Relevancy extracting method and device

Abstract

The invention discloses a relevancy extracting method and a relevancy extracting device. The relevancy extracting method comprises the following steps: S1, performing linear processing on a present relevancy matrix so as to obtain a novel relevancy matrix; S2, acquiring initial offset according to the novel relevancy matrix; S3, performing speed tracking so as to obtain final offset; S4, acquiring the size of a window for calculating the relevancy of a next frame according to the present offset data. On the basis of an original speed tracking mechanism, speed weight and linear processing are introduced to adjust the relevancy matrix and the window range size, the relevancy with high reliability between two frames is obtained, the search range is small, the splicing speed is greatly improved, and an image can be truly reduced according to the speed when an acquired original image is poor in effect. The offset data can be also applied to image detection and navigation. The image detection sensitivity and precision can be improved. The spliced image is good in effect, and phenomena of conspicuous image malposition, image deformation, image shortness, image streaking and the like can be corrected.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a correlation extraction method and device. Background technique [0002] Image stitching is to combine multiple small-size images with overlapping areas from the same scene into a large-size high-quality image. Image mosaic technology solves the problem that large pictures cannot be produced due to the limitation of the viewing angle and size of the acquisition window. Image mosaic technology was first used in the aviation field, and has been widely used in digital video, dynamic analysis, fingerprint reconstruction and other fields. [0003] The prerequisite for image stitching is that a part of the adjacent images is logically the same, and there must be a certain overlap. Image stitching is mainly to determine the degree of overlapping of adjacent images in width and height. This degree of overlap is called the correlation of the two images. [0004] ...

AI Technical Summary

Problems solved by technology

[0004] In the existing method, a tracking mechanism is also added in the correlation calculation, but the invention of this mechanism is not perfect, resulting in obvious image dislocation, image deformation, short image, image tailing and other phenomena in the spliced ​​image, which is in It will have a serious impact on many links in the back-end processing, directly affecting the image recognition rate

Images