Identification And Positioning Of Corn Leaves Based On Binocular Stereo Vision

Modern agricultural development technology has entered the high and new technology application research stage.Machine vision technology,image processing technology,3S technology and computer technology are gradually applied to agricultural production.Agricultural production needs to get the location,physiological parameters and environment of field crops timely to regulate the growth management of crops.Crop leaves are important organs for photosynthesis of crops,which can also reflect the information of crop diseases and insect pests,growth status and so on.In this paper,the identification and positioning of maize leaves were studied by using binocular stereo vision technology and plant modeling technology.In this paper,a binocular stereo vision system is established under the background of the field,and the research method of image processing of maize leaves is put forward,and the mathematical model and three-dimensional model of the heading stage of the field maize plant are established.In this paper,the blade mathematical model combined with the incomplete blade outline in visual image can deduce the shape of the whole blade.The identification and location of each leaf of maize plant were realized.The contents of this research paper have a certain reference value for the identification and localization of other crop leaves.In order to realize the monitoring ofcropstimely,sprayingaccurately,intelligentharvestingandother technologies,build the foundations.The main contents and conclusions as follows:(1)the binocular stereo vision system was established and the application was written.The binocular system is composed of two models of the same type oflarge MER-500UC-Lcamera,two M1224-MPW lenses,a cloudplatform,a bracket and a laptop computer.By using Opencv and Visual studio2012,the calibration of binocular stereo vision system is realized,and precise binocular camera calibration parameters can be obtained.(2)This paper presents a research method for image processing of maize leaves under the background of field.On color space model and gray level image,image threshold,morphological operation,edge detection and image processing theory in the(3)premise,corn leaf image processing method is that the best combination channel G-R channel is selected in color space model;the image thresholding algorithm uses the maximum inter class variance method.the optimal morphological operation is to first open operation and close operation.the edge contour is extracted with the Canny operator,and the centroid center of maize is extracted with the basic rectangular centroid algorithm.(4)The three-dimensional model and the mathematical model of each leaf were established.According to the method of plant modeling,combined with the actual situation of field of corn growth,the quadratic functions of the leaves of the maize plants were obtained by curve fitting by using the measured coordinates and the ordinate data of the measured 20 equal parts of the leaves of the field maize plants.The R~2 was above 0.95 and the fitting degree was higher.Model has two important purposes,the one is that in view of the machine vision system and processed image discontinuity in the blade,combine to establish the mathematical model of blade to infer the purpose of continuous contour form the whole leaves.Second,it is possible to infer the entire three-dimensional coordinates of each leaf of the maize plant by combining with the centripetal coordinates of the maize leaves.(5)Application can realize the identification and positioning of maize plant leaves.Combining image processing technology,modeling technique,feature point matching and three-dimensional reconstruction technique,the identification and positioning of maize leaves based on binocular stereo vision were realized.