一种新的血管造影图像Hessian矩阵增强算法

A Novel Hessian Matrix Enhancement Algorithm for Angiography Images

针对实际血管造影图像存在的噪声和光照不均,本文提出一种新的Hessian矩阵增强算法来提取血管图像。算法主要通过无抽样方向滤波器组进行血管方向图分解、同态滤波器滤波、Hessian特征矩阵和血管方向图合成。实验表明,与传统Hessian算法相比,提出的算法能够连续提取血管的精细结构,对噪声不敏感,处理的图像质量更高。

A novel Hessian matrix algorithm is presented to enhance angiography images considering their noise and non-uniform illumination. The algorithm is performed in four steps..decimation-free directional filter bank decomposing,homomorphic filtering, Hessian eigenvalues matrix, and directional image reconstruction. Experimental results show that the proposed algorithm can detect filamentous image structures continuously,and has great advantages of both noise insensitioity and image visual quality,in comparison to the traditional Hessian algorithms.