Posture information determination method and apparatus, visual point cloud construction method and apparatus

Abstract

Disclosed are a posture information determination method and determination device, a visual point cloud construction method and a visual point cloud construction device. The posture information determination method includes determining relative posture information of an image acquisition device when acquiring an image of a current frame relative to when acquiring an image of a previous frame; Determining, by a sensor with an absolute scale, a first set of translation parameters for movement of the image acquisition device during acquisition of images of the current and previous frames; Adjusting the relative posture information based on the relative posture information and a first set of translation parameters; And determining posture information of the image acquisition apparatus when acquiring an image of a current frame based on the adjusted relative posture information. In order to obtain more accurate posture information, the translation vector of the camera posture information isdirectly calibrated by the scale of the external sensor using the posture information determination method.

Description

technical field [0001] The present application relates to the field of computer vision. Specifically, the present application relates to a method for determining pose information, a device for determining pose information, a method for constructing a visual point cloud, a device for constructing a visual point cloud, electronic equipment, and a computer-readable storage medium. Background technique [0002] Maps are the foundation of the autonomous driving field. However, in the monocular camera SLAM, due to the scale uncertainty of the monocular camera, it is impossible to construct a vector map with a consistent global scale, and due to the scale uncertainty of the monocular camera, the monocular SLAM cannot track images in multiple frames. It is easy to accumulate tracking result errors due to scale drift, which eventually leads to tracking failure. [0003] In the existing technology, the real three-dimensional scale of the map point is directly obtained at each moment ...

AI Technical Summary

Problems solved by technology

However, although the real scale of inter-frame images can be obtained by using binocular vision technology or IMU modules, due to the high cost of sensors, high computing costs, high production costs, complex calibration, complex algorithms, and the cost of the IMU itself Larger, making the use of visual point clouds a huge obstacle

Images