Rice Row Recognition And Location Based On Machine Vision

Rice field management is one of the important links in rice production.In the 4 age of agriculture,the use of automatic machinery instead of artificial weeding and spraying will be necessary for the mechanization of rice production.In order to extract rice line reference line with machine vision and automatic control agricultural machinery to track rice line for field operation,this paper carried out the research on line recognition and location of paddy field image collection platform based on machine vision.The main contents are as follows:(1)According to the characteristics of rice images,a rice row image segmentation algorithm is studied.Rice rows in different color models were analyzed,and RGB model was used to study rice row images.In the rice grayscale transformation,the traditional 2G-R-B algorithm is improved,and the Otsu method is applied to binary image processing.(2)An artificial bee colony clustering least squares algorithm for rice row recognition is proposed.The artificial bee colony clustering algorithm and the least square method are combined to detect the line of rice line,and the results of the linear treatment of rice line by least square method and Hough transform are compared and analyzed.280 images of rice rows were tested and the correct recognition rate was 91%.(3)An image acquisition system for paddy fields is designed.A paddy field wheel walking platform with brushless DC speed motor driving rear wheel and electric push rod motor driving front wheel steering is designed.A walking driving control system and an image acquisition system controlled by wireless remote control are developed.The Labview program obtains the rice row image from the CCD camera,and calls the image processing program written in MATLAB to realize the rice row diagram recognition.(4)The location parameters of paddy field walking platform in paddy field were determined.A board lattice calibration board is used to calibrate the image acquisition camera of the paddy field.The camera internal parameter and the average projection error of 0.13 Pixels are obtained,and the parameters of the offset and the angle of the paddy field image acquisition platform relative to the rice line are analyzed and determined.(5)Field experiment of paddy field image acquisition platform,rice row recognition algorithm and field experiment of image acquisition platform were carried out.The experimental results show that the paddy field image acquisition platform can walk smoothly in the paddy field,the speed of walking is 0-1.12m/s,the paddy field wheel has little damage to the rice plant,and 400 images of rice are collected.The algorithm correctly identifies 374 images and the correct recognition rate is 86%.By measuring the paddy field image collection platform,the platform is relative to rice when walking between the lines.The offset and deviation parameters are 12.18 mm,maximum error 25.8mm and root mean square error 8.66 mm,the average deviation is 6.75,the maximum error is 9.7,and the root mean square error is 2.53 degrees.