CNES studies of on-board compression for multispectral and hyperspectral images

Abstract

Future high resolution instruments planned by CNES for space remote sensing missions will lead to higher bit rates because of the increase in resolution, dynamic range and number of spectral channels for multispectral (up to 16 bands) and hyperspectral (hundreds of bands) imagery. Lossy data compression is then needed, with compression ratio goals always higher and with low-complexity algorithm. For optimum compression performance of such data, algorithms must exploit both spectral and spatial correlation. In the case of multispectral images, CNES (in cooperation with Thales Alenia Space, hereafter TAS) studies have led to an algorithm using a fixed transform to decorrelate the spectral bands, the CCSDS codec compresses each decorrelated band using a suitable multispectral rate allocation procedure. This lowcomplexity decorrelator is adapted to hardware implementation on-board satellite and is under development. In the case of hyperspectral images, CNES (in cooperation with TAS/TeSA/ONERA) studies have led to a full wavelet compression system followed by zerotree coding methods adapted to this decomposition. We are investigating other preprocessors such as Independent Component Analysis which could be used in both approaches. CNES also participates to the new CCSDS Multispectral and Hyperspectral Data Compression Working Group.

Regression Wavelet Analysis for Lossless Coding of Remote-Sensing Data

“…To find the estimate W j of the details components W j based on the approximation components to the Exogenous model proposed for the PCA [56] or for the orthogonal optimal spectral transform (OST) [18], [19].…”

Lossy and lossless compression of MERIS hyperspectral images with exogenous quasi-optimal spectral transforms

“…However, the main drawback of the data-dependent KLT (or PCA) is its heavy computational cost (due to the calculation of a covariance matrix across the data), therefore low-complexity computations of the covariance matrix have been proposed to reduce the computational burden [4,7]. To reduce the complexity of KLT based codecs, another strategy has been successfully applied in [6] for on-board compression of multispectral images. This strategy is to calculate first the KLT on a set of images (called the learning basis) from one (and only one) spectrometer in order to obtain an efficient, although sub-optimal, spectral transform, called exogenous KLT, that is then applied to any image from the same sensor.…”

“…They used 1-D DWT for reducing spectral redundancies, however the result still holds with a linear transform. Moreover, when the spectral decorrelation is achieved via the Karhunen-Loève Transform (KLT) or a Principal Component Analysis (PCA), the performances are significantly improved at low, medium and high bit-rates [3,4], even when the DWT adapts to the encoded image [5,6]. However, the main drawback of the data-dependent KLT (or PCA) is its heavy computational cost (due to the calculation of a covariance matrix across the data), therefore low-complexity computations of the covariance matrix have been proposed to reduce the computational burden [4,7].…”

“…Moreover, codecs based on these methods have generally a lower complexity than codecs using vectorial quantization before entropy coding. Thus, research activities on multicomponent image codecs based on transform and/or subband coding have been strengthened these last years [1][2][3][4][5][6][7][8][9][10][11]. The JPEG2000 (JP2K) Part 2 standard implements such a method with two variants for reducing spectral redundancies: by applying either a 1-D linear transform, or a 1-D discrete wavelet transform (DWT) between the components of the encoded image.…”

Adaptive multispectral GPU accelerated architecture for Earth Observation satellites

“…Image compression is achieved by exploiting correlations in the data to achieve a more efficient representation of the image information. The first known satellite to implement onboard image compression was SPOT-1, launched in 1986 [1]. Since then, many satellites have featured onboard compression capabilities [2].…”