Method and device for extracting high-precision map elements

Abstract

The invention relates to a method and device for extracting high-precision map elements. The method includes: acquiring a one-to-one correspondence between a binocular image and a laser point cloud; and performing matching and depth estimation on the binocular image to obtain a corresponding binocular image. Depth map; according to the binocular image and its corresponding depth map and laser point cloud, the image features of the binocular image and the corresponding three-dimensional features of the laser point cloud are obtained; according to the three-dimensional features of the laser point cloud, the independent 3D bounding box information of the element; independent elements are extracted according to the 3D bounding box and the laser point cloud information in the box. The invention integrates the laser point cloud based on deep learning and the dual-purpose element extraction method, improves the extraction accuracy through multi-scale feature fusion, reduces the computational complexity and improves the extraction speed by filtering the non-element point cloud.

Description

technical field [0001] The invention relates to the field of high-precision map making, in particular to a high-precision map extraction method and device based on binocular estimation and point cloud. Background technique [0002] In the extraction of high-precision map elements, the semi-automatic element production method is time-consuming, requires many manual interactions, and is inefficient. Traditional pure laser point cloud extraction can provide less information and high computational complexity. SUMMARY OF THE INVENTION [0003] Aiming at the problems of high labor cost, high computational complexity and low accuracy in the traditional method of making high-precision maps by using laser point clouds, the present invention provides, in a first aspect of the present invention, a method for extracting high-precision map elements, including the following Steps: obtaining a one-to-one correspondence between the binocular image and the laser point cloud; performing ma...

AI Technical Summary

Problems solved by technology

[0003] The present invention aims at the problems of high labor cost, high computational complexity and low accuracy of the traditional method of using laser point cloud to produce high-precision maps. In the first aspect of the present invention, a method for extracting high-precision map elements is provided, including the following Steps: Obtain a one-to-one correspondence between the binocular image and the laser point cloud; perform matching and depth estimation on the binocular image to obtain a depth map corresponding to the binocular image; according to the binocular image and its corresponding depth map and laser point cloud , acquire the image features of the binocular image and the corresponding three-dimensional features of the laser point cloud; determine the 3D bounding box information of independent elements in the laser point cloud according to the three-dimensional features of the laser point cloud; Independent elements of laser point cloud information extraction

Images