21 December 1994 Image quality measures to assess hyperspectral compression techniques
Author Affiliations +
Abstract
The term 'multispectral' is used to describe imagery with anywhere from three to about 20 bands of data. The images acquired by Landsat and similar earth sensing satellites including the French Spot platform are typical examples of multispectral data sets. Applications range from crop observation and yield estimation, to forestry, to sensing of the environment. The wave bands typically range from the visible to thermal infrared and are fractions of a micron wide. They may or may not be contiguous. Thus each pixel will have several spectral intensities associated with it but detailed spectra are not obtained. The term 'hyperspectral' is typically used for spectral data encompassing hundreds of samples of a spectrum. Hyperspectral, electro-optical sensors typically operate in the visible and near infrared bands. Their characteristic property is the ability to resolve a large number (typically hundreds) of contiguous spectral bands, thus producing a detailed profile of the electromagnetic spectrum. Like multispectral sensors, recently developed hyperspectral sensors are often also imaging sensors, measuring spectral over a two dimensional spatial array of picture elements of pixels. The resulting data is thus inherently three dimensional - an array of samples in which two dimensions correspond to spatial position and the third to wavelength. The data sets, commonly referred to as image cubes or datacubes (although technically they are often rectangular solids), are very rich in information but quickly become unwieldy in size, generating formidable torrents of data. Both spaceborne and airborne hyperspectral cameras exist and are in use today. The data is unique in its ability to provide high spatial and spectral resolution simultaneously, and shows great promise in both military and civilian applications. A data analysis system has been built at TRW under a series of Internal Research and Development projects. This development has been prompted by the business opportunities, by the series of instruments built here and by the availability of data from other instruments. The products of the processing system has been used to process data produced by TRW sensors and other instruments. Figure 1 provides an overview of the TRW hyperspectral collection, data handling and exploitation capability. The Analysis and Exploitation functions deal with the digitized image cubes. The analysis system was designed to handle various types of data but the emphasis was on the data acquired by the TRW instruments.
© (1994) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Joan B. Lurie, Joan B. Lurie, Bruce W. Evans, Bruce W. Evans, Brian Ringer, Brian Ringer, Mathew Yeates, Mathew Yeates, } "Image quality measures to assess hyperspectral compression techniques", Proc. SPIE 2313, Microwave Instrumentation and Satellite Photogrammetry for Remote Sensing of the Earth, (21 December 1994); doi: 10.1117/12.197349; https://doi.org/10.1117/12.197349
PROCEEDINGS
13 PAGES


SHARE
RELATED CONTENT

Low-bit-rate subband image coding with matching pursuits
Proceedings of SPIE (January 09 1998)
Content-based image classification
Proceedings of SPIE (December 20 1999)
Approach to assessing hyperspectral data quality
Proceedings of SPIE (September 25 2001)

Back to Top