Unmanned aerial vehicle image splicing method based on grid optimization and global similarity constraint

Abstract

The invention provides an unmanned aerial vehicle image splicing method based on grid optimization and global similarity constraint. The unmanned aerial vehicle image splicing method specifically comprises the steps of extracting feature points of a reference image I and a target image I'and the corresponding relation of the feature points through an SIFT feature point extraction algorithm; meshing the reference image and the target image; solving a local homography matrix of the reference image and the target image after meshing by adopting a local DLT method; performing grid optimization byusing a minimized energy function to align the gridded target image with the reference image so as to eliminate ghosting of an overlapping region and obtain a preliminary spliced image; taking the ground as a landmark, and performing global similar coplanar constraint on the preliminary spliced image to eliminate projection distortion of a non-overlapping part to obtain a final spliced image. Thebeneficial effects of the invention are that the method can achieve the better matching of the features of the images of the unmanned plane through combining the features of the images photographed bythe unmanned plane, achieves the precise registration, and achieves an ideal splicing effect.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a method for mosaicing UAV images based on grid optimization and global similarity constraints. Background technique [0002] UAVs have the functions of automatic take-off and landing, automatic driving, automatic navigation, automatic fast and accurate positioning, automatic information collection and transmission, etc. It is especially suitable for replacing humans to complete tasks in difficult, harsh or extreme environments. Drones have a wide range of applications in military mapping, aerospace and commerce. However, it is difficult to cover the entire target area due to the limitation of a single view. To obtain a complete scene of the desired target, multiple images need to be stitched together. Image stitching is a technique that combines overlapping areas of multiple images to form a panorama. [0003] For drone images, it has some characteristics. UAVs are small in ...

AI Technical Summary

Problems solved by technology

When the drone shakes slightly, the camera cannot be adjusted in time, and the collected images will definitely be affected

The most important problem is image parallax

Also, there are different depth differences in the image due to the undulations of the ground

When using the above method to stitch UAV images, due to the influence of these characteristics, some ghosting and distortion will occur

Images