Abstract
Spectral signature coding is an effective means to characterize spectral features. This paper presents a novel progressive spectral signature coding, called Multistage Predictive Coding Modulation (MPCM) for hyperspectral imagery. It not only can be used for signature coding, but also can provide a progressive profile for each signature in terms of identification. The idea is to consider the spectrum of a hyperspectral image pixel vector as a 1-D signal, then the proposed MPCM encodes such a 1-D spectral signal progressively based on the priorities assigned to signal points for spectral reconstruction. The MPCM generates priority codewords for a spectral signature in a specific manner that the characteristics of this particular signature can be captured and used as its own fingerprint. Despite that the success of MPCM in spatial coding has been demonstrated in image progressive reconstruction, progressive edge detection and text detection, its utility in spectral coding has not been investigated and is yet to be explored. This paper demonstrates that the MPCM-based progressive spectral coding can be useful not only in data compression, progressive reconstruction, but also in progressive material identification.
Original language | English |
---|---|
Pages (from-to) | 252-261 |
Number of pages | 10 |
Journal | Proceedings of SPIE - The International Society for Optical Engineering |
Volume | 5268 |
DOIs | |
Publication status | Published - 2004 |
Event | Chemical and Biological Standoff Detection - Providence, RI., United States Duration: 2003 Oct 28 → 2003 Oct 30 |
All Science Journal Classification (ASJC) codes
- Electronic, Optical and Magnetic Materials
- Condensed Matter Physics
- Computer Science Applications
- Applied Mathematics
- Electrical and Electronic Engineering