The Research Of Observation Clearness Algorithm For Night Aerial Image To Earth

In recent years,with the outstanding advantages of small and medium unmanned aerial vehicle drones in various fields,it has played an increasingly important role in aerial observation of Earth,especially in low-altitude observations.In addition,as aerial observation applications become more and more widely used,the demand for observation under the night environment is increasing.Due to the particularity of aerial observation at night,this paper deeply analyzes the difficulty of image clarity,such as low contrast and poor visual visibility.It also carries out theoretical research on related algorithms and has also made useful explorations in terms of hardware feasibility.The main research content is as follows:First,in view of the general characteristics of the aerial observation images for night aerials are farther than the ordinary natural images.The main edge contour information of the night aerial observation images is seriously lost.This article is based on the theory of excellent guided filter algorithms in recent years,The key color transfer coefficients in color transfer are highlighted,and the problem of the structure weakening of aerial observation images in the night is solved.Second,for the night aerial observation image low-light background,and accompanied by complex artificial light source interference.This paper adopts the idea of color transfer to design the reference image flexibly to change the background illumination level of night aerial images.The improved color transfer algorithm is used to solve the problem of the color distortion of aerial observation images at night.Third,for the existing image sharpening methods,the algorithm verification of the computer platform is mainly used.This paper also studies the hardware implementation of the entire algorithm.The hardware verification of the improved color transfer algorithm is realized on the DSP.Finally,some contrast experiments was carried out based on the proposed algorithm,the experimental results show that this algorithm is superior to the reference algorithm both in terms of visual perception and objective evaluation,especially for the performance of clear observation of night-to-earth aerial observations.The study can provide some help for the rapid clearing of low-altitude fear-seeing images in the night.