Panoramic video camera unit spatial pose calibration method

Abstract

The invention relates to a panoramic video camera unit spatial pose calibration method. The method includes the following steps that: step 1, spatial pose calibration is performed on all adjacent cameras; step 2, independent constraint relations are listed according to the relative pose relations of the adjacent cameras; step 3, a correction equation is listed according to the independent constraint relations, adjustment is performed on constraints generated by redundancy calibration according to an adjustment method, so that a correction vector can be obtained; step 4, a measured value is modified according to the correction vector; and step 5, a correction result is substituted into the relative pose relations, and the step 2 is repeated until a corrected result satisfies all independent constraint relations. The panoramic video camera unit spatial pose calibration method of the present invention can satisfy the requirements of larger and more comprehensive field-of-view image splicing techniques, has higher pose determination relation accuracy and require fewer cameras.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a method for calibrating the spatial pose of a panoramic video camera group. Background technique [0002] 360-degree panorama is a low-cost virtual reality technology based on image stitching technology, and it is a research hotspot in the field of virtual reality and computer vision technology. Panoramic stitching technology, as a cheap and intuitive implementation, has been widely used in many fields, such as real estate, decoration design, tourist attractions, engineering reports, virtual campuses, street view services, video surveillance, etc., and has a broad market demand . [0003] Before performing panoramic stitching image processing, the relative pose of the camera group needs to be calibrated. The spatial relationship of the cameras is fixed. Therefore, different cameras have different internal and external parameters and height angles. In order to obtain panoramic v...

AI Technical Summary

Problems solved by technology

[0004] There are many methods for calibrating the internal parameters of the camera, and the method of calibrating two cameras is difficult to apply to the environment with a large field of view. Traditional dual-camera calibration or multi-camera calibration generally requires the cameras to be calibrated to be A target (such as a calibration board) is imaged at the same time, so that the target coordinate system can be used as a reference to unify each camera to be calibrated, and then the relative pose relationship between the cameras to be calibrated can be obtained

The above-mentioned calibration method is not suitable for occasions that require a larger field of view, because if it is necessary to use fewer units to obtain a larger field of view, then there must be fewer overlapping image parts, so that there will be a greater impact on the calibration results. It is very difficult to use the calibration target to calibrate the pose relationship between cameras in this limited overlapping field of view. When using a calibration block with a high-precision three-dimensional structure, although each camera can collect a picture , but the use of the calibration block is very inconvenient, and when using the calibration board, it is required to collect at least three calibration board images with different attitudes, and it is also very difficult to change the calibration board to different attitudes in a limited field of view; in addition, due to Image distortion, wide-angle lens imaging generally has a large distortion at the edge of the image, and the common field of view of adjacent cameras is just at the edge of the field of view of each wide-angle lens camera, so the accuracy obtained by traditional calibration methods is also Not guaranteed

Images