Vision-based cleaning area detection method

Abstract

The invention discloses a vision-based cleaning area detection method, which comprises the following steps of: processing a video through a visual mileage calculation method, constructing a local map, classifying local point cloud data based on an interested area, performing plane fitting on ground point cloud by using a plane fitting algorithm based on a random sample consensus (RANSAC) algorithm, and detecting a cleaning area according to the plane fitting algorithm. Carrying out structured processing on the edge detection point cloud data, and carrying out radius filtering and centroid calculation; a traditional adjacent point slope method is improved, road boundary points are extracted, curve fitting is carried out on road edges, adaptive angular point extraction is carried out on different areas by applying a dynamic grid strategy, and continuous detection of the road edges is realized. According to the method, the contradiction between the calculation speed and the calculation precision is balanced, accurate segmentation of the road edge is realized, and thus accurate extraction of the sweeping area of the sanitation vehicle is realized.

Description

technical field [0001] The invention relates to an area detection method, in particular to a vision-based cleaning area detection method, which belongs to the field of visual detection. Background technique [0002] With the rapid development of the economy, traffic accidents are showing a high incidence trend, and advanced driver assistance systems and automatic driving play an extremely important role in ensuring road traffic safety. Road edge detection is an important part in the field of environmental perception research. The determination of the road edge isolates the drivable area and other surrounding non-drivable road areas for the vehicle. The drivable area detection is the basis for vehicle decision-making and execution, and provides a basis for subsequent assisted driving. Or automatic driving tasks provide a strong guarantee, and are an important part of vehicle intelligent navigation and assisted driving technology, which further guarantees the driving safety of...

AI Technical Summary

Problems solved by technology

However, this solution has the following deficiencies: first, the solution uses lidar for data collection, and it is often difficult to use expensive lidar in the field of assisted driving, which increases the deployment cost

Second, the scheme uses a method based on the neighborhood elevation feature to extract ground points, and the neighborhood elevation feature method has poor tolerance to the smoothness of the road surface, and it is easy to cause false detection when the road surface is poor in smoothness and there are foreign objects on the road surface

However, the scheme has the following problems: First, the scheme extracts the road area through the semantic segmentation network, but in complex environments, such as cracks on the road surface and large areas of shadows, there are problems of inaccurate segmentation, and the environmental stability of this segmentation method is relatively low. Difference

Second, although this solution can segment the edge of the road, it cannot perceive the distance from the vehicle to the edge of the road, which has obvious limitations in application scenarios

However, this scheme has the following problems: First, this scheme uses the canny edge detection algorithm to extract road edges, but in complex environments, such as when there are cracks on the road surface and large areas of shadows, there are problems of large calculation and high false detection rate. At the same time, the solution is easily affected by changes in ambient light, so the accuracy of edge detection and the ability to adapt to the environment need to be further improved

Second, although this scheme can calculate the actual distance from the edge of the road to the vehicle, it needs to calibrate the correspondence between the actual distance between the edges of the road and the pixel distance between the edges of the road in the image coordinate system in advance. The accuracy of distance calculation also depends on the calibration accuracy. Since the internal and external parameters of the camera will inevitably change over time, the accuracy of distance calculation will decrease.

Images