Reduction of Signal-Dependent Noise From Hyperspectral Images for Target Detection

Hierarchical Sparse Bayesian Learning with Beta Process Priors for Hyperspectral Imagery Restoration

“…To investigate the performance of the proposed method, we choose four different methods for comparison, including K-SVD [6], BM3D [7], ANLM3D [11] and BM4D [12]. The necessary parameters in the four compared methods are finely tuned or automatically selected to generate the optimal simulated results.…”

“…However, the proper kinds of wavelet transform are hard to choose. Considered the HSI as a three-order tensor, tensor analysis methods are proposed for noise reduction in [7] and [8], with the spectral and spatial information completely preserved. Besides, nonlocal approaches can exploit spatial information on the whole image directly.…”

A Bayesian error model for quantifying methane and carbon dioxide column densities from hyperspectral measurements

“…Measurement noise in IFTS data comes from multiple sources, including sensor-related uncertainties, e.g. photon shot noise, dark current noise, quantization noises, thermal noise, detector read-out noise [27,28], as well as interferometer [29], optics, and calibration-related uncertainties [30,31,19].…”

“…Radiometric noise sources may be classified depending on whether they observe a fully random or a fixed-pattern behavior [31]. For example, photon shot noise and thermal noise are random noises, whereas optics and calibration errors are fixed-pattern noises.…”

“…Indeed, SD noise often obeys a Poisson distribution, while SI noise is normally-distributed white noise; these noise sources are often combined into a Poisson-Gaussian noise model [28,32]. In the literature, this model has been used for noise reduction for hyperspectral images [30,31,32], parametric sensitivity analysis [19,28], and inversion procedures for identification or quantification of scene properties (e.g. gases and material properties) [14,21,33].…”