Research On Recognizing And Locating Of Fruit Based On Binocular Stereo Vision

Robot technology has entered a period of rapid development and the robot intelligent picking operation for improving the traditional manual labor is a very big advantage. China is a major agricultural production and fruit industry occupies the important proportion. In order to reduce labor intensity of picking fruit, improve the picking speed and realize intelligent picking, using robot technology to picking fruit is an effective way. In this paper, the grape fruit as the analysis object. Using binocular stereo vision technology of image acquisition, target recognition, picking location, spatial orientation and picking robot virtual simulation aspects were studied to guide the robot to realize intelligent picking. The main research contents of this paper are as follows:(1) Through the analysis of the color space of the grape fruit image, this paper uses the histogram fuzzy clustering algorithm combined with the H component to achieve a better effect on grape target segmentation. For H component in complex conditions produce incorrect segmentation phenomenon, this paper presents based on H-? ×I component combined with threshold segmentation technology to improve the recognition rate of XiaHei grape segmentation. According to the model of the single color by threshold segmentation technique is easy to be influenced by the ambient light, the histogram back projection segmentation contains grape fruit image, and the method of specific color quality grapes has better segmentation effect.(2) Grape fruit image as the research object, through the analysis of the grapes in the natural environment of growth morphology, this paper on fruiting pedicels unconcluded state of fruit picking point identification and location were studied, by a straight angle threshold, trusted zone, point to linear minimum distance and depth distance constraints of traditional grape picking point positioning fruiting pedicels mode is improved, the realization of the picking point located in the effect of grape stems Central.(3) Based on binocular stereo vision, the calibration and positioning principle of monocular camera and binocular camera were studied. The space positioning principle is studied of no parallel structure of binocular cameras and parallel structure of binocular camera calibration, and ultimately selected binocular parallel structure of space positioning coordinates as grape harvesting robot picking point’s scheme in three-dimensional space positioning.(4) The spatial position calculation of fruit picking point and stereo matching of grape fruit picking point were researched. In view of the existing stereo matching algorithm is analyzed. In order to improve accuracy of spatial location of grape picking point by selecting the appropriate stereo matching function as the stereo matching method.(5) The virtual experiment simulation of harvesting robot picking process was studied. The camera image acquisition, picking point positioning, robot kinematics and reasonable trajectory planning were studied by using virtual robot model with binocular vision. The realization of the robot picking process virtual simulation provides some references for research robot virtual experiment with binocular vision.