摘要
介绍了一种基于色彩统计的图像分割方法,该方法主要包括两步:色彩量化和区域分割.在第一步中,选择一种基于统计聚类的算法来实现色彩量化.将所有的图像像素以少数对应的色彩标记来代替,进而形成图像级别图.此方法的重点在第二步,对得到的图像级别图做出色彩标记分布和色彩直方图的统计,并据此来寻找种子区域,同时分析图像中的色彩特征,实现种子区域的快速增长和合并.实验结果表明,此分割算法在分割效果和计算时间上均有较好性能,可以将其应用于目前系统稳定性要求和用户需求都较高的图像检索领域.
An image segmentation method based on color st-atistics is presented.This method,mainly consists of two steps： color quantiza-tion and region segmentation.In the first step,a statistical cluster algorithm i-s introduced to quantify the color image.All the image pixels are then repla-ced by several corresponding color class labels;thus forming a class-map of the image.The focus of this approach is on the second step.By making statistics of color labels distribution and color histogram of the class-map,and（using th-e statistic data to select seed regions,simultaneously,analyzing the color chara-cteristics of image,the seed region growing and segmentation are fulfilled quickly.The experimental results show that the proposed algorithm performs well both in segmentaion result and calculation time,and can be applied to the image r-etrieval field,which has high system stability requirements and user needs.
出处
《武汉大学学报（工学版）》
CAS
CSCD
北大核心
2011年第5期663-667,共5页
Engineering Journal of Wuhan University
基金
国家自然科学基金项目(编号:50478058)
关键词
色彩统计
色彩量化
图像分割
color statistics
color quantization
image segmentation
