Wavelet-curvelet-contourlet based remote sensing data mining model

Abstract

: Remote sensing applications such as change detection, multispectral classification, environment monitoring, image mosaicking, weather forecasting, super resolution images and integrating information into geographic information system (GIS), image registration is a required process. Such natural images contain intrinsic geometrical structures that form the key features in visual information. Satellite data thus delivered/received in the form signals/images have a wide coverage with multi-temporal and multispectral capabilities. In such problems, a prime objective is to improve the quality of transmitted signals/images composed of desired signal plus additive random/Gaussian noise, by employing efficient feature extraction and denoising techniques with efficient representation of visual information. The experimental results and performance factor analysis based on of each of the multiresolution transforms show that contourlet transform produces relatively better result in terms of capturing directional information, reconstruction, noise restraints. The modelling and simulation: The feature extraction and denoising process is aimed at removing the noise with the help of a matched filter (either using wavelet, curvelet or contourlet), and is composed of three major steps viz. Decomposition of the transmitted signal, Thresholding to demise noisy elements, and Reconstruction of the processed signal. Signal is represented as