Method for intelligently classifying and splicing multiple disordered images

Abstract

The invention provides a method for intelligently classifying and splicing a plurality of disordered images. The method comprises the following steps of: 1, carrying out feature matching on the plurality of disordered images by adopting an improved BBF-based K-D tree feature matching method; 2, defining splicing relevancy to describe the correlation between the images, and defining splicing credibility to describe the correlation strength between the images; 3, performing sorting priority width traversal on the related images to realize high-speed intelligent classification of the disordered images; and 4, reducing image splicing errors by optimizing a transmission type splicing method, and combining an image splicing line and a fusion algorithm to realize image splicing. Practice shows that faster feature matching is realized, more accurate image correlation description quantity is defined, higher-speed intelligent classification and splicing of the images are realized, and the methodis successful practice of intelligent classification and splicing of the disordered images, is practical, efficient, accurate and rapid, can be easily expanded, and solves the problems of intelligentclassification and accurate splicing of multiple disordered images.

Description

technical field [0001] The invention relates to a disordered image classification and splicing method, in particular to a method for intelligently classifying and splicing several disordered images, and belongs to the technical field of disordered image splicing. Background technique [0002] Image stitching is widely used in remote sensing image processing, virtual reality and other fields. It is an important content in the field of computer vision and image processing, and has good market value and application prospects. With the popularization and enrichment of image acquisition equipment, image storage and network information sharing are more convenient, the number of images has increased sharply, and different scenes and disordered images are more common. These images may be taken by different users in different scenes, different equipment, and different perspectives. The relative relationship between them is unknown, and they belong to the multi-scene disordered image ...

AI Technical Summary

Problems solved by technology

[0010] The second is that the existing image stitching technology is mainly aimed at a single sequence of sequential images. The stitching correlation between the sequence images is known or relatively simple, mainly for feature point extraction, matching between adjacent sequence images, etc.

How to achieve the matching of a large number of features more quickly; how to define a more precise quantity to describe the correlation between images; if the images are not from the same group of scenes, how to realize the intelligent classification and splicing of disordered images at a high speed remains to be solved ;

[0011] The third is that the existing technology cannot calculate the image stitching matrix based on the image classification results, and cannot realize the stitching of several disordered images; it cannot quickly and accurately realize the intelligent classification and stitching of disordered images, and cannot use disordered images, so the visual effect of stitching is better wide-angle images; cannot intelligently and reliably realize low-altitude image data processing such as drones; cannot process images of large amounts of data and different qualities shot at different times, different users, and different scenes, and cannot realize classification and retrieval of massive images , cannot meet the needs of the industry;

[0012] Fourth, the existing technology cannot achieve fast feature matching, and cannot realize high-speed intelligent image classification and stitching. The design and implementation of complex and large-scale disordered image classification and stitching is weak, and it is only suitable for small-scale image classification and stitching. And the accuracy is not high, the portability is poor, generally only used in specific fields, and there are defects such as weak interactive performance, low intelligence, low scalability, slow classification speed, and low splicing accuracy.

Images