基于颜色特征的图像分割方法及其在足球机器人中的应用  

Application of Color Image Segmentation Based-on Color Feature for Robot Soccer Vision

作　　者：蔡叶菁[1] 龙永红[1] 

机构地区：[1]湖南工业大学电气与信息工程学院,湖南株洲412008

出　　处：《计算技术与自动化》2008年第1期70-73,共4页Computing Technology and Automation

基　　金：国家自然科学基金资助项目(60774069)

摘　　要：提出一种新的适用于跟踪移动机器人的视觉系统的彩色图像分割技术,这种方法采用符合人类视觉特性的HIS颜色空间表示图像的颜色特性,利用人对颜色的感知来对颜色分量进行非等间隔的量化并形成特征矢量,根据HIS颜色空间三个基本分量合成的一维特征矢量直方图进行分割,并克服HIS颜色空间存在不可避免的奇点对图像分割的影响。另外采用一种基于背景图像的非均匀光照修正算法,应用在足球机器人中。实验结果表明该方法不仅使目标定位和方向角更精确,而且提高动态环境下的鲁棒性。In this paper, a novel color image segmentation technique for mobile robot tracking is proposed. HIS color space that is compatible with human＇s visual sense is used to represent image＇s color space. It takes advantage of human＇s perception to quantize color components as non - equal interval. The technique implements thresholding approach based on the analysis of the one dimensional featur^vector histogram, which is formulated by three components of HIS color space and correct influence made by nonremovable S^hgulaL^ty which is one of the disadvantages of HIS color space. A non- uniformation illumination correction method based on the analysis of the background image of playground for robot soccer is also discussed. Experimental results show that a more accurate and precise estimation of object position are achieved in robot soccer. The proposed techniques improve robustness to changing environmental conditions.

关 键 词：彩色图像分割 移动机器人跟踪 非均匀光照修正 彩色空间转换 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]