Weak connection image stitching method based on line features

Abstract

The invention relates to a weak connection image stitching method based on line features. The weak connection image stitching method based on line features includes the steps: measuring the image coordinates of N pairs of uniformly-distributed homonymous dot pairs in the homonymous region of left and right images, and finding homonymous intersecting line pairs or parallel lines in the left and right images; according to the image coordinates of the measured image points, calculating the equation of the lines, and calculating the image coordinates of the intersection point of the lines; regarding the corresponding intersection point of the lines as the homonymous point, substituting the image coordinates into the equation of the vertical parallax of the images, and solving the relative orientation parameter between the images together with the common homonymous point; and according to the relative orientation parameter obtained through solution, performing geometric transformation on the right image, obtaining a new image overlapped region, taking the center line of the overlapped region as the stitching line to stitch the two images. The weak connection image stitching method based on line features not only constructs more homonymous point observed values, but also enables distribution of point location to be more ideal so as to improve the stitching accuracy of weak connection images, and is suitable for stitching processing of various weak connection condition images.

Description

technical field [0001] The invention relates to a method for solving the relative positional relationship between images, in particular to a method for determining and splicing the relative positional relationship of weakly connected images, and belongs to the fields of surveying and mapping and computer vision. Background technique [0002] When using aerial images for topographic mapping or constructing panoramic images from images taken in scenes, it is necessary to stitch the images to construct a full-frame image of the entire shooting area. In order to ensure the accuracy of image stitching, the core step is to determine the relative positional relationship between each image . At present, the image coordinates of N pairs (N>5) of point pairs with the same name between two images are obtained mostly through image matching, and the relative positional relationship of the images is solved by the relative orientation of the images. Generally speaking, to obtain high r...

AI Technical Summary

Problems solved by technology

However, in some special cases, the connection conditions between images are very weak, and the degree of overlap does not even exceed 10%. It is difficult to obtain relatively high-precision relative orientation results using the traditional relative orientation method of corresponding points of the same name.

Images