The Recognition And Reconstruction Of Camellia's Main Branches Based On Machine Vision

Camellia oleifera is a unique high-quality oil crop in China and plays an important role in the national oil security strategy.Aiming at the low efficiency and high cost of Camellia oleifera fruit harvesting,which restrict the development of Camellia oleifera industry,this paper presents a scheme of Camellia oleifera fruit picking based on machine vision.We focus on the identification and reconstruction of main branches of Camellia oleifera,combined with the experience of solving the problem of fruit picking at home and abroad,which provides a visual basis for the automatic picking of Camellia oleifera fruit.The research contents of this paper include:(1)A vibration picking scheme of Camellia oleifera fruit based on machine vision is proposed.Traditional fruit picking mainly focuses on identifying fruits and single picking.Considering the characteristics of Camellia oleifera,this paper puts forward a scheme of identifying branches and vibration picking,which can simplify the picking process and improve the efficiency of Camellia oleifera fruit picking.(2)A binocular vision system was built to acquire the branch images of Camellia oleifera.The system is composed of a calibrated parallel binocular camera,which is the hardware basis for spatial localization.(3)An algorithm flow for identifying and reconstructing the main branches of Camellia oleifera was proposed.The stereo matching of binocular image was carried out to obtain the disparity map of branches.The foreground and background of the disparity map are separated by the thresholding segmentation method.An improved parallel fast refinement algorithm is used to process the segmentation results and generate the branch skeleton.The burrs were removed according to the neighborhood features of the skeleton pixels to maintain the original tree structure.The key points of the skeleton were detected,and then the branches were graded and reconstructed to obtain the model of branches.The research shows that the vibration picking scheme has many advantages,such as high efficiency,easy realization and so on.This paper solves the problems of branch identification,localization and reconstruction in the process of vibration picking of Camellia oleifera fruit,and provides a visual basis for the whole automatic picking system of Camellia oleifera.