Abstract

Hyperspectral images (HSIs) are unavoidably polluted by various kinds of noise, which decrease the potential of subsequent processes in HSIs applications. Due to the diversity and complexity of HSIs mixed noise, including impulse noise, Gaussian noise, stripe noise and deadlines, traditional restoration technology cannot be used directly. In this paper, a novel HSIs restoration approach is proposed that integrates low-rank (LR) prior and spatial-spectral total variation with directional information. Specifically, by analyzing the characteristic of spatial-dependent edge and texture directional structure, a spatial-spectral directional total variation (SSDTV) regularization is defined. Then, considering the HSIs as a cube data, the proposed method utilizes SSDTV regularization to characterize spatial-spectral smoothness, as well as LR regularization to constrain spectral consistency. Finally, an extended alternating direction method of multipliers algorithm is designed to achieve simple and fast implementation, in which the complex optimization problem is separated into several easier subproblems. Both simulated and real-world HSIs experiments indicated that the proposed method is effective and numerically feasible for HSIs Restoration.

Highlights

  • H YPERSPECTRAL image (HSI) data are obtained from imaging spectrometers [1]

  • All of these TV-based approaches only use the magnitudes of the gradients to calculate the spatial weights of each pixel, aiming to eliminate the influence of largegradient-magnitude values at boundaries and edges. They still fail to preserve image detail information very well. Based on these previous works, a novel HSIs restoration method is proposed in this paper that involves low-rank prior information and spatial-spectral directional total variation (SSDTV), termed as LRSSDTV, which simultaneously removes some kinds of noise, such as impulse noise, Gaussian noise, stripe noise and deadlines

  • The diversity and complexity of mixed noise brings a great challenge to HSI processing and analysis

Read more

Summary

INTRODUCTION

H YPERSPECTRAL image (HSI) data are obtained from imaging spectrometers [1]. These measure the energy of light between the ultraviolet and infrared wavelengths that is reflected off the earth’s surface in the same region, providing several hundreds narrow, contiguous spectral bands. For the mixed-noise population recovery problem associated with HSI, He et al [35] introduced a uniform framework by combing the nuclear norm, the TV term and the L1-norm, proposing a HSI denoising approach termed ’total-variationregularized low-rank matrix factorization’ (LRTV), which combined multiple prior information in order to strengthen the spatial structural constraint. They still fail to preserve image detail information very well Based on these previous works, a novel HSIs restoration method is proposed in this paper that involves low-rank prior information and spatial-spectral directional total variation (SSDTV), termed as LRSSDTV, which simultaneously removes some kinds of noise, such as impulse noise, Gaussian noise, stripe noise and deadlines.

RELATED WORK
HYPERSPECTRAL IMAGE RESTORATION MODEL
LOW-RANK REGULARIZATION
SPATIAL-SPECTRAL DIRECTIONAL TOTAL VARIATION REGULARIZATION
OPTIMIZATION PROCEDURE
PARAMETER DETERMINATION
EXPERIMENTAL RESULTS AND DISCUSSION
SIMULATED HYPERSPECTRAL IMAGE EXPERIMENT
DISCUSSION
CONCLUSION
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call