Effects of Lossy Compression of Hyperspectral Imagery

摘要

Hyperspectral imaging (HSI) sensors provide imagery with hundreds of spectral bands, typically covering VNIR and/or SWIR wavelengths. This high spectral resolution offers promise for many applications, but it also produces enormous volumes of data, which may be problematic for storage and transmission. Lossy compression may therefore be necessary, but application performance degradation that results from compression is of concern. This report documents results for a spectral-spatial lossy compression scheme and a variety of applications: normalized difference vegetation index (NDVI), integrated column water vapor (CWV), and background classification. The compression scheme first performs principal-components analysis spectrally, then discards many of the lower-importance principal-component (PC) images, and then applies JPEG2OOO spatial compression to each of the individual retained PC images. Two different rate-allocation methods, which select the spatial compression ratios, are considered. The assessment of compression effects considers general-purpose distortion measures, such as root-mean-square difference. It also examines changes in NDVI and CWV data products and proposes statistical tests for deciding whether compression causes significant degradations in classification results.