一种改进的类人足球机器人彩色目标识别算法  

Improved color object recognition method of humanoid soccer robot

作　　者：潘亚宾[1] 刘国栋[1] 

机构地区：[1]江南大学物联网工程学院,江苏无锡214122

出　　处：《计算机应用研究》2011年第3期1195-1197,1200,共4页Application Research of Computers

摘　　要：针对类人足球机器人视觉需求,提出一种结合区域生长和基于形状判别的阈值自适应更新的彩色目标识别算法。该算法在HSI空间基于S分量把图像分为高饱和与低饱和区域,在高饱和区域基于H分量采用区域生长算法识别目标;通过目标形状判别自适应更新阈值,并用新阈值更新区域生长中原来的阈值,以稳定准确地识别彩色目标。在类人足球机器人系统中的成功应用表明,该算法能在不同光照条件下稳定地识别出彩色目标,对光照环境有良好的适应性和鲁棒性,具有良好的识别效果。As the vision demands of humanoid soccer robot,this paper proposed a color object recognition method that combined region growing with the adaptive threshold update algorithm based on shape discrimination.The method divided the image into the high saturation region and the low saturation region by the S component in the HSI color space.Used the region growing to recognize the object by H component in the high saturation region.It adaptively updated the threshold by shape discrimination and used in region growing in order to accurately recognize the color object stably.The successful applications of the method on a real humanoid soccer robot show that the method has well adaptable and robust for changing illumination because color objects are recognized stably under different illumination condition.So the method has good recognition effects.

关 键 词：类人足球机器人 彩色目标识别 区域生长 阈值自适应更新 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]