CAJ | 학술논문

提出一种基于Hessian矩阵的多尺度视网膜血管集成增强滤波算法。将Hessian矩阵的特征值和特征向量应用于血管特征的响应函数及形态学和非线性各向异性扩散处理的各个环节，选择合理的尺度空间范围及尺度空间增量和调节因子进而达到平滑非线状区域和锐化增强血管区域的效果。与其他血管增强方法相比，该方法在同等准确率下具有较高的稳定鲁棒性。
제출일충기우Hessian구진적다척도시망막혈관집성증강려파산법。장Hessian구진적특정치화특정향량응용우혈관특정적향응함수급형태학화비선성각향이성확산처리적각개배절，선택합리적척도공간범위급척도공간증량화조절인자진이체도평활비선상구역화예화증강혈관구역적효과。여기타혈관증강방법상비，해방법재동등준학솔하구유교고적은정로봉성。
We propose a Hessian matrix-based multi-scale integrated enhancement filtering method for retinal vessels.It applies theeigenvalue and eigenvector of Hessian matrix to response function of vessel feature and morphology as well as all to aspects of nonlinearanisotropic diffusion processing,selects reasonable scale space range and scale space increment and adjustment factor so as to achieve theeffects of smoothing the non-linear region and sharpening and enhancing the vessel region.Compared with other vessel enhancement methods,this one has higher stable robustness with same accuracy.