Navigation Line Extraction Method for Ramie Combine Harvester Based on U-Net

摘要

To solve the problem that when the ramie combine harvester is operating, a small amount of ramie's tilt and omission will interfere with the extraction of the ramie boundary and affect the accuracy of extracting navigation line, this paper proposes an algorithm for extracting navigation line for ramie combine harvester based on U-Net neural network. The algorithm uses the U-Net neural network to segment the collected images, which obtain the segmentation results quickly and accurately. In addition, this paper further improves the accuracy of the navigation line by preprocessing the boundary points obtained by the segmented images. Using the algorithm in this paper to process 300 images in the test set, and the results show that the algorithm proposed in this paper can accurately segment the ramie area and obtain the navigation line of the combine harvester. The average angle error of the extracted navigation lines is 1.16°, and the average processing time is 0.17s. The algorithm proposed can provide a reliable and real-time navigation line for the automatic driving of the ramie combine harvester.