Research On The Key Techniques Of Tomatoes Harvesting From Greenhouse Based On Binocular Stereo Vision

Tomatoes and fruit from Greenhouse are now commonly picked by manual picking.With the rising cost of labor and the continuous development of agricultural robot technology,new requirements are proposed to solve the technique of unmanned tomato picking in greenhouses.This research has designed the software control system of tomato picking robot,based on Binocular Stereo Vision,used by the key technology of visual system identification and positioning technology.The research contents are as follows:For the problem of tomato image recognition and extraction,under the condition of dynamic change of light intensity and the shade of the fruit leaves,the dissertation proposes an optimized circular Hough transform algorithm which is combined with chromatic aberration to dynamically identify tomato images.The problem of low accuracy of tomato recognition is solved under the condition of the change of illumination,environmental and sheltered fruit.The optimized Hough transform is used to extracting the tomato profile and optimizing the unit size setting of accumulator,to extract the the target tomato and remove the extraneous noise preferably.The accumulator in Hough transform is used to finding center point in the unit size(Fault tolerance)to reduce operation time.After the improvement,time performance is improved about 40% and the accuracy of recognition is 95% with the chromatic aberration and optimized Hough circle algorithm by the tested 100 samples.To determine the three-dimensional spatial coordinates of tomato fruit and analyze the two-dimensional information in the image,the dissertation designs and realizes the three-dimensional coordinates of tomato based on principle of binocular stereo vision.The software of tomato picking and identification system was tested and the error analysis is made on the three-dimensional coordinates of the obtained tomato.The test results show that the detection distance is the highest in the range from 300 mm to 400 mm,which meets the operational requirements of the actual picking robot.Based on the above research results,the dissertation designs the upper computer software system of tomato picking robot using by VS2010.The software is tested and could accurately obtain the three-dimensional coordinates of the tomato under laboratory conditions.This dissertation completed the research on the key techniques of tomatoes harvesti ng from greenhouse based on Binocular Stereo Vision.The results provide a foundation f or the fruit and vegetable picking robots and play an important role in improving fruit and vegetable picking efficiency and product quality in the shed environent.