Regression Wavelet Analysis for Lossless Coding of Remote-Sensing Data

Abstract

A novel wavelet-based scheme to increase coefficient independence in hyperspectral images is introduced for lossless coding. The proposed Regression Wavelet Analysis (RWA) uses multivariate regression to exploit the relationships among wavelet-transformed components. It builds on our previous nonlinear schemes that estimate each coefficient from neighbor coefficients. Specifically, RWA performs a pyramidal estimation in the wavelet domain thus reducing the statistical relations in the residuals and the energy of the representation compared to existing wavelet-based schemes. We propose three regression models to address the issues concerning estimation accuracy, component scalability and computational complexity. Other suitable regression models could be devised for other goals. RWA is invertible, it allows a reversible integer implementation, and it does not expand the dynamic range. Experimental results over a wide range of sensors, such as AVIRIS, Hyperion and IASI, suggest that RWA outperforms not only Principal Component Analysis (PCA) and Wavelets, but also the best and most recent coding standard in remote sensing, CCSDS-123. Index TermsTransform coding via regression, wavelet-based transform coding, remote sensing data compression, redundancy in hyperspectral images.

Clustering Regression Wavelet Analysis for Lossless Compression of Hyperspectral Imagery

“…To further improve the performance of DWT based HSI coding, Regression Wavelet Analysis (RWA) [6] was proposed to reduce remaining redundancies between wavelet coefficients after spectral wavelet transformation. This is achieved by performing a linear regression at each spectral DWT scale to generate a model which allows estimates of detail components to be generated using approximation components.…”

Diffusion-Based Inpainting for Coding Remote-Sensing Data

“…As said, the idea of PDE-based image compression relies on storing a small fraction (a density d) of image pixels, has been decorrelated losslessly by the RWA transform [5], and compressed by the PAQ software [16] . Figure 4 provides the rate-distortion performance for the transform-based methods applied to the Hawaii uncalibrated image.…”

“…Other recently proposed works combine transform-and predictive-based methods using regression analysis to tackle the dependencies that still remain among the data in the transform domain. These techniques provide promising results for lossless and progressive lossy-to-lossless coding [5], [6], [7], [8].…”

Hyperspectral IASI L1C Data Compression

“…For RWA, two different estimation models [38] could be adopted: Maximum model and Restricted model . Here, we use a variant of the Maximum model, the Exogenous variant , which considerably reduces the computational cost and does not entail transmission of any side-information.…”