Research Of Remote Sensing Change Detection Based On Fusion Of Difference Images For SAR Image

The change detection of remote sensing images refers to the qualitative or quantitative analysis of the characteristics and processes of the change by monitoring the change information of the same scene on the surface of the earth and the image of the different phase.In recent decades,with the rapid development of remote sensing technology,the change detection of remote sensing images have been applied to many fields,and it has become the key technology for people to update the geographic database.Synthetic aperture radar(SAR)has the advantages of penetrating cloud and fog and surface vegetation,not affected by bad weather conditions such as rain and snow,and collecting information all day and all weather.How to quickly and accurately obtain the change information of ground observation through SAR remote sensing technology is very important for the application and development of SAR remote sensing technology,and it is also one of the hot issues in the current research.This paper mainly studies the change detection technology of SAR remote sensing images.The research contents are summarized as follows:(1)An unsupervised SAR remote sensing image change detection algorithm based on difference image fusion and fuzzy C means clustering is proposed.Firstly,the Frost filter based on the logarithmic domain is carried out on the original images.Two difference images are obtained by using the difference method and the mean-ratio method.In order to make full use of the information of two difference images,the two difference images are fused to get a final difference image.Finally,we use fuzzy C mean algorithm to get change classes and non-change classes.(2)An adaptive SAR threshold denoising algorithm based on difference image fusion and nonsubsampled shearlet transform(NSST)is proposed for unsupervised remote sensing image change detection.First,the logarithmic domain denoising is applied to the original image using mean filter,and then two difference images are generated by log-ratio method and mean ratio method.In order to get the difference image with better change information,we fuse two difference images.However,the obtained difference image still contains a lot of noise.In order to reduce its impact on the detection results,the adaptive threshold denoising based on for NSST is carried out.Finally,K-means clustering is used to cluster the difference image to get the detection results.Experiments on the proposed algorithm and the reference algorithm in this thesis verify that the proposed algorithm has higher detection accuracy and detection efficiency.