Design And Implementation Of Multi-exposure Fusion System Based On Aerial Image

Aiming at the exposure problem of aerial images,a multi-exposure fusion system based on aerial images is designed and implemented.Based on the multi-exposure image fusion function,the single image is adjusted and the exposure parameters are fed back by automatic dimming,and image registration is added before the multi-exposure image fusion to deal with the moving characteristics of aerial images.The specific contents of this topic are summarized as follows:Firstly,an automatic dimming method based on human visual characteristics is proposed to solve the problem that the dimming method can not adjust the aerial image to suitable exposure when aerial photography is of complex brightness.Firstly,the gray scale information of the image is counted to get the gray scale range and specific position of the region of interest,and different weights are allocated according to its size,so as to reduce the influence of the region of interest on dimming.Finally,the adjusted image with visual characteristics is obtained,and the average gray value of the adjusted image is used as the exposure feedback value in this scene to call back the imaging equipment.Secondly,in view of the relative movement of the camera and the imaging target in aerial multi-exposure images,the method of registration before fusion is adopted for image correction.In view of the characteristics and registration requirements of aerial multi-exposure images,SURF algorithm is selected as the basis,and UAV POS information is applied to aerial image registration.Through the transformation relationship between image coordinates and ground coordinates,the overlapping areas between aerial multi-exposure images are calculated,so that registration operations can only be carried out in this area,thus improving the matching accuracy and shortening the registration time.Thirdly,in view of the fact that the dynamic range of aerial photography equipment is much smaller than that of real aerial photography scene,the Laplacian pyramid multi-resolution fusion method is used for exposure fusion,and the method is converted into YUV space to preserve the color information of aerial photography image and prevent chromaticity deviation.On this basis,three new measurement factors are proposed to construct the weight map of aerial photography image.To further improve the efficiency of the algorithm and the final fusion effect.Fourthly,the multi-exposure fusion system based on aerial images is designed and implemented.Firstly,the overall scheme of the system is completed based on the actual needs,and the above three improved algorithms are modularized.Then,the system interface is designed and built to make the exposure processing of aerial images more convenient.Finally,the relevant functional items of aerial images are added to the system to realize a multi-functional aerial image fusion system.In summary,firstly,this paper makes a theoretical analysis of aerial images with distinct light-dark differences,and makes a thorough study on auto-dimming,aerial image registration and multi-exposure image fusion technology,and achieves relevant theoretical achievements such as auto-dimming based on human visual characteristics,image registration combined with flight data and multi-exposure fusion in YUV space.Finally,based on these research results,a complete multi-exposure fusion system for aerial images is developed and implemented.