Volumetric image compression with JPEG2000

Abstract

A 3D extension of the standard will support cuboid volumetric data sets with no time component. Medical and scientific imaging equipment generate a significant amount of volumetric data. Because such data sets tend to be very large, powerful compression technology is crucial for efficient storage and transmission, random access, region of interest (ROI) support, and resolution/quality scalability. JPEG2000, the current standard devised by the Joint Photographic Expert Group (JPEG), provides this functionality for 2D data sets through Part 1 (Core Coding Systems) 1 and Part 2 (Extensions). 2 Although the latter indirectly supports a 3D wavelet transform by means of the extended multiple component transform, a complete volumetric coding scheme is still essential to functionality, optimal rate-distortion performance, and isotropic behavior for all dimensions.

Hyperspectral Image Compression Using a Two-Stage KLT

“…It has particularly useful applications in such fields as medical imaging, where storage of large volumetric data sets is necessary. The JP3D specifications essentially do not introduce new coding concepts but simply extend those of JPEG2000 parts 1 & 2 (embedded block coding by optimized truncation (EBCOT) based on the principles of layered zero coding (LZC)) to a 3 rd dimension [11]. The JP3D codec first partitions the volumetric data set (typically a stack of 2D images) into tiles of volumetric blocks that are independently encoded as separate grayscale datasets.…”

“…The JP3D codec first partitions the volumetric data set (typically a stack of 2D images) into tiles of volumetric blocks that are independently encoded as separate grayscale datasets. All the tile sample coefficients are then filtered via a discrete wavelet transform, using the Mallat decomposition pattern [11]. From here, the codec partitions the resulting sub-bands into dyadically-sized code-blocks, independently encoded by the EBCOT coder ( Figure 6-1).…”

“…The code-block coefficients are processed four adjacent bits at a time within a bit plane, before moving down to the next layer ( Figure 6-2). The bit plane pattern is identical to that used in JPEG2000 part 1 [11]. The JP3D compression scheme could also be applied to spectral images used in remote sensing, where the 3 rd axis could simply be replaced by the spectral information of a 2D scene.…”

“…The algorithm does have a conceivable application advantages over the 2-stage KLT. Because JP3D uses a wavelet transform approach in all three dimensions, as opposed to a linear transform, such as the KLT, it is more suitable for transmitting data a few lines at a time [11]. Our approach does not have this kind of scan-based-mode flexibility over the spectral dimension.…”

“…JP3D's process of partitioning, subpartitioning, and then filtering at the tile sample coefficient level, can be very involved [11]. If spectral linearity can be assumed, computation can be greatly, and sufficiently, simplified.…”