Research On Remote Sensing Image Cloud Detection Algorithm Based On Dark Channel

Through the satellite remote sensing technology,people can observe the ground in real-time,and access to remote sensing image data.Analysis of image data and extracted from the information contained in,it can be widely used in various fields,such as the Earth resource exploration,environmental climate change prediction,and even military assistance to develop strategic plans.However,the existence of cloud in the air is a big obstacle,and the cover of clouds will seriously affect the utilization of remote sensing image data.Therefore,remote sensing image cloud detection research has important application value.In this paper,we propose two cloud detection algorithms based on dark channels processing.Through experiments,the accuracy of the two algorithms is 93% or more and the Kappa coefficient is over 0.9.The main research work and content of this paper are as follows:Firstly,this paper introduces the research purpose and significance of remote sensing image cloud detection.It is divided into three categories to describe the basic principles and algorithms of cloud detection both at home and abroad,and grasp the current cloud detection technology and research direction as a whole.Through the related research,we analyze the physical characteristics and different characteristics of the cloud area and the area,summed up the five points with the separability of the characteristics,and for the cloud detection algorithm design to improve the theoretical basis and basis.Aiming at the great advantage of dark channel a priori processing in image fogging,image detection and so on.In this paper,we propose a novel remote sensing image cloud detection algorithm based on dark channel processing.The use of dark channel processing can greatly reduce the brightness values of pixel points in the vast majority of images,while the pixel values of cloud pixels do not change much.Therefore,through the two in the brightness of the separability,the cloud area and the area can be detected separately.However,dark channel processing may bring about error to the edge of the cloud and reduce the detection effect.For this problem,this paper proposes a cloud detection algorithm based on adaptive shape dark channel processing.By adding the superpixel segmentation algorithm to solve the edge problem,that is,combining the superpixel segmentation results,adding the similarity of the pixels in the dark channel processing Degree judgment.Compared with the literature[52] and the K-means and fuzzy C-mean cloud detection algorithm,the two algorithms proposed in this paper have high detection effect and comprehensive performance.In summary,the two remote sensing image cloud detection methods studied in this paper have made some advantages in practical application.They can be used as the actual remote sensing image for selection and deletion,subsequent image transmission,classification and tracking.Theoretical basis or method of application.