Research On Implementation And Optimization Of Aerial Image Mosaic Algorithm Based On GPU Architecture

In recent years,with the rapid development of UAV remote sensing technology,UAV aerial technology has been applied to many scenes of our lives,such as: security patrol,terrain survey,disaster assessment.However,the avatar image contained in the image information is limited,which in some specific tasks can not meet the application requirements.The specific contents of this paper are summarized as follows:(1)In this paper,digital image mosaic technology is used to stitch the images acquired by UAVs,so as to obtain a large-angle aerial image and expand the field of view of aerial images.(2)In this paper,we use GPU parallel computing technology and CUDA C language to realize the aerial image mosaic based on SIFT algorithm.Experiments show that the combination of GPU parallel computing technology and SIFT algorithm can greatly improve the execution speed of SIFT algorithm,and also indirectly improve the efficiency of aerial image stitching.(3)the dark channel a priori algorithm is used to dehaze the aerial image,and the brightness of the dehazed image is adjusted by using the automatic color scale algorithm and contrast-limited adaptive histogram equalization(CLAHE)algorithm.The CLAHE algorithm has a better effect on the dehazed image brightness adjustment,and also improves the accuracy of the subsequent image registration and the quality of the stitch image.