Research On Cloud Removal Processing Of Gaofen Series 2 Satellite Remote Sensing Image

With the continuous development of remote sensing technology,the application of remote sensing images is becoming more and more extensive,and the requirements for image quality are becoming higher and higher,but there are many external factors that affect the data quality during actual shooting,of which atmospheric cloud and fog have the largest impact.Cloud noise not only affects the interpretation accuracy of remote sensing images and the accuracy of the interpretation of target features,but also increases the difficulty of obtaining effective remote sensing image data and reduces the timeliness.Therefore,if effective methods can be used to reduce or remove the impact of atmospheric clouds on the quality of remote sensing images,it is of great significance to improve the effective use of remote sensing data.The main content of the thesis is:(1)Summarize and analyze the causes of cloud formation and the characteristics of cloud in remote sensing image,and on this basis elaborate the model of thin cloud imaging in remote sensing image and the principle of cloud imaging.(2)Taking Gaofen2 remote sensing image as the research object,using homomorphic filtering,wavelet transform method,and Laplace transform method to conduct suitability experiments and obtain experimental results,comparing the cloud removal effects of the three methods,homomorphic filtering to remove After the cloud image,some detailed information is lost and the edges are blurred;after the Laplacian method removes the cloud,the clouded area transitions to the cloudless area blunt;although the wavelet transform method is better in comprehensive effect,some areas also appear in the resulting image become blurred.(3)A cloud removal algorithm using a combination of two methods is proposed: wavelet transform-low frequency homomorphic filtering method.For the cloud layer in the low frequency part,the low frequency part is homomorphic filtered after wavelet decomposition,and then the low frequency coefficient is reduced to increase High-frequency coefficients are reconstructed to obtain the resulting image;wavelet transform-homomorphic filtering method,considering that the clouds are not only concentrated in the low-frequency part,but also have a small amount of clouds in the high-frequency part.After wavelet transform reconstruction,the whole is homomorphic filtered to obtain the result image.Compared with the cloud removal effect,the wavelet transform-low frequency homomorphic filtering method is not ideal for the Gaofen2 remote sensing image,and the high-frequency part of the cloud information has not been removed;the wavelet transform-homomorphic filtering method has the best cloud removal effect,although the information entropy The value increase is not the largest but the closest to the original image value under the premise of growth,indicating that the amount of information has been improved under the premise of ensuring the original information...