| In the unstructured orchard ecosystem,the growth environment of apples is very complicated,the fruits often overlap each other or the fruits are blocked by leaves,which makes the apple picking robot unable to quickly and accurately identify and locate the target apple,which seriously affects the picking efficiency of the apple picking robot.In order to solve this problem.This paper proposes a new method for apple overlap and occlusion.The main research contents and conclusions are as follows:This paper introduces three kinds of segmentation algorithms,K-means algorithm,OUST threshold segmentation and color difference method,and compares the three algorithms to compare their effects.The experimental results show that the average similarity of the K-means segmentation algorithm is 96.886%;the average segmentation error is 2.562%;the average running time is 0.754 s.In addition to the running time,the segmentation performance of the K-means algorithm is higher than the other two algorithms,and the running time of the algorithm is less than 1 second,which indicates that the algorithm can extract the targets in the apple image more stably,quickly and accurately.Therefore,the K-means segmentation algorithm is selected to segment the Apple image.A method based on graph theory segmentation algorithm Nuct and curvature contour method combined with cubic spline interpolation algorithm is proposed to identify and locate overlapping apples.The steps of the method are as follows: in the first step,the K-means segmentation algorithm is segmented and the pre-processed restored color image is segmented twice using the graph theory segmentation algorithm Nuct to extract the contours of the two apples,and the second step uses the curvature The contour is extracted without obscuring the apple outline and the occluded apple outline,and finally the occluded apple outline is reconstructed using a three-spline interpolation algorithm.The experimental results show that the average error of the algorithm is 3.156% and the average similarity is 96.083% for the apple profile without occlusion.For the occluded apple outline,the average error of the algorithm is 5.131%,and the average similarity is 94.709%,which proves that the method can segment and reconstruct overlapping apple targets stably and accurately.In the natural environment,the apple target is often blocked by branches,which will affect the picking efficiency of the picking robot.To solve this problem,this paperfirst USES the k-means segmentation algorithm to extract the contour of the apple target,then USES the convex shell to extract the real contour segment,and USES the method that three points not on the same line can be used to determine the circle.In order to avoid the problem of large difference in parameter estimation caused by different selection points,three points were randomly selected on the real contour segment.After multiple selection,the average values of all centers and radii were calculated and used to fit and locate the occlusion apple target.According to the experimental results,the average positioning error of the algorithm is 7.795%,and the overlap coefficient is89.477%,which proves that the algorithm can locate the occluded apple target stably and accurately.Use the GUI Edit Toolbox to identify and locate Apple target software systems in MATLAB with overlapping and occlusion design.The system has three modules which are the effect comparison module of the segmentation method,the overlapping apple target segmentation and fitting module and the occlusion apple target positioning module. |
