Method for Detecting Obstacles of Riceplanter Based on Machine Vision

Rice planting machine in the paddy field automatic operation project, the realization of the automatic steering of the machine at the field stalk is the key technology to realize the autonomous navigation of the whole process of the rice planting machine. Aiming at the problem of discriminating the turning position in the visual navigation of the rice planter, this paper proposes a discriminating method of the rice field stem based on machine vision. Obtain distortion parameters through camera calibration to correct the original image, and calculate the deviation of the gray average value of the image line through python to determine whether the field stem appears. Binary and morphological processing of the field stem image, the processed image is scanned in the height direction to obtain the characteristic points of the field stem boundary line fitting, and finally the least square method is used to fit the feature points to draw the field stem borderline. The test results show that the discrimination rate of whether the field stem appears is not less than 90%, and the average pixel error of the field stem boundary is 4.954 pixels. It can meet the real-time and accuracy requirements of the rice planter during the operation process.