Camera calibration is the process of obtaining the internal and external parameters of the camera. In the actual calibration process, the results of multiple calibrations may be different.
Possible reasons are:
- The experimental environment is different: the accuracy of camera calibration is affected by many factors, including shooting distance, angle, lighting, etc. If the experimental environment changes, the results may be different, such as different shooting distances, and the calibrated internal and external parameters of the camera are different. This is because the real camera projection process is not completely corresponding to the small hole model. There are papers After analyzing this problem, the real lens group model is used to calculate the corresponding optical center of the pinhole model, and finally a conclusion is drawn: the equivalent camera optical centers at different distances are different.
- Calibration image quality is different: the higher the quality of the image used for calibration, the more accurate the calibration result. If the images used for calibration differ in quality, the calibration results may vary.
- Calibration uses different algorithms: Different camera calibration algorithms have different performance and scope of application, and may produce different calibration results.
- Influence of the camera itself: when calibrating a monocular camera, a small hole model is usually used, but this model is not a real camera projection model (real camera lenses are usually more complicated and do not have an absolute optical center point) , is only an approximation. If the manufacturing process of a camera is relatively reliable, then the parameters of its corresponding pinhole model (focal length, principal point) will be relatively stable, which is reflected in the calibration results that the fluctuation of the internal parameter value of each calibration will be relatively small.
Solution:
- Introduce more control variables to ensure that the experimental environment is as stable as possible. Pay attention to some points when actually collecting calibration images: the flatness of the calibration board, the size of the checkerboard (as large as possible), image quality, whether the distribution of corner points is uniform, whether the corner points are Covers the entire screen. These factors will affect the stability of the internal parameters of the single calibration camera.
- Use higher quality calibration images to ensure accurate calibration results.
- Use a more accurate camera calibration algorithm and make appropriate adjustments according to the actual situation.
- The focal length of the collected calibration images should be as close as possible to the actual focal length. For example, when a camera used for intelligent driving collects calibration images, the distance between the calibration board and the camera can be 3 to 10 meters instead of within 1 meter.