荔枝采摘机械手果实识别与定位技术

针对采摘机械手在自然环境下对荔枝的识别和精确定位问题进行研究.通过分析自然环境成熟荔枝的颜色特征,选取HSV颜色模型进行阀值分割,去除荔枝图像的复杂背景,并利用模糊C-均值聚类法(FCM)对图像中荔枝果实和果梗进行分割,试验结果表明:有效识别果实和果梗的正确率为93.3％.通过计算果实“质心”与果梗的距离最大值确定荔枝采摘点,利用基于色调空间的彩色图像匹配法和极限约束法进行果梗采摘点的立体匹配,实现了采摘点的空间定位,定位试验结果表明:定位坐标中深度值误差小于3 cm,深度值误差率小于5.64％,能满足实际作业中荔枝采摘机械手的定位精度要求.

Fruit recognition and positioning technology of iitchi picking manipulator

The color features of mature litchi in natural environment were analyzed,and the HSV color model was chosen for threshold segmentation to remove complex background.The fuzzy C-means clustering method(FCM) was used to segment litchi fruit and peduncle.The experimental results show that the effective recognition accuracy of litchi fruit and peduncle reaches 93.3%.The maximum distance between fruit centroid and peduncle was calculated to determine litchi picking point.The color image matching method based on hue component of color image and the limiting binding method were used to realize stereo match of picking point.The positioning experimental results show that the depth value error is less than 3 cm with error rate less than 5.64%,which can meet the visual accurate position requirements of litchi picking manipulator in practical work.