基于语义的图像物体提取的新方法  

A Novel Extracting Image Semantic Objects Based on Semantic

作　　者：余卫宇[1] 曹燕[1] 余英林[1] 

机构地区：[1]华南理工大学,广州510640

出　　处：《微计算机信息》2007年第21期211-213,共3页Control & Automation

基　　金：国家自然科学基金(#60372068)资助;广东省自然科学基金(#06300098)资助

摘　　要：在图像语义研究中,提取图像中的语义物体或区域是重要的。本文首先对图像预处理,通过颜色空间的转换,在空间对图像进行K-均值分类,提取出具有语义性质的物体和区域。实验结果表明该方法是可行的,而且很有效的。The color classification process requires to partition a color image into uniform color regions. It is very important that extract interesting region or object in image semantic analysis. In this paper we propose an approach to extract semantic region based on color feature. First, we translate RGB space into Lab space. Second, we use the k-means algorithm solve clustering problem. In the end, we extract semantic object in terms of color information. Experimental results show that the color clustering give superior results in increases in cluster effectiveness.

关 键 词：彩色分类 K-均值算法 语义信息 物体提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]