood vessels; then selects the appropriate radius of the structure elements combining the width of small blood vessels, makes the gray-scale expansion on the reconstructed image to realize small blood vessels detection. The result shows that the morphology of the detection al- gorithm with DWT can accurately detect small blood vessels, compared with the common edge detection algorithm detects a high success rate.