Methods Of Rotary Tillage Path Identification In High Stubble Paddy Cultivation

Because of the difference of the driver’s proficiency, it will easily appear heavy plow tillage leakage phenomenon during the high crop stubble field rotary tillage, in order to eliminate this kind of human error, improve work efficiency, we can join navigation path recognition system to rotary cultivator. At present it has few about high crop stubble field rotary tillage visual navigation path identification methods of research in China, In this tudy we choose high crop of stubble field rotary tillage image as the research object, choose Dongfanghong tractor LX-854 as the test platform, and carried out for high crop stubble field rotary cultivator visual navigation path recognition method for the research, the concrete research content is as follows:(1) The image acquisition system platform was design on Dongfanghong LX-854 tractor, the platform allows free adjustment of the installation location of cameras and shooting Angle, finally according to the requirement of film camera installation height of 2.4 m, shooting Angle and horizontal plane into 23.5° Angle downward.(2) Different denoising method was proposed to deal with high crop of stubble field rotary tillage image was researched, by comparing the treatment results finalized choose type mask smoothing is the most suitable for high crop of stubble field rotary tillage image denoising method.(3) The characteristics about high crop of stubble field rotary tillage image in HSI color space, found that only color characteristic values can be used to distinguish between soil area in the image and I straw area boundary.(4) Histogram about the color values I of high crop of stubble field rotary tillage image was analyzed, proposed over and compensation arithmetic, by setting the threshold to distinguish border straw with soil area in the image region, applied the algorithm to identify the image area soil with straw regional boundary information error is not more than 1%.(5) The boundary threshold analysis software was written, the use of this software can quickly detect the image edge threshold information under different experimental environment.(6) The photo camera was calibration, identified the camera’s internal and external, internal average error is 0.15 pixel, distortion coefficient is 0.18.(7) The least square method for high crop straw with stubble fields tillage soil areas in the whole image area boundary point was used to carry on the fitting, combining with the camera internal and external parameters calibration results finally got the navigation path information.(8) The tractor control system was designed and the path recognition program was written, and conducted on campus navigation path recognition system of road surface test and field experiment, road experiment in real time is 5.63 mm, lateral deviation caused by the handle turning Angle of deviation from the mean is 1.38°, field experiment real-time lateral deviation error is 8.03 mm, caused by the handle turning Angle deviation is 1.28°...