类人足球机器人球识别方法的研究

Robocup类人组比赛中，对球的快速、准确识别是占据比赛主动权的关键。针对实际比赛中，场地光照在空间和时间上分布不均匀，常用的颜色阈值分类方法不具有鲁棒性，容易出现机器人找不到球或丢球现象的问题，在降维的HSI颜色空间模型上，提出了一种非线性支持向量机结合查找表的分类方法。仿真表明，此方法能有效、快速地进行球识别，在应对光照条件变化的情况下，也具有很好的鲁棒性。

Research on ball identification method of humanoid robot soccer

In the Robocup Soccer Humanoid League, the fast and accurate identification of the ball is key to take the initiative in competition. As the light of the field is different in the distribution of space and time, the robots cannot find the ball or eassly lose the ball using the common classifying method of color-threshold. So,based on the color space model of reducing dimension HSI, a classifying method of nonlinear - SVM combined with look-up table is proposed. Simulation results show that this method is effective on ball identification quickly, and it also has good robustness in the case of variable illuminations.