Hyperspectral Data Compression provides a survey of recent results in the field of compression of remote sensed 3D data, with a particular interest in hyperspectral imagery. Chapter 1 addresses compression architecture, and reviews and compares compression methods. Chapters 2 through 4 focus on lossless compression (where the decompressed image must be bit for bit identical to the original). Chapter 5, contributed by the editors, describes a lossless algorithm based on vector quantization with extensions to near lossless and possibly lossy compression for efficient browning and pure pixel classification. Chapter 6 deals with near lossless compression while. Chapter 7 considers lossy techniques constrained by almost perfect classification. Chapters 8 through 12 address lossy compression of hyperspectral imagery, where there is a tradeoff between compression achieved and the quality of the decompressed image. Chapter 13 examines artifacts that can arise from lossy compression.
The interest in remote sensing applications and platforms (including airborne and spaceborne) has grown dramatically in recent years. Remote sensing technology has shifted from panchromatic data (a wide range of wavelengths merged into a single response), through multispectral (a few possibly overlapping bands in the visible and infrared range with spectral width of 100-200«m each), to hyperspectral imagers and ultraspectral sounders, with hundreds or thousands of bands. In addition, the availability of airborne and spaceborne sensors has increased considerably, followed by the widespread availability of remote sensed data in different research environments, including defense, academic, and commercial.
Remote sensed data present special challenges in the acquisition, transmission, analysis, and storage process. Perhaps most significant is the information extraction process. In most cases accurate analysis depends on high quality data, which comes with a price tag: increased data volume.