The Study On Atomization Removal Method For Aerial Image Of UAV

UAV remote sensing technology,as a new important means of spatial data acquisition,has developed rapidly in recent years.Compared with traditional satellite remote sensing and manned aerial remote sensing,UAV remote sensing technology has higher resolution and high efficiency.However,the image acquired by uav remote sensing in fog will be interfered by the scattering of airborne particles,resulting in the reduction of the final image clarity and contrast,color distortion,and great difficulties in the subsequent application of image target recognition and tracking.Therefore,it is of great practical significance to study the image atomization removal technology of UAV.The research work of this paper is as follows:First of all,this paper studies two kinds of defog models of UAV aerial images.In the study of image enhancement method based on non model defogging,it is concluded that this method mainly improves the color and contrast of the image to achieve defogging,although the algorithm is simple and easy to achieve,but the effect of defogging is not good.The model-based image restoration method uses the prior theory such as dark channel to get the transmission image and global atmospheric light,and then combines the imaging model of fog image to restore the fog free image.The model-based image restoration method has advantages over the image enhancement method in time complexity and effect.Therefore,this paper studies and compares the advantages and disadvantages of two kinds of defogging methods,and finally selects the prior defogging method based on the model of image restoration.Secondly,when using the fog image degradation model to restore the fog free image,it is necessary to optimize the rough atmospheric transmission image obtained by the dark channel prior theory.However,there is still a problem of "halo" in the edge position of non fog image.To solve this problem,this paper proposes an adaptive weight guided filtering method.Image of the local characteristics of the structure tensor main eigenvalues can be accurate description of the structure characteristics of the image and the edge,therefore,this article will not main characteristic value of local characteristics of the structure tensor as weights to join the guided filter,get structure characteristic and the edge more accurately guided imagery to optimize transmission diagram,again without through atmospheric scattering model to recuperate the fog image,can solve the problem of no fog "halo" of the image,and can improve image to fog fog effect and after the image quality.Finally,the use of drones to obtain foggy images is designed.The methods proposed in this paper are compared with traditional guided filtering,bilateral filtering and other anti-fogging algorithms for subjective and objective comparison and analysis experiments.The experiment of defogging effect with different concentration and the experiment of building recognition in defogging image are designed.The experimental results show that this method can effectively suppress the "halo" phenomenon at the edge of the fog free image while maintaining the same performance,and the restored fog free image is closer to the original fog image in terms of color and contrast.