一种鲁棒的彩色地图分割算法  

A New Segmentation Algorithm for Color Map Image

作　　者：李晓峰[1] 曾平[1] 

机构地区：[1]西安电子科技大学外部设备研究所,西安710071

出　　处：《计算机工程与应用》2003年第22期62-64,共3页Computer Engineering and Applications

基　　金：部委预研基金;陕西省自然科学基金(编号:2001x06)资助

摘　　要：研究了一种按颜色分割彩色地图的新算法。根据人类初级视觉的特性,先大致将图像分割为低频区和高频区,然后进行颜色聚类和区域生长。该方法不需人为指定聚色中心和类数,阈值的选定不对最终分割质量起决定性作用,具有较好的鲁棒性。A new segmentation algorithm for color map image is presented in this paper.The image is firstly segmented into aeras of low frequency and high frequency with the consideration of the characteristic of HVS.Then color clustering and region growing are made.This algorithm doesn't need artifically specified centers and quantity for the color cluster-ing.Meanwhile,it has a good robustness that the choice of the thresholds doesn't make crucial effect on the quality of the final segmentation

关 键 词：彩色地图分割 人类初级视觉 聚类 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]