Translator Disclaimer
27 August 2008 Block truncation signature coding for hyperspectral analysis
Author Affiliations +
This paper introduces a new signature coding which is designed based on the well-known Block Truncation Coding (BTC). It comprises of bit-maps of the signature blocks generated by different threshold criteria. Two new BTC-based algorithms are developed for signature coding, to be called Block Truncation Signature Coding (BTSC) and 2-level BTSC (2BTSC). In order to compare the developed BTC based algorithms with current binary signature coding schemes such as Spectral Program Analysis Manager (SPAM) developed by Mazer et al. and Spectral Feature-based Binary Coding (SFBC) by Qian et al., three different thresholding functions, local block mean, local block gradient, local block correlation are derived to improve the BTSC performance where the combined bit-maps generated by these thresholds can provide better spectral signature characterization. Experimental results reveal that the new BTC-based signature coding performs more effectively in characterizing spectral variations than currently available binary signature coding methods.
© (2008) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Sumit Chakravarty and Chein-I Chang "Block truncation signature coding for hyperspectral analysis", Proc. SPIE 7086, Imaging Spectrometry XIII, 70860D (27 August 2008);

Back to Top