Abstract: Accurate extracting of litchi fruits’ complete contour is very important to automatic recognition and picking of robots. On the basis of ant colony and Fuzzy C-Means（FCM） clustering algorithm, an effective color images segmentation algorithm using the suitable L*a*b* color space is proposed. The algorithm firstly uses of L*a*b* color space that the positive and negative values of a* channel represent red and green color for initial segmentation, and then determines the FCM initial clustering center according to ant colony algorithm, Finally, using the space constraints FCM completely segments litchi fruit. The results show that the proposed algorithm not only completely segments the litchis, but also meets the robot subsequent litchi recognition and picking, and the correct segmentation rate is up to 87%.

Key words: color image segmentation, litchi image, color space, Fuzzy C-Means（FCM）, ant colony algorithm

摘要： 准确地提取荔枝果实的完整轮廓对采摘机器人自动识别与采摘至关重要。以蚁群和模糊C均值（FCM）聚类为理论基础，选用符合荔枝颜色特性的L*a*b*颜色空间，提出一种基于蚁群和带空间约束FCM的荔枝图像分割算法。该算法利用L*a*b*颜色空间的a*通道正轴代表红色和负轴代表绿颜色进行初始分割，然后利用蚁群聚类算法全局性和鲁棒性的优点确定FCM的聚类中心，用引入空间约束的FCM完整地分割出荔枝果实。实验结果表明此方法实现了荔枝图像完整地分割，并且满足了采摘机器人后续的荔枝识别与采摘，对成熟荔枝分割的正确率达到了87%。

关键词: 彩色图像分割, 荔枝图像, 颜色空间, 模糊C均聚类, 蚁群算法