采摘机器人扰动状态下柑橘动态识别

Dynamic recognition of oscillating orange for harvesting robot

为了提高柑橘采摘速度和采摘效率,提出一种采摘机器人在柑橘扰动并伴有遮挡状态下的快速动态识别方法.首先对所采集的两帧图像进行分析,通过颜色空间的选取以及阈值分割,将水果与背景进行分离;其次通过帧间差分法、水平最小外接矩形法、正方形内切圆法将处于扰动状态下的水果标识出来,然后对振荡水果进行识别.试验结果表明:该算法能很好地适应采摘机器人实际工作中碰到的、各种原因引起的水果扰动及遮挡现象;单个扰动水果算法识别时间小于0.4 s.

In order to improve picking speed and picking efficiency, a fast dynamic identification method was proposed for harvesting robot with orange oscillating and covering. The two acquired successive oscil- lation orange images were analyzed to separate the fruit image from background by color space selection and threshold segmenting method. The dynamic region of oscillation fruit was labelled by the inter-frame difference method, the minimum enclosing rectangle method in horizontal direction and the square in- scribed circle method. The oscillation orange was located and identified. The experimental results show that the outstanding performance of the proposed method can be used to solve the harvesting problems caused by various reasons in practice. The recognize time of sole target is less than 0.4 s.