| Parameter measurement based on machine vision is an important direction in the development of intelligent harvester.Focusing on the measurement principle of cutting width and rice and wheat height,crop boundary line extraction,crop height line extraction and algorithm verification,real-time cutting width measurement and rice and wheat height measurement of intelligent rice and wheat harvester are studied The main research content of the paper is:The principle of height measurement and width measurement is studied and the measurement image is acquired.The conversion principle of world coordinate system to pixel coordinate system is studied.Inverse perspective transformation is used to measure the cutting width.Aiming at the problems of low efficiency and waste of labor in measuring the height of the crop using traditional measurement methods,the operating characteristics of the harvester were analyzed and a visual-based measurement method was proposed.The camera is selected and the appropriate image processing module is also selected according to the complexity of the algorithm.Crop boundary extraction method based on improved particle filter and back-projection algorithm was studied.Aiming at the problem that the crop boundary line is difficult to extract and is easily affected by some factors,the change law of particles’ weights near the crop boundary are analyzed and a crop boundary line location method based on improved particle filtering is proposed.To solve the problem of inaccurate positioning of the boundary,a method of relocating the boundary position by the back-projection algorithm is proposed.Firstly,the improved particle filter is used to determine the new region of interest.Then the back-projection algorithm is used to obtain the probability map.Finally,the crop boundary is obtained through image processing and the crop boundary extracted by the two algorithms are optimally selected.The method of measuring the height of rice and wheat based on color character is studied.To solve the problem of high time complexity,the Gaussian pyramid is used to downsample the image.To obtain the target region,the HSV gray value change law of the image is analyzed and the H channel is used to segment the target region.The inter-frame sum method is used to enhance the image,and the large-scale window’s median filtering and morphological operations are used for hole filling.The inter-frame sum method is used to enhance the image,and the large-scale window’s median filtering and morphological operations are used for hole filling.The complete upper boundary line of rice and wheat is obtained and height measurement is realizedExperimental verification.Based on the characteristics of field experiments,a reasonable verification method for evaluating the measurement accuracy of the cutting width and the height of the crop is designed.To analyze the accuracy of different algorithms for extracting crop boundary line and crop height line,videos in different environments were collected.Experiments show that the improved particle filter and back-projection algorithm can be used to measure the cutting width of the harvester accurately,and the method based on the color fteature can be used to measure the height of rice and wheat accurately and quickly.According to the test results of the National Agricultural Machinery Quality Supervision and Inspection Center,the average absolute error of width measurement is less than 70 mm,and the average absolute error of height measurement is less than 20 mm. |
