摘要
分形是实现自然纹理描述的一种有效方法.本文从分数布朗随机场模型的描述出发,证明了具有不同分形特性的两种纹理边缘处的H参数估计值大于1的一般结论、将H参数泛化用于描述一般图像的特性,从而提出了一种基于广义多尺度分形参数的图像边缘检测方法.该方法通过构造多分尺度的广义分形参数H值,从而可以实现图像边缘的有效提取.
Fractal is an effective tool for the description of nature textures. A general conclusion that the Hparameter estimated in two different textures boundary is greater than 1 is proved which based on fractional Brownian random field model, Hparameter is generalized for the description of image properties and a method for edge detection is proposed based on generalized multiscale fractal parameter. Through construction of generalized multiscale fractal parameter H. image edges can be detected effectively.
出处
《系统工程与电子技术》
EI
CSCD
1997年第1期39-44,共6页
Systems Engineering and Electronics
基金
国家自然科学基金资助课题
关键词
图像分析
机器视觉
边缘识别
图象处理
Fractal, Generalized mulliscale fractal parameter. Fractional Brownian random field, Edge detection.
