Lightweight indoor modeling method based on vertical plane and vertical line

Abstract

The invention discloses a lightweight indoor modeling method based on a vertical plane and a vertical line. The method comprises the following steps: A, extracting a vertical plane in a scene, establishing a plane-based model of the scene through region segmentation and geographical registration, and optimizing the model by using a merging optimization algorithm to obtain a complete plane-based model with scene information; B, establishing a spatial vertical line model, correcting the image by using the attitude angle of the camera, and extracting a spatial vertical line after filtering processing, matching the homonymous straight lines through a DTW algorithm and tracked through a Kalman filter, calculating the position of the vertical line in the space based on the plane model, and thenestablishing a line-based model; C, generating a hybrid model composed of a vertical plane and a line. According to the method, the defects in the prior art can be overcome, and the required equipmentcost and time complexity are reduced.

Description

technical field [0001] The invention belongs to the technical field of indoor positioning and navigation, in particular to a lightweight indoor modeling method based on vertical planes and vertical lines. Background technique [0002] In recent years, indoor positioning and robot navigation have become hot topics due to their wide application value. The premise of precise navigation and positioning is to establish an accurate indoor space model. In the field of indoor space modeling, image-based modeling methods are more effective than graphical modeling and can meet the needs of large-scale scenes with rich texture information. Today, image sensors are more common than other sensors such as laser sensors and ultrasonic sensors, which makes image-based modeling methods widely adopted by intelligent systems. A point cloud is a common model, but it has a large storage capacity and suffers from memory constraints, so it cannot be used in large scenes. Most areas of the interi...

AI Technical Summary

Problems solved by technology

Point cloud is a common model, but it has a large storage capacity and suffers from memory constraints, so it cannot be used in large scenes

Most areas of the interior scene lack sufficient texture, so without enough point features, it is difficult to model

An alternative solution is to use lightweight line and surface features to build models, but most algorithms require expensive equipment (such as radar and depth cameras) and have high time complexity

These methods are difficult to implement due to the above reasons

Images