Fast self-calibration method for fish-eye lens binocular camera

Abstract

The invention relates to a fast self-calibration method for a fish-eye lens binocular camera. And internal parameters and external parameters of the fish-eye lens binocular camera are quickly calibrated by utilizing polar constraint provided by a common three-dimensional scene. Firstly, a fish-eye lens binocular camera synchronously shoots images of a common three-dimensional scene, and feature point extraction and matching are carried out. Secondly, focal lengths and radial distortion parameters of the two cameras and a basic matrix between the two cameras are solved at the same time by utilizing at least nine image matching point pairs, and an image matching inner point set is determined; and then a rotation translation relation between the coordinate systems of the two cameras is obtained through focal lengths of the cameras and basic matrix decomposition. And finally, the internal parameters and the external parameters of the binocular camera are optimized according to the image matching internal point set determined by the RANSAC. The fisheye lens binocular camera with serious radial distortion can be efficiently and accurately calibrated under the common three-dimensional scene condition, and the method is suitable for equipment with limited computing power such as a mobile phone and an unmanned aerial vehicle and has important research significance and wide application prospects.

Description

technical field [0001] The invention relates to a fast self-calibration method for a fisheye binocular camera, in particular to a method for calibrating internal parameters and external parameters of a fisheye binocular camera using epipolar constraints provided by a common three-dimensional scene. Background technique [0002] Binocular camera is a commonly used structural form in image measurement applications. Binocular camera can directly obtain the 3D geometric information of the target according to the parallax principle, so it is widely used in the fields of visual measurement, 3D reconstruction and robot navigation. Compared with conventional lenses, fisheye lenses have a larger field of view, and their imaging has greater distortion, and the closer to the edge of the image, the greater the image distortion. Therefore, before using the fisheye lens binocular camera for measurement, it must be calibrated with high precision, and the calibration accuracy directly affec...

AI Technical Summary

Problems solved by technology

However, nonlinear optimization requires a large number of correct image matching point pairs

If the influence of radial distortion is ignored in the first step of estimating the geometric relationship between binocular cameras, it may be difficult to find a sufficient number of correct image matching point pairs, so that the second step of nonlinear optimization is likely to fall into a local optimum excellent

Images