Study Of Denoising Method Of Remote Sensing Image Based On Dual Tree Complex Wavelet And Fourth-Order Partial Differential Equation

With the advent of the large data age of satellite industry in China,the research work related to remote sensing application is developing vigorously.The quality of remote sensing image as the source of data has always been the focus of people's attention.Because remote sensing images are inevitably disturbed by a large number of noises in remote sensing imaging,image transmission,storage,decoding and other links,noise removal has always been a hot issue in the field of image processing.It is of great significance in the application of remote sensing image segmentation,classification,and change detection to study how to remove the noise in remote sensing images,so as to improve the readability,discrimination and purity of the images.It is proved that the image gray level tends to be piecewise constant in the low order partial differential equation denosing model.Considering the applications of wavelet multi-scale decomposition in image processing,a kind of remote sensing image denoising model combining dual tree complex wavelet transform and fouth order partial differential equation is put forward.Firstly,the noise images were multiscalely decomposted by DTCWT,then the low frequency components were reserved and the noise in high frequency components of other layer were removed by fourth order PDE model.Finally,the high and low frequency components of the corresponding layerwere reconstructed in order to get the final image.According to the denoising experiments results of the ZY-3 satellite images which have different noise intensities,it is proved that the average PSNR of results by the proposed model increases 1-2dB and structural similarityincreases as well.The denoising model proposed can effectively preserve the image details while removing noise.