Method for detecting autonomous navigation path of plant protection robot, server and storage medium

Abstract

The invention discloses a method for detecting an autonomous navigation path of a plant protection robot. The method comprises the following steps: acquiring an image; carrying out segmentation preprocessing on the image; carrying out region division on the ridge row, carrying out noise reduction by adopting morphological operation, extracting the edge of the ridge row, and then carrying out further noise reduction by utilizing a dynamic area threshold noise reduction method; extracting feature point clusters by adopting a mean value method, and optimizing the feature point clusters to obtainan optimal feature point set; and extracting a navigation path by using a least square method in combination with an RANSAC rejection algorithm.

Description

technical field [0001] The invention relates to a method for detecting an autonomous navigation path of a plant protection robot, a server and a storage medium including the method. Background technique [0002] The inventors found that the feature of the ridge line is the core visual feature based on the autonomous navigation of the plant protection robot, and the Hough transform and the least square method are the mainstream methods for detecting the feature of the ridge line, that is, the navigation path. Although the methods themselves are relatively mature, they are not effective when applied to navigation path detection. The reason is that the interference characteristics caused by the irregular size of weeds and crop stems and leaves seriously affect the detection accuracy of navigation paths. Expressed as: [0003] 1) For plant protection robots operating in the field, crop row can provide structured navigation information. In image preprocessing, due to the influe...

AI Technical Summary

Problems solved by technology

In the image preprocessing, due to the influence of weeds between the ridges, it brings a large error to the feature point extraction, which in turn affects the final detection accuracy.

At present, the traditional image morphology processing method is mainly used to solve the weed interference, but the effect is not ideal

[0004] 2) The Hough transform method to detect the straight line features of discrete point clusters pays more attention to the collinearity of local feature points, which leads to the fact that some area points in the row form a line alone, and it is difficult to detect a complete straight line. The detection result is composed of multiple straight lines. , which is not conducive to the extraction of plant protection robot navigation parameters; and the processing speed of Hough transform is slow in the use process, the parameter conditions are difficult to set, and it is difficult to meet the real-time navigation requirements of plant protection robots

[0005] 3) Due to the existence of outliers, the line features extracted directly by the least squares algorithm cannot accurately and reliably reflect the actual direction of the row, which will bring non-negligible errors to the navigation of the plant protection robot

Images