Joint Sparse and Low-Rank Multitask Learning with Laplacian-Like Regularization for Hyperspectral Classification

Abstract

Multitask learning (MTL) has recently provided significant performance improvements in supervised classification of hyperspectral images (HSIs) by incorporating shared information across multiple tasks. However, the original MTL cannot effectively exploit both local and global structures of the HSI and the class label information is not fully used. Moreover, although the mathematical morphology (MM) has attracted considerable interest in feature extraction of HSI, it remains a challenging issue to sufficiently utilize multiple morphological profiles obtained by various structuring elements (SEs). In this paper, we propose a joint sparse and low-rank MTL method with Laplacian-like regularization (termed as sllMTL) for hyperspectral classification by utilizing the three-dimensional morphological profiles (3D-MPs) features. The main steps of the proposed method are twofold. First, the 3D-MPs are extracted by the 3D-opening and 3D-closing operators. Different SEs are adopted to result in multiple 3D-MPs. Second, sllMTL is proposed for hyperspectral classification by taking the 3D-MPs as features of different tasks. In the sllMTL, joint sparse and low-rank structures are exploited to capture the task specificity and relatedness, respectively. Laplacian-like regularization is also added to make full use of the label information of training samples. Experiments on three datasets demonstrate the OA of the proposed method is at least about 2% higher than other state-of-the-art methods with very limited training samples.

Hyperspectral Anomaly Detection via Discriminative Feature Learning with Multiple-Dictionary Sparse Representation

“…For LRX method, in order to avoid covariance matrix singular problem, we finally use four pairs of window sizes (outside, inner) including (17,7), (17,9), (19,7), and (19,9) by comprehensively considering all the spectral dimensions of all the hyperspectral images. As for SRD and CRD, we define six kinds of different sizes including (13,7), (15,7), (17,7), (17,9), (19,7), and (19,9). In addition, the regularized parameter λ involved in CRD method is set as 10 −6 referring to its original work [45].…”

“…These spectra are divided into hundreds of approximately continuous and very narrow spectral bands which have a strong ability to precisely characterize different objects and accurately recognize the subtle differences between surface materials [2,3]. Benefiting from its high spectral resolution, hyperspectral image has been successfully used in many applications [4], such as target detection [5,6], image classification [7,8], band selection [9,10], and hyperspectral unmixing [11,12].…”

Hyperspectral Anomaly Detection Based on Separability-Aware Sample Cascade

“…These spectra are represented by hundreds of continuous bands that can meticulously describe the characteristics of different materials to recognize their subtle differences [3]. Therefore, owing to this good discriminative property of hyperspectral image, it has been widely used in many remote sensing research fields [4,5], such as image denoising [6,7], hyperspectral unmixing [8,9], band selection [10,11], target detection [12,13], and image classification [14,15]. They all have important practical applications in geological exploration, urban remote sensing and planning management, environment and disaster monitoring, precision agriculture, archaeology, etc.…”

A Multiscale Hierarchical Model for Sparse Hyperspectral Unmixing

“…Hyperspectral images possess abundant spectral information, which makes target detection and classification become feasible [1,2]. However, due to a low spatial resolution of hyperspectral sensor and the complex background, amounts of mixed pixels exist in the image and that makes it impossible to determine the material directly from the pixel level.…”