Remote Sensing Image Change Detection Algorithm Based On Sparse Representation

In recent years,the development of high-resolution synthetic aperture radar has increased.The use of remote sensing images has become more widespread in military and people's daily lives.It is mainly used in the military to evaluate military strikes,and it is mainly used in people's daily lives.As for the location of natural disasters and the changes in urban areas,the accuracy of most of the change detection methods cannot fully meet the increasing demand for remote sensing imagery in response to various changes in geomorphological landscapes before and after geological disasters,and need to deal with large amounts of data.This paper focuses on how to use sparse representation to reduce the amount of data needed to detect changes in remote sensing images of geomorphological changes in different periods before and after geological disasters and to improve their detection accuracy.The following two tasks are completed:(1)The research on the change detection method of remote sensing image using double sparse representation and its effectiveness demonstration.This method first needs to perform some related preparations,in order to prevent the image from being interfered by the speckle noise,to process the image in advance.Then,the two time-phase pictures are double-sparsely represented and sparse twice using different two dictionaries to obtain a reconstructed image containing only useful information,which is used as an input in the following change detection process.Next,the Mean-Shift method is used toextract the features of the images,and the difference image is constructed by using the regression method.Finally,the obtained difference image is divided by the threshold value to obtain the final change detection result image.The validity of this algorithm is verified by comparison with other literature algorithms.(2)This paper studies the change detection method of double sparse representation remote sensing image using difference image fusion and demonstrates its effectiveness.This method first needs to perform some related preparations,in order to prevent the image from being interfered by the speckle noise,to process the image in advance.Then,the difference maps are constructed using the difference method and the ratio method for the two remote sensing images.Then,the two reconstructed difference images are fused into a single image by the stationary wavelet fusion method.Then,using the double sparse representation method proposed in the first work to process the difference image to extract the useful information in the image,reduce the number of pixels to be processed,and use the reconstructed image as the input value of the algorithm in the next step.Next,use the Mean-Shift method to segment the image and extract features.Finally,the obtained difference image is divided by the threshold value to obtain the final change detection result image.Experiments show that the stationary wavelet fusion can make up for the problem that the double sparse representation can not extract the edge feature of the image,and the combination of the two makes the detection accuracy improved.