Bundle adjustment algorithm with photographic scale

Abstract

The invention discloses a bundle adjustment algorithm with a photographic scale. Firstly, three-dimensional coordinate initial values of measured object points are obtained by using a conventional three-dimensional reconstruction algorithm, then a photographing scale of imaging of each object point in a corresponding camera is calculated according to the three-dimensional coordinate initial values of the measured object points, an objective function with the photographing scale is established, and camera parameters and the three-dimensional coordinate initial values of the measured object points are optimized to obtain final camera parameters and three-dimensional coordinates. The method is suitable for three-dimensional coordinate reconstruction when the distance between a camera and a mark point changes greatly in industrial photogrammetry, and the calculation result is reliable.

Description

technical field [0001] The invention relates to the technical field of visual measurement, in particular to a bundle adjustment algorithm with a photographic scale. Background technique [0002] In the fields of aerospace, automobile manufacturing, etc., industrial photogrammetry technology often requires a certain number of coded marker points to be arranged on the measurement site, and cameras are used to collect images of marker points from multiple directions, and then 3D reconstruction of marker points is performed. In order to improve the reconstruction accuracy, it is necessary to use bundle adjustment to perform nonlinear optimization on the initial reconstruction value to obtain high-precision three-dimensional coordinate values. However, the conventional bundle adjustment has the following limitations in the practical application of the system: [0003] In the actual system application, the distance between the optical center of the camera and the same marker poin...

AI Technical Summary

Problems solved by technology

[0003] In the actual system application, the distance between the optical center of the camera and the same marker point is different in each orientation, and the spatial positioning error corresponding to the reprojection error of the same size on different imaging planes of the marker point will also be different. Therefore, The spatial positioning error represented by the reprojection error of each imaging plane cannot be unified on the same scale

The conventional bundle adjustment uses the reprojection error on the imaging plane as the error equation, and it is difficult to evaluate the spatial positioning error of the final reconstruction result when the distance between the camera and the marker point changes greatly

Images