本算法综合了传统的利用毯子维数提取图像边缘的方法和Fuzzy聚类的方法,对具有不同维数的物体边缘进行提取.通过改进Fuzzy聚类方法,克服了传统Fuzzy聚类方法中只考虑分维谱图中像素自身而忽略分维谱图中邻近像素聚类特点的缺点;改进后的Fuzzy聚类方法同时弥补了传统的利用毯子维数提取图像边界时迭代次数过少的不足.实验结果表明,利用该算法较传统算法更好地描述了物体的边界.%This paper proposes to combine the traditional ways of using blanket method and Fuzzy clustering method, extract the edge of image, whose fractal dimensions are different. The new way of Fuzzy clustering overcome the default of neglecting the feature of agency pixel while doing the cluster. The new way also makes up the default of not doing enough iteration, while using the traditional ways of using blanket method to extract the edge of image. The result of experiment shows that the new algorithm does much better than the traditional algorithm while extracting the edge of image.
展开▼