Research On Adaptive Dehazing Algorithm For Remote Sensing Images Based On The Combination Of Bright And Dark Channels

Nowadays,remote sensing technology has made huge strides and has been applied in many fields.At the same time,the quality requirements of remote sensing image have become much more stringent.In fact,the quality of optical remote sensing image in fog will be seriously degraded by the turbidity factors(particles,dust,etc.)in the atmosphere,which will result in a significant decline in the degree of scene recognition,the distortion of image details and color shift.Thus,with the decrease of human sense quality,system functions such as target recognition,remote sensing navigation and land mapping are directly influenced.Consequently,it is of great importance to do research on remote sensing image defogging technology.This paper aims at solving the problem of quality degradation for remote sensing image in foggy conditions.Based on the study of the principle of image degradation in haze,some characteristics of fog images are deeply explored.The prior principle of the bright channel is proposed,the demisting image is obtained by combining the bright and dark channel principle.In addition,the experimental results prove that it has better performance in the proposed method than the classical dark channel defog method whether in defog efficiency or the original image structure information.Finally,OpenCL is introduced to realize and optimize the image in parallel.The main work of this paper is as follows:(1)First,the paper is to understand the causes of fog formation and the influence of fog on remote sensing imaging.Finally,a series of models in the atmospheric scattering model is described in this paper.It provides a good theoretical basis for the study of spatial remote sensing image defogging.(2)The paper analyzes the dark channel priori(DCP)algorithm,how to obtain the fogless image by DCP algorithm and some disadvantages of the algorithm in practical application.A priori theory of bright channel is also proposed.(3)The algorithm of this paper is proposed,the algorithm of dark channel is improved,and the problem of obtaining the primary color of the bright and dark channel after improvement is solved.Then,the atmospheric light is obtained through the combination of bright and dark channel,and the transmittance is obtained and refined through the guide filter.Afterwards,the fogless image is obtained,and the brightness of the image after demisting is adjusted.At last,the algorithm is compared with other algorithms based on image quality evaluation system,which proves that the algorithm proposed in this paper is better than other algorithms to some extent.(4)To solve the problem of the large amount of calculation and low speed of execution,the article will use the OpenCL technology to process the algorithm in parallel and perform simulation experiments on GPU and CPU platform.The degree of parallelism of each step in this algorithm is analyzed in detail,and the inappropriate steps are split,transformed and optimized.The parallel experiment of the algorithm is carried out by using OpenCL programming,which speeds up the operation and implementation of the algorithm and lays a solid foundation for the later engineering application.