Research On UAV Aerial Image Mosaic Algorithm Based On Improved SPHP Algorithm

UAVs are extensively used in many fields by virtue of their low mass,agility and convenience.Due to the restrictive viewing angle of the camera carried by the UAV,it is difficult for a single photo to reflect the actual situation on the scene.Therefore,drones are required to take multi-directional shooting to obtain images taken from various angles,and then stitch them together to obtain an Omnidirectional image,that is,a panoramic image,which is used to restore the actual situation of the scene.Therefore,the research on UAV image stitching technology is very important.The traditional image stitching algorithm takes too long to operate,and the stitching image efficiency is low.In addition,the quality of the drone itself is small,and it is susceptible to the influence of external natural force majeure,causing parallax problems in the images to be spliced.To attack these problems,We have done the following four parts:(1)This paper discusses the research background and current situation of UAV aerial image mosaic technology,the research status is divided into three stages,and the advantages and disadvantages of existing algorithms are analyzed.The basic process of image stitching is analyzed,research on image distortion correction and image denoising,image enhancement technology to achieve preprocessing.It also analyzes the image registration,the solution of transformation models,the principle of image fusion and common algorithms.(2)Based on the Delaunay triangulation method,a FAST-Tomasi feature and Delaunay triangulation elimination algorithm is proposed.After the algorithm extracts the feature points quickly and accurately,it uses the triangulation net to eliminate the mismatched points.It not only improves the speed of the whole image stitching,but also obtains the fine matching points,which ensures the subsequent stitching accuracy.The results indicate that,in this paper,the proposed algorithm is more efficient in rejection,and the matching accuracy is also enhanced,which is more suitable for the stitching of UAV aerial images.(3)An improved SPHP image stitching algorithm is proposed.The target image to be spliced is deformed according to the TPS function model,the deformed image can be accurately aligned with the source image,eliminating parallax.The SPHP algorithm is used to smoothly extrapolate the image to a non-overlapping area to obtain a stitched image.The results indicate that compared with other algorithms,this paper effectively solves the ghosting problem caused by parallax.Improve the accuracy of image registration during registration,and it improves the distortion problem of the edge image mainly due to the shape stretching,so that the final image is smooth and natural.(4)Based on the above research results,the functions of UAV image preprocessing,feature extraction,feature point elimination,image registration and image fusion have been realized,and a software system has been developed by integrating these functions.The software interface is simple and easy to understand.Users can easily select different images to complete the stitching.