CAJ | 학술논문

基于偏微分方程的增强算法是解决图像边缘结构增强问题最有效的方法之一。本文提出一种将非线性结构张量与非线性边缘增强相结合的算法。其中，非线性结构张量的邻域选择方式能更好地描述图像局部几何结构，非线性边缘增强的各向异性扩散方式更有利于在增强边缘的同时有效地降低噪声水平。实验结果表明，本文算法不仅有效地增强了图像边缘，而且在噪声水平较高的环境下也有着良好的表现。
기우편미분방정적증강산법시해결도상변연결구증강문제최유효적방법지일。본문제출일충장비선성결구장량여비선성변연증강상결합적산법。기중，비선성결구장량적린역선택방식능경호지묘술도상국부궤하결구，비선성변연증강적각향이성확산방식경유리우재증강변연적동시유효지강저조성수평。실험결과표명，본문산법불부유효지증강료도상변연，이차재조성수평교고적배경하야유착량호적표현。
PDE-based method is considered as one of most efficient solutions to edge enhancement. In this article, an algorithm combining the characteristic of the nonlinear structure tensor as well as the behavior of nonlinear diffusion is proposed. The way of the local neighborhood selection during the computation of the nonlinear structure tensor qualifies itself as a better descriptor of the local structure, and the anisotropic behavior of nonlinear diffusion filter offers more advantages in edge enhancement while eliminating the noise. The experiments show that our algorithm is more than capable of enhancing the edge, and it also performs well even applied to the images with heavy noise.