Tic-tac-toe structured light camera system and method for three-dimensional reconstruction of target with scale

Abstract

The invention discloses a well-shaped structured light camera system and a method for performing scaled three-dimensional reconstruction of a target, belonging to the field of computer vision; the system of the invention combines a well-shaped structured light emitter with a monocular industrial camera, and can reconstruct the target 3D point cloud with scale; the reconstruction of 3D point cloud mainly includes the following steps: 1. Calibration of the parameters of the well-tac-toe structured light camera system; 2. Image recognition of the well-tac-toe mark; 3. Scale factor estimation and optimization; The scale factor between real objects, thus restoring the scale-free 3D model to the real object size.

Description

Tic-tac-toe structured light camera system and method for scaled 3D reconstruction of target technical field The present invention relates to the three-dimensional reconstruction field in computer vision, be specifically related to a kind of well-shaped structured light camera system and A method for performing scaled 3D reconstruction of objects. Background technique 3D reconstruction technology is the most popular and classic research direction in computer vision, and it has a variety of methods. and methods, in which the monocular multi-view reconstruction method has been more and more widely used in long-range 3D reconstruction of spatially non-cooperative targets. , because of its simple device, low power consumption, not limited by baseline, suitable for long-distance reconstruction, and the reconstructed point cloud contains Color and texture information is beneficial to subsequent applications. What the camera obtains when shooting is a two-dimensional imag...

AI Technical Summary

Problems solved by technology

[0003] What the camera obtains when shooting is a two-dimensional image of the target object or scene at a specific angle. The target image lacks real depth information, so only a normalized three-dimensional model can be obtained by using monocular reconstruction, and the model cannot reflect the real size of the target.

However, a 3D model that lacks a scale does not have much application value. For example, in a long-distance scene, the normalized 3D model of a non-cooperative target can be obtained. In a medium-distance scene, the normalized 3D model information cannot be used to obtain the motion of a non-cooperative target. and location information

Usually monocular cameras are configured with more complex depth cameras to obtain the required size information, but these depth cameras also cannot work effectively at long distances

Images