Study On Extraction Of Saline Pasture Information Based On Cloud Removing Remote Sensing Image

The pasture salinization is the land degeneration which is caused by the unreasonable human activity and the frail ecological environment. The pasture salinization is comprehensively concerned by more and more people in the whole world because it brings the destructive effect to the social economy, environment and the ecosystem. It is very meaningful to govern saline pasture, achieve the sustainable development of husbandry using remote sensing technology, which is characteristic by its macroscopic, comprehensive, dynamic and fast quality, to study the quality, the scope, the area, the geographic distribution, the dynamic change and the degree of saline soils. It reduces efficiency in the use of remote sensing data, affecting the thematic information extraction. Therefore, the remote sensing image cloud remote has been an important research filed in the remote sensing image preprocessing.The purpose of the remote sensing image thick cloud remove is to revert the information of the obscured areas and ensure that the information of other areas is not distortion, the key point is to locate the cloud areas. This article uses a new method bases on Mean Shift and Wavelet to automatic locate the cloud areas of the TM remote sensing image after the studying on the cloud remove progress and common methods. First, analysis the spectral characteristics of the TM image and select band 5 which is sensitive of cloud to locate cloud areas. Then process wavelet transform after the Mean Shift cut and computing the wavelet energy of the pixels in the cloud areas. Finally, obtain the wavelet energy experience threshold of the cloud pixels through training. After locate the cloud areas automatically, using the pixel replacement method to achieve the multitemporal TM image cloud remove. The experiment shows this method can remove the thick clouds effectively.This paper use color composite to stress the salinization information during the study of the salinization information extract. Based on the study of the common method, process a improved method of color composite during the analysis form the perspective side. Choose bright, wetness and TM4 process the RGB color composite. This method stress the salinization information compare with the bands color composite. Reduce the mix of the slightly salinization with the non-salinization compare to the traditional K-T transform. The experiment shows the accuracy increased by 7%.This article starts from different research methods, using the advantages of TM image to improve the accuracy of the thematic information extraction based on the thick cloud remove then evaluate the pasture salinization. Also thinking of the related fields during the study. On the cloud remove field, how to remove the thin cloud and fog without affect other information is the key point. On the salinization information extraction field, automatically interpret the remote sensing data using computer and improve the accuracy of thematic information extraction has a broad space for development.