The Influence Of Different Lighting Conditions On The Recognition And Location Of Apple Picking Robots

As the output of fruits and vegetables is increasing year by year and the work intensity of picking workers is too great,it is urgent to study the picking robots of fruits and vegetables.At present,the image segmentation and positioning when the difference between target and operating environment color is not obvious becomes a difficult point for picking machine vision research.Based on binocular stereo vision,this paper takes different types of apples as research objects,realizes effective segmentation of apples of different colors,and applies the best calibration environment obtained by experiments to improve the positioning accuracy of apples.The main research contents are as follows:Research on apple images,based on color difference theory,a method for converting an image from an RGB color model to an HSV color model and reweighting the three elements of the image under the model is proposed,and an iterative criterion is finally designed to effectively segment the image.Through the experiments in the orchard operation environment and simulated lighting conditions of fruit and vegetable greenhouses during the night,the applicability of this method to the coexistence of red,green,yellow,and mature(red)and unripe(green)apples was demonstrated.Using MATLAB to calibrate the binocular camera,the Zhang Zhengyou calibration method was adopted under the sun illumination conditions,and the effects of the position,posture,luminous flux,luminous flux,and calibration plate size on the calibration accuracy of the dual-targeting and centering camera were tested.Through the analysis of the experimental results,the best calibration conditions in the outdoor natural lighting environment are finally determined,and compared with the calibration experiment under the artificial light source to obtain the best calibration environment that ultimately improves the positioning accuracy of the picking robot.Aiming at the relationship between artificial lighting conditions at night and calibration accuracy,simulate the night lighting environment of fruit and vegetable greenhouses,select the most commonly used lighting sources in machine vision,collect calibration images and use MATLAB2015 b to achieve binocular camera calibration.Comparing the best calibration environment under natural light and artificial light conditions,back-projection error was used as an index to measure calibration accuracy.A suitable solution is provided for all binocular vision as the apple picking vision system,which can guarantee the accuracy of camera calibration,thereby improving the positioning accuracy.By comparing and analyzing the advantages and disadvantages of the global optimization matching method and the feature point matching method,the feature point matching method based on the polar line matching principle is adopted.The least square method was used to locate the coexistence of red,green and mature and immature apples.