Self-calibration method for binocular stereo vision device

Abstract

The invention provides a self-calibration method for a binocular stereo vision device, which comprises the following steps of: firstly, matching corresponding points between two view pictures acquired by the binocular stereo vision device; secondly, realizing projective reconstruction according to the matching relationship of the corresponding points; thirdly, calibrating out a transformation matrix from a projective space to a metric space according to prior constraints of parameters in a camera; and finally, calculating the internal and external parameter matrixes of two cameras. The self-calibration method has the advantages of restoring the internal and external parameter matrixes of the two cameras from the two view pictures by adopting absolute dual quadric-based self-calibration technology; guaranteeing the robustness of algorithm by adopting a ransack method to estimate the basic matrix and select matched feature points obtained by SIFT algorithm; and enabling all the restraints to internal parameters in the calibration to become linear restraints by decomposing the absolute dual quadric so as to avoid solving a non-linear equation.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a self-calibration method of a binocular stereo vision device. Background technique [0002] Binocular stereo vision is an important branch in the field of machine vision. It directly simulates the way human eyes process scenes, which is reliable and simple. Therefore, binocular stereo vision devices are of great application value in many fields such as pose detection and control of micro-operating systems, autonomous navigation and aerial survey of robots, three-dimensional metrology and virtual reality. [0003] The current binocular stereo device generally adopts the traditional camera calibration technology, that is, the camera is calibrated by using a calibration reference object. Specifically, the internal and external parameters of the camera are obtained by calculating the image coordinates of the calibration reference object in the image with known shape and size. In pra...

AI Technical Summary

Problems solved by technology

Therefore, the camera self-calibration technology based on the kruppa equation cannot meet the requirements for self-calibration of binocular stereo vision devices

[0005] The self-calibration method based on the absolute dual quadric surface is a commonly used method in camera self-calibration technology. Compared with other methods, it has many advantages such as ensuring the consistency of the infinite plane in space. Calibration technology faces a very difficult problem in practical applications, that is, it cannot guarantee that the calibrated absolute dual quadric surface has positive semidefinite (or vice semidefinite) properties

Images