Splicing method suitable for video images of multi-focal-length camera

Abstract

The invention discloses a splicing method suitable for video images of a multi-focal-length camera. The splicing method comprises the following steps: (1) establishing coordinates for the position ofa single camera image of the multi-focal-length camera with a known arrangement mode in a panoramic image; (2) determining an overlapping region between adjacent images, and determining feature pointsof the overlapping region; (3) matching the feature points of the overlapping region, and removing the feature points with wrong matching; and (4) screening the feature points which are correctly matched, and removing the feature points which are wrongly matched again. The panoramic coordinate system is established according to the arrangement positions and the focal lengths of the camera pictures for multi-focal-length camera picture splicing, the specific overlapping area of the adjacent pictures is determined, error matching prone to occurring is avoided to a large extent, and the matchingprocess is accelerated; furthermore, an effective error detection mechanism is provided for a matching result, and the accuracy of multi-focal-length camera image feature matching can be improved.

Description

Technical field: [0001] The invention belongs to the technical field of high-definition digital video, and in particular relates to a splicing method suitable for video images of multi-focus cameras. Background technique: [0002] For fusion and stitching of multiple images with a certain overlapping area, the core technology is to find common feature points between two adjacent images. Obtain a certain number of feature matching pairs, and two camera coordinate transformation matrices can be calculated through the coordinates of the matching pairs. When there are more cameras, one camera is selected as a reference, and the images of other cameras are mapped to the plane where the camera image is located. Therefore, accurately finding feature matching pairs is the key to determine the quality of the final panoramic image. Currently, the commonly used technologies for obtaining feature points include SUFR, SIFT, and ORB. Then, by pairing the feature points between the imag...

AI Technical Summary

Problems solved by technology

However, once the image feature points are relatively similar and the number of image paths is large, it is easy to cause wrong matching.

Possible error situations are: 1. A certain number of feature matching pairs can be determined between two actually non-adjacent images due to the similarity of the scene; 2. There are some obvious mismatches between adjacent images, although currently The RANSAC technology is used to make mistakes, but it is not guaranteed to completely eliminate such mistakes

The above-mentioned similar mis-matching will lead to a large calculation deviation of the transformation matrix between images or the calculation will not converge, which will directly affect the effect of the final panorama

Images