Research Of The Key Techniques Of Apple Automatic Bagging Based On Machine Vision

The agricultural robots’ research began in1982, in Kyoto University, after a shortperiod of more than20years, such as harvesting、transplanting、spraying、 baggingvarious robot has been extensively studied for the purpose. Application of machinevision in agricultural robot research is a step up research topic of the world’s countries,there are two difficult problems have always been the focus of:(1)、the unevenillumination;(2)、the plants grow random cause part of the fruit is not visible. Applebagging robots based on machine vision can identify and positioning on young Applefruit, thus guiding the robot arm to bagging young fruit. Research on automatic baggingrobot vision system has an important practical significance in accelerate the process ofagricultural robot research、increase agricultural robot type、change the current situationof mainly rely on manual for the sack at present.Apple’s young fruits are studied in this paper, aimed at the visible light images,which are acquired under the natural environment, the key technology of imagerecognition and stereo vision-based localization are researched of the young fruit. Themain research content is as follows:1、Young Apple fruit image pre-processing. To preprocess the young apple images,through comparison of experimental results and analysis on a wide range of imagesegmentation methods, experiments showed that Otsu segmentation algorithm is simple,faster, better segmentation, based on the considerations of real-time of the robot visionsystem to be able to meet a variety of complex environmental, the Otsu algorithm isused in this paper of image segmentation algorithm.2、Young Apple fruit image recognition. Recognize the young apple’s image on thebasis of Preprocessing. At first, using the improved connected components labelingalgorithm to marking and labeling the preprocessed young apple image, removing smallareas that do not meet the requirements, reduce the interference, and then extracting the image contour. Using the shape characteristics of circular degree to identify the youngapple fruit and automatically tag.3、The Visual positioning of young Apple fruit. The MATLAB Calibration toolboxis used for camera calibration in this paper, and the Census transform matchingalgorithm is used to match the two images captured by the binocular camera. Finallydisplayed the three dimensional image of the young apple based on the OpenGL.4、Verifying and comparing the above algorithm under the MATLAB and VC++programming design environment, and the Simulation and experimental results showthat the method is effective.