Camera online calibration method based on small target blurred image

Abstract

The invention relates to a camera online calibration method based on a small target blurred image, which comprises the following steps: enabling a high-precision small circular light-emitting point target to be close to a lens as much as possible, and ensuring that the target obtains a large image proportion and a placement angle; adjusting a camera aperture and exposure time to obtain an identifiable target image; establishing a target feature point positioning uncertainty mathematical model, and solving the positioning uncertainty of each feature point; By taking target feature point positioning and parameter uncertainty as constraints, solving high-precision solutions of internal and external parameters of the camera through a nonlinear optimization method; according to the method, thefocusing space and the target size of the camera do not need to be considered, and the ultrahigh-precision calibration of the camera can be realized under the conditions of large image blurring and noise only by placing a very small high-precision target at a position close to a lens. The method is suitable for online calibration of large-view-field and shallow-depth-of-field cameras and even cameras in use, and has great flexibility and adaptability.

Description

technical field [0001] The invention relates to the technical field of sensor calibration, in particular to an online camera calibration method based on a small target fuzzy image. Background technique [0002] The camera is the most important core component in the field of computer vision, and camera calibration plays an important role in the computer vision system. Existing camera calibration methods require the target to be placed in the focal space of the camera to ensure a clear target image, and it needs to occupy a large area in the image. At the same time, the position of the target needs to have a large angle change, so that high-precision Camera calibration results. However, for large-field cameras represented by fisheye and catadioptric cameras, high-precision small and medium-sized targets in the camera’s focal space occupy a small area in the image, while large targets have low precision and are difficult to process. Placement is extremely inconvenient; for te...

AI Technical Summary

Problems solved by technology

However, the calibration methods based on these two types of targets are limited by the algorithm itself and have the disadvantage of insufficient calibration accuracy.

In addition, there are also many self-calibration techniques that do not require targets. These methods have the advantage of not using targets, and are widely used in video tracking, mobile robot visual navigation and other fields, but the calibration accuracy is average, and they are not suitable for applications that require high-precision camera calibration. occasion

[0004] However, in the existing main camera calibration methods, the target must be placed in the focus space, and the target size is required to be large enough. At the same time, the imaging is required to be clear and the feature point extraction is accurate, so as to achieve relatively high calibration accuracy.

Images