Hyperspectral Image Compression Using Three-Dimensional Wavelet Coding: A Lossy-to-Lossless Solution†

We propose an embedded, block-based, image wavelet transform coding algorithm of low complexity. The embedded coding of Set Partitioned Embedded bloCK (SPECK) algorithm is modified and extended to three dimensions. The resultant algorithm, three-Dimensional Set Partitioned Embedded bloCK (3D-SPECK), efficiently encodes 3D volumetric image data by exploiting the dependencies in all dimensions. 3D-SPECK generates embedded bit stream and therefore provides progressive transmission. We describe the use of this coding algorithm in two implementations, including integer wavelet transform as well as floating point wavelet transform, where the former one enables lossy and lossless decompression from the same bit stream, and the latter one achieves better performance in lossy compression. Wavelet packet structure and coefficient scaling are used to make the integer filter transform approximately unitary. The structure of hyperspectral images reveals spectral responses that would seem ideal candidate for compression by 3D-SPECK. We demonstrate that 3D-SPECK, a wavelet domain compression algorithm, can preserve spectral profiles well. Compared with the lossless version of the benchmark JPEG2000 (multi-component), the 3D-SPECK lossless algorithm produces average of 3.0% decrease in compressed file size for Airborne Visible Infrared Imaging Spectrometer images, the typical hyperspectral imagery. We also conduct comparisons of the lossy implementation with other the state-of-the-art algorithms such as three-Dimensional Set Partitioning In Hierarchical Trees (3D-SPIHT) and JPEG2000. We conclude that this algorithm, in addition to being very flexible, retains all the desirable features of these algorithms and is highly competitive to 3D-SPIHT and better than JPEG2000 in compression efficiency.