Unmanned aerial vehicle image splicing method

Abstract

The invention discloses an unmanned aerial vehicle image splicing method comprising the steps that SIFT feature matching pairs of images under matching photographed by an unmanned aerial vehicle are acquired so that the initial transformation parameters of other images under matching are obtained; supposing that the transformation parameters of the images are set as x, x refers to a column vector formed by eight independent parameters of a homographic matrix, T refers to transformation represented by x, X=(X<T><1>, X<T><2>,...,X<T><M>)<T>, and a target function E(X)=cE(oX<r>)+omega(EX)r is defined, wherein omega refers to constant weight, Ecor(x) refers to the corresponding energy term based on feature points and Erig(x) refers to the bound term of the transformation parameters x; the features of the SIFT feature matching pairs are correspondingly substituted in the target function, and optimization is performed by using an LM algorithm; and fusion processing is performed on the transformed images. According to the unmanned aerial vehicle image splicing method, all the transformation parameters are maintained to be rigid transformation as much as possible so that the splicing result is enabled not to have global deformation, each image is enabled to maintain the original shape, the global deformation of the outputted splicing image is enabled to be effectively controlled and the great splicing effect can be acquired with no requirement for parameter and attitude information of a camera.

Description

technical field [0001] The invention relates to the technical field of UAV image processing, in particular to a method for mosaicing UAV images. Background technique [0002] UAV reconnaissance technology is a kind of remote sensing technology, which has its own advantages compared with satellite reconnaissance, such as low cost, flexible reconnaissance area control, no access time and cycle restrictions, and high resolution of ground targets; compared with manned reconnaissance aircraft, It has the advantages of continuous work day and night, regardless of pilot fatigue and casualties. In recent years, due to the advantages of high resolution, high flexibility, high efficiency and low cost of low-altitude remote sensing data, UAVs have been widely used in natural disaster area assessment, battlefield reconnaissance, environmental monitoring and other fields. Although UAV reconnaissance technology has played an important role in areas such as assessment of natural disaster ...

AI Technical Summary

Problems solved by technology

Although this algorithm brings a good registration effect to image registration, it does not improve the image stitching speed much. For large-scale UAV reconnaissance images, it is necessary to change the parameters of the SIFT algorithm. It is more difficult to achieve the effect of real-time image processing, and UAV image stitching usually requires sensors such as GPS / IMU to provide the attitude information of the camera, or use image-based 3D reconstruction methods (such as structure-from-motion) to restore the camera's attitude. Attitude, the calculation efficiency is relatively low, and the image effect after splicing is not good

Images