Hyperspectral image mixed noise removal via tensor robust principal component analysis with tensor-ring decomposition

Abstract

ABSTRACTDue to the instability of sensors and other factors, hyperspectral images (HSIs) are inevitably polluted by various types of mixed noise. To explore a better denoising method based on the existing research, combining the denoising advantages of tensor-ring (TR) decomposition and tensor robust principal component analysis (TRPCA), a mixed-noise removal method for HSIs is proposed in this paper. First, TRPCA maintains the tensor structure of the image itself, accurately recovers the low-rank part and the sparse part from their sum and separates the sparse noise in the form of sparse tensors. Then, TR decomposition is introduced to denoise the low-rank tensors. To verify the effectiveness and superiority of this method, experiments are carried out on two simulated data sets and two real data sets. Compared with the traditional denoising methods and several existing improved denoising methods from both visual and quantitative aspects, the proposed TRPCA-TR method provides better denoising results.KEYWORDS: Hyperspectral image (HSI)mixed noise removaltensor ring (TR) decompositiontensor robust principal component analysis (TRPCA) AcknowledgementsThe authors would like to thank the anonymous reviewers for their constructive comments.Disclosure statementNo potential conflict of interest was reported by the author(s).Data availability statementThe simulated Indian Pines data set used in this study were obtained from the MultiSpec, Purdue University, West Lafayette, IN (https://engineering.purdue.edu/~biehl/MultiSpec/hyperspectral.html). The Washington DC Mall data set (WDC Mall), the Salinas-A data set and the real Indian Pines data set used in this study were obtained from the Remote Sensing Laboratory, School of Surveying and Geospatial Engineering, Tehran, Iran (https://rslab.ut.ac.ir/en/data).Notes1. https://engineering.purdue.edu/~biehl/MultiSpec/hyperspectral.html. (He et al. Citation2016).2. https://rslab.ut.ac.ir/data.3. https://rslab.ut.ac.ir/data.4. https://rslab.ut.ac.ir/data.Additional informationFundingThis work was supported in part by the Sichuan Science and Technology Program under Grant Nos. 2021YJ0351 and 2021YFG0319, in part by the Opening Fund of Geomathematics Key Laboratory of Sichuan Province under Grant Nos. scsxdz2021yb02 and scsxdz2019zd03.