Multi-view camera pose estimation method, apparatus, device and medium

By using the Plucker line method and the random consistency sampling algorithm, the limitation that each camera needs to observe three points in a multi-camera model is overcome, expanding the application scenarios of visual positioning and improving the robustness and accuracy of attitude estimation.

CN117291975BActive Publication Date: 2026-06-23GUANGZHOU SHIYUAN ELECTRONICS CO LTD +1
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
GUANGZHOU SHIYUAN ELECTRONICS CO LTD
Filing Date
2022-06-20
Publication Date
2026-06-23

AI Technical Summary

Technical Problem

In existing technologies, multi-view camera models require each view to observe three points in order to perform pose estimation, which limits the application scenarios of visual positioning.

Method used

The Plucker line method combined with the random consistency sampling algorithm is adopted. By obtaining the position of each camera relative to the body and the two-dimensional and three-dimensional coordinates of the observation point in the multi-view camera, the coordinates and depth of the straight line formed by the observation point and the camera are calculated. The attitude combination is calculated by combining the Umeyama method and the random consistency sampling to reduce the influence of noise.

Benefits of technology

It expands the application scenarios of visual positioning, eliminating the need for each camera to observe three points, thus improving the robustness and accuracy of attitude estimation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117291975B_ABST
    Figure CN117291975B_ABST
Patent Text Reader

Abstract

The application relates to the technical field of computer vision, and provides a pose estimation method, device and equipment of a multi-view camera and a medium. The method comprises the following steps: acquiring the positions of each camera relative to the body of the multi-view camera; acquiring the two-dimensional coordinates of observation points observed by each camera and the three-dimensional coordinates in a world coordinate system; calculating the straight line coordinates of straight lines formed by each observation point and a corresponding camera by using a plucker line method according to the two-dimensional coordinates of each observation point and the position of the corresponding camera relative to the body; calculating the depths of each observation point according to the straight line coordinates of the straight lines formed by each observation point and the corresponding camera; calculating the coordinates of each observation point in a body coordinate system according to the depths of each observation point; and calculating all possible pose combinations according to the coordinates of each observation point in the world coordinate system and the coordinates of each observation point in the body coordinate system. According to the application, each camera does not need to observe three points, so that the use scene of visual positioning is expanded.
Need to check novelty before this filing date? Find Prior Art