基于偏微分方程与一致锐度的图像边界检测  

A hybrid method of partial differential equations and region homogeneity measure for image edge detection

作　　者：房少梅[1] 钟澎洪[2] 

机构地区：[1]华南农业大学理学院,广东广州510642 [2]云南民族大学数学系,云南昆明650091

出　　处：《佛山科学技术学院学报（自然科学版）》2008年第1期1-6,共6页Journal of Foshan University(Natural Science Edition)

基　　金：国家自然科学基金(10576013);广东省自然科学基金(5300889);广东省科技计划项目(2005B109)

摘　　要：首先导出一致锐度的概念,并将一致锐度与偏微分方程结合起来,摒弃了梯度算子和传统PDE模型的一些缺陷,建立了一个既能根据不同图像区域性质而调整检测尺度,又能有效降低噪声对检测判断影响的新的边界检测模型。在理论上,该PDE模型适定可靠,在图像处理实例上也得到了比较理想的效果。On the basis of the combination of partial differential equation and the homogeneity acutance, we get rid of many drawbacks caused by either the gradient or the traditional PDE model. After that, we establish a new edge detection model which can not only adjust the detection scale according to the actual properties of image, but also effectively reduce the impact of noise.Theoretically, this PDE model is suitable and reliable. Meanwhile, the results of practice and application are satisfactory, too.

关 键 词：边界检测 偏微分方程 一致锐度 噪声 

分 类 号：O241.82[理学—计算数学]