Visual positioning method and device based on map construction

Abstract

The invention discloses a visual positioning method and device based on map construction, and the method comprises the steps: S1, obtaining an image of a surrounding environment, extracting the feature information of the image, and constructing a three-dimensional point cloud map with spatial information and the feature information; and S2, acquiring the two-dimensional image in real time by the imaging equipment, extracting feature information of the image, constructing a mapping relationship between the two-dimensional image acquired in real time and the three-dimensional point cloud map according to the extracted feature information, and positioning the position of the imaging equipment in the three-dimensional point cloud map according to the mapping relationship. The method has the advantages of being simple in implementation method, high in positioning precision, high in robustness and the like.

Description

technical field [0001] The present invention relates to the technical field of visual positioning, in particular to a visual positioning method and device based on map construction. Background technique [0002] Visual positioning is to use the image information collected by image acquisition equipment to achieve positioning. In the prior art, visual positioning methods are usually implemented based on optical flow or feature points, but this type of visual positioning method has the following problems: [0003] 1. Since the feature points are generated by the image, the location accuracy of such feature points is low, which will directly affect the positioning accuracy. [0004] 2. The feature points directly extracted by vision still have unstable problems. For example, the feature points may be obtained from the reflection of fallen leaves and tall buildings on the road. The position of this feature will change greatly or even disappear in a short period of time. Positi...

AI Technical Summary

Problems solved by technology

[0003] 1. Since the feature points are generated by images, and the location accuracy of such feature points is low, it will directly affect the positioning accuracy

[0004] 2. The feature points directly extracted by vision still have unstable problems. For example, the feature points may be obtained from the reflection of fallen leaves and tall buildings on the road. The position of this feature will change greatly or even disappear in a short period of time. Unstable features are used for positioning, which will make the positioning less robust

[0005] In application scenarios such as unmanned driving, it is necessary to provide map data for vehicles. The map data is usually laser point cloud data, but laser point cloud images cannot be directly applied to visual positioning

Patent application 201910587663.0 discloses a calibration method for visual mapping. The method obtains the image data collected by the visual sensor, determines the motion trajectory of the visual sensor based on the image data, generates an initial visual map based on the motion trajectory and image data of the visual sensor, and then To calibrate the initial visual map, this method can be used for visual positioning by constructing a visual map, but the map construction method is complicated, and the actual positioning accuracy of the visual positioning achieved by relying on the map construction method is not high

Images