提出了一种新颖的图像分割算法,用于提高仿人机器人足球比赛中机器人获取的颜色质量.为了使仿人机器人快速获得颜色图像,采用HSI颜色空间,并对现有算法进行了改进,根据每个像素点HSI3值中有关I值的公式对当前像素的主要特征进行判断,采用了模糊K均值聚类的方法进行图像分割.根据每个像素点S值的分布,动态地决定当前选用的是H值还是I值,这样就能够有效地消除HSI模型的不足,使得机器人的视觉能够适应复杂的比赛环境.实验结果表明所提方法有效. A novel image segmentation algorithm is proposed to improve the color quality of robots in the humanoid robot soccer game.In order to make the humanoid robot quickly obtain the color image, the HSI color space is adopted and the existing algorithms are improved. According to The formula of HSI3 value in each pixel is used to judge the main features of the current pixel, and the method of fuzzy K-means clustering is used to segment the image.According to the distribution of S value of each pixel, the current selected Is the value of H or I, which can effectively eliminate the shortage of HSI model and make the robot’s vision adapt to the complicated competition environment.The experimental results show that the proposed method is effective.