Research Of Wavelet Analysis With Its Application For Remote Sensing Images

The transit, analysis and processing of Remote sensing image is important innational economics construction and military application. By introducing waveletanalysis and statistical analysis in modern mathematical analysis theory and method,this paper analyses the statistical characteristic of the color, texture and the waveletcoefficients of remote sensing image, compares remote sensing data compressionalgorithms, and gets some results as follows:(1) The characteristic of remote sensing data is analyzed in the terms of theircontents, structure and data management. The characteristic of color and texture isdepicted based on statistical method. The depiction of this characteristic is thefundamental of constructing model of remote sensing image and connecting the basiccharacteristic and advanced semantics.(2) By introducing wavelet analysis and performing wavelet transform on remotesensing image by fast analysis and reconstruction algorithm, using statisticalparameter estimation and hypothesis testing, we find the statistical distribution of thewavelet coefficients of remote sensing images, and plot the statistical distributionfigure for experimental images. Further more, we find that the high frequency waveletcoefficients of remote sensing image have the same non-Gaussian distribution asnormal images. However, the high frequency wavelet coefficients of remote sensingimage contain the coefficients which have more significant magnitude and higherpower.(3) We compare several remote sensing image compression methods, especiallythe main coding methods--the prediction method, DCT method and WT method. Bycomparing the computational complexity, memory requirement and computationaltime, we propose a compression algorithm based on biorthogonal lapped transform,which is of low complexity, low memory requirement and parallel computation, andeasy to implement the remote sensing image compression on satellite.