3D Reconstruction of Camellia Oleifera Fruit Recognition and Fruit Branch based on Kinect Camera

摘要

In order to study the design of the 3D vision system of the camellia fruit picking robot and realize the outdoor operation of the camellia fruit picking robot, this paper takes the mature camellia fruit as the research object to identify the mature camellia fruit within the field of view of the image collection device, and use point cloud data to identify the camellia. Fit the contour of the camellia fruit, and the 3D model of the connected branches, determine the position of the picking center point of the camellia fruit in the 3D space, and finally determine the direction angle of the camellia fruit to provide it to the camellia fruit picking robot. The posture angle can smoothly avoid the fruit branches and achieve the purpose of picking mature camellia fruits by imitation. Experimental results show that the algorithm in this paper realizes the three-dimensional reconstruction of branches while realizing the precise segmentation of fruits, which provides ideas and methods for outdoor branch segmentation.