基于极值的重叠苹果识别方法研究

针对采摘机器人对重叠果实无法识别采摘问题,提出了一种基于极值的重叠苹果识别定位方法。首先,利用色差法对图像进行分割;然后采用OTSU分割和孔洞填充提取轮廓;最后利用一种快速计算圆内的点到边缘最小距离的算法找到局部极大值,从而确定圆心和半径。实验表明:这种方法对提取的苹果轮廓较完整的情况定位效果较好,且实时性高,具有较强的实用性。

Research on Localization Method for Overlapped Apples Based on Maxima

In natufral scenes , the apple picking robot can not recognize overlapped apples .To solve this problem , an o-verlapped apple recognizing and locating algorithm was proposed in this paper .First, an image is segmented by chromatic difference method .Then OTSU segmentation and holes filling was used to get the outline of the apples .In the end , the center point and radius was determined by a fast calculating method , in which the minimum distance of the points in the circle to the edge of the circle was calculated and the local maximal value was found .The experiment results showed that this method worked very well when the outline of the apples is complete , and the real-time performance and practicability was very good .