Remote Sensing Image De-noising Algorithm Based On Wavelet Transform

Remote sensing images are obtained from sensors on the remote sensing platform. Different sensors have different spectral and spatial resolution, so it causes different remote sensing images. They contain a lot of information, which present the reflection spectrum reflected from the objects on the Earth’s surface, and contain abundant spatial structure information as well. However, they are often interfered by a variety of noise during the imaging and transmission processes, which causes a great deal of impact on the information process and storage. Looking for a way to reduce noise and retain the image edge information is always the research focus.Wavelet transform is an effective method for image processing. It has the characteristic of time-frequency localization analysis, and be able to detect the edge feature of local mutation. It has the multi-scale detailed analysis of remote sensing image by scaling, translation and other computing functions. So it can effectively extract information from images, then the structure and texture of images can be manifested in different resolution levels. Therefore, in recent years, wavelet transform has been widely used in remote sensing image de-noising. People have put forward a number of different wavelet de-noising methods. Compared with the traditional methods, it proposes some improved de-noising algorithms on the basis of wavelet transform, and studies them in depth in this paper. The main research contents and results are as follows:(1) It puts forward an improved algorithm based on wavelet transform and semi-soft threshold. First, it analyzes the advantages and disadvantages of the traditional threshold functions. In view of the shortcomings of the hard and soft threshold functions, a semi-soft threshold function is established. Second, it gives the thresholds of different scales and directions on the base of the wavelet transform. Then the semi-soft threshold is carried out on the image processing, and it uses mean filter to the high frequency coefficients of a layer of image reconstruction. Finally, it performs the inverse wavelet transform. Experimental results show that, the proposed algorithm can obtain better visual effect and higher peak signal to noise ratio on the remote sensing image de-noising.(2) It proposes a remote sensing image de-noising algorithm based on wavelet transform and2D-PCA.Through the2D-PCA method, it can deal with the matrix directly and calculate the eigenvalues and eigenvectors of the matrix. Then it selects the appropriate threshold to remove the small eigenvalues, and performs the normalization. Finally it calculates the projection matrix after dimensionality reduction. The2D-PCA method can reduce the characteristic vector linear correlation and the amount of computation, which increases the computing speed significantly. The experimental results show that, the algorithm can effectively remove Gaussian noise of remote sensing images and keep the edge details better.(3) To verify the effectiveness of the algorithms in the above, it establishes a remote sensing image de-noising detection system. Concrete design and the analysis of the system are given. It also describes each function module and using process briefly. With the system, users can get different de-noising images according to their choice and compare the advantages and disadvantages between the different algorithms.