摘要： 提出一种新的血管图像去噪和锐化方法。在对血管多尺度分析的基础上，利用血管响应函数替代梯度，构建血管的一致性扩散系数。该扩散系数包含前向和后向扩散操作，在去除图像噪声的同时，可以锐化血管。在合成图像、肺小血管图像上进行仿真实验，结果表明，该方法适用于模糊的医学血管噪声图像，不仅能较好地去除噪声，而且能锐化血管边缘。

关键词: 前后向扩散, 多尺度分析, Hessian矩阵, 二阶方向导数, 图像锐化, 去噪

Abstract: A new method of vessel image enhancement is presented. Based on the multi-scale image analysis of vessel image, the vascular response function instead of gradient is applied to construct a vessel image diffusion coefficient. The constructed diffusion coefficient contains Forward and Backward(FAB) diffusion operation. While de-noising the image, it also sharpens the edge of the vessel image. Experiments on synthesis image and pulmonary image describe the performance of the method. This new method has a better result in image restoration than conventional diffusion methods.

Key words: Forward and Backward(FAB) diffusion, multi-scale analysis, Hessian matrix, second order directional derivative, image sharpening, de-noising

中图分类号: 

• N945