摘要
本文在离散分形布朗增量随机场(DFBIR)的图象模型基础上,利用在图象处理中常用的金字塔技术,通过提取特性参数——H值和方差,并结合传统的聚类分割方法,实现了图象的快速分割。通过对自然背景中的人造物体进行分割实验,结果表明:这种快速算法不仅能得到较好的分割效果,而且可以将速度提高将近三分之一倍。
In this paper, we first introduce a fractal image model of the discrete fractional Brownian increment random field,then we perform fast image segmentations by extracting two property parameters H and variance and using pyramid technology, where the traditional clustering technique is combined. Experiments on the man-made objects in natural background show that this fast algorithm not only can work in a running pyramid,but also save much time.
出处
《计算机应用与软件》
CSCD
1998年第2期34-40,60,共8页
Computer Applications and Software
基金
国家自然科学基金
