Translator Disclaimer
Paper
15 November 2010 A total variation denoising algorithm for hyperspectral data
Author Affiliations +
Abstract
Since noise can undermine the effectiveness of information extracted from hyperspectral imagery, noise reduction is a prerequisite for many classification-based applications of hyperspectral imagery. In this paper, an effective three dimensional total variation denoising algorithm for hyperspectral imagery is introduced. First, a three dimensional objective function of total variation denoising model is derived from the classical two dimensional TV algorithms. For the consideration of the fact that the noise of hyperspectral imagery shows different characteristics in spatial and spectral domain, the objective function is further improved by utilizing two terms (spatial term and spectral term) and separate regularization parameters respectively which can adjust the trade-off between the two terms. Then, the improved objective function is discretized by approximating gradients with local differences, optimized by a quadratic convex function and finally solved by a majorization-minimization based iteration algorithm. The performance of the new algorithm is experimented on a set of Hyperion imageries acquired in a desert-dominated area in 2007. Experimental results show that, properly choosing the values of parameters, the new approach removes the indention and restores the spectral absorption peaks more effectively while having a similar improvement of signal-to-noise-ratio as minimum noise fraction (MNF) method.
© (2010) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Ting Li, Xiao-mei Chen, Bo Xue, Qian-qian Li, and Guo-qiang Ni "A total variation denoising algorithm for hyperspectral data", Proc. SPIE 7854, Infrared, Millimeter Wave, and Terahertz Technologies, 785432 (15 November 2010); https://doi.org/10.1117/12.869982
PROCEEDINGS
8 PAGES


SHARE
Advertisement
Advertisement
RELATED CONTENT


Back to Top