Remote Sensing Image Cloud Detection And Thin Cloud Removal

Satellite remote sensing technology plays an important role in the field of earth resources,agriculture and forestry and natural disaster monitoring. In the process of satellite remote sensing images,cloud have seriously affected the application of remote sensing image. Therefore, the cloud detection and removal for the use of remote sensing image is significant. This paper using the Landsat data to research and analyze cloud detection and cloud removal.This paper first introduces the significance of cloud detection. Summarizing and analyzing the various algorithms about cloud detection in the world.Satellite remote sensing image of Landsat are analyzed and explained. The pretreatment process of remote sensing data, and a variety of features and cloud spectral characteristics are discussed, algorithm and OTSU cloud detection algorithm are introduced.Artificial neural network is the network structure of a simulation of the brain neurons. It is widely used in pattern recognition, function approximation and other fields.It has the very strong learning function.Because the accuracy of some kind of cloud is very low of the ACCA algorithm. Neural network, which has strong learning ability of artificial, combined with the use of ACCA for cloud detection. This paper uses 3 kinds of artificial neural network, BP neural network, RBF neural network and probabilistic neural network. ACCA was combined with the three kinds of artificial neural network, and compared with the ACCA algorithm.As results,ACCA combined with the artificial neural network is better than ACCA.About the removal of thin cloud, this paper studies about the wavelet transform. The Mallat algorithm and 2-D Mallat algorithm are analyzed in detail. Also lifting wavelet are discussed. According to the principle of lifting wavelet transform can quickly transform, wavelet transform is uesd in the thin cloud removal field. Through the lifting wavelet transform and the traditional wavelet transform experiment, founding that the speed of lifting wavelet transform for cloud removal compared with the traditional wavelet transform has greatly improved.