Research On Cloud Detection And Removal Of Remote Sensing Image

One of the major limitations of remote-sensing imaging could be its sensitivity to weather conditions during the image acquisition process. The resulting images are frequently contaminated by clouds,which is affected the information quality of remote sensing to a large extent,and is not conducive to the analysis and interpretation of remote sensing image. Remote sensing has to cope with the so-called cloud removal problem which is an important difficulty affecting the observation of the Earth surface.The main research content is to eliminate cloud cover of single remote sensing images from the perspective of digital image processing,and is divided into the following aspects:(1) Gain the satellite remote sensing image and analysis the fuzzy mechanism.(2) Summarize and analysis the basic ideas of removing cloud cover.(3) Through the optimal threshold value which is selected by the smallest cross-entropy criteria to detect cloud area. The experimental results present that this method can detect more details of the cloud area boundary,and make the smallest difference between the cloud region and the underlying surface.(4) The morphology opening operation is used to exclude smaller bright pixels of land cover not connected to the cloud area, which are mistaken as the cloud pixels.The experimental results present that this method can distinguish the cloud area from the underlying surface clearly.(5) The concept of digital image inpainting technique is introduced to the image processing of removing cloud cover in remote sensing.(6) Based on the original Curvature-driven diffusion inpainting model(CCD model),a dynamic method of calculating (?) values is designed to the purpose of removeing the cloud area in remote sensing,and basically to meet the visual effect of human eyes,without obvious signs of inpainting. (7) Intuitive visual evaluation and objective evaluation of statistics two sides is adopted to analyze the algorithm of remove the cloud cover in remote sensing,and to evaluate the experimental results from the multi-perspective comprehensively.