SLIC Superpixel-Based l2,1-Norm Robust Principal Component Analysis for Hyperspectral Image Classification.

Baokai Zu,Tiejun Li,Wei Du,Yafang Li,Jingzhong Hou,Kewen Xia,Ziping He

doi:10.3390/s19030479

Abstract

Hyperspectral Images (HSIs) contain enriched information due to the presence of various bands, which have gained attention for the past few decades. However, explosive growth in HSIs’ scale and dimensions causes “Curse of dimensionality” and “Hughes phenomenon”. Dimensionality reduction has become an important means to overcome the “Curse of dimensionality”. In hyperspectral images, labeled samples are more difficult to collect because they require many labor and material resources. Semi-supervised dimensionality reduction is very important in mining high-dimensional data due to the lack of costly-labeled samples. The promotion of the supervised dimensionality reduction method to the semi-supervised method is mostly done by graph, which is a powerful tool for characterizing data relationships and manifold exploration. To take advantage of the spatial information of data, we put forward a novel graph construction method for semi-supervised learning, called SLIC Superpixel-based -norm Robust Principal Component Analysis (SURPCA2,1), which integrates superpixel segmentation method Simple Linear Iterative Clustering (SLIC) into Low-rank Decomposition. First, the SLIC algorithm is adopted to obtain the spatial homogeneous regions of HSI. Then, the -norm RPCA is exploited in each superpixel area, which captures the global information of homogeneous regions and preserves spectral subspace segmentation of HSIs very well. Therefore, we have explored the spatial and spectral information of hyperspectral image simultaneously by combining superpixel segmentation with RPCA. Finally, a semi-supervised dimensionality reduction framework based on SURPCA2,1 graph is used for feature extraction task. Extensive experiments on multiple HSIs showed that the proposed spectral-spatial SURPCA2,1 is always comparable to other compared graphs with few labeled samples.

Highlights

IntroductionHyperspectral Images (HSIs) provide comprehensive spectral information of the materials’physical properties [1], which is applied in ecosystem monitoring [2], agricultural monitoring [3,4], environmental monitoring [5], forestry [6], urban growth analysis [7,8], and mineral identification [9,10].The abundant hyperspectral image bands bring rich spectral information, but they result in the Sensors 2019, 19, 479; doi:10.3390/s19030479 www.mdpi.com/journal/sensors “Hughes phenomenon” [11,12] that reduces the accuracy and efficiency of the classification task.By reducing the dimensionality of the data, more compact low-dimensional hyperspectral images can be obtained
The confusion matrix [71] is often used in the remote sensing classification field, and its form is defined as: M = [mij ]n×n, where mij indicates that the number of pixels labeled by the j class should belong to the i class. n is the class number
We present a novel graph for Hyperspectral Images (HSIs) feature extraction, referred to as Simple Linear Iterative Clustering (SLIC)

Summary

Introduction

Hyperspectral Images (HSIs) provide comprehensive spectral information of the materials’physical properties [1], which is applied in ecosystem monitoring [2], agricultural monitoring [3,4], environmental monitoring [5], forestry [6], urban growth analysis [7,8], and mineral identification [9,10].The abundant hyperspectral image bands bring rich spectral information, but they result in the Sensors 2019, 19, 479; doi:10.3390/s19030479 www.mdpi.com/journal/sensors “Hughes phenomenon” [11,12] that reduces the accuracy and efficiency of the classification task.By reducing the dimensionality of the data, more compact low-dimensional hyperspectral images can be obtained. Hyperspectral Images (HSIs) provide comprehensive spectral information of the materials’. The abundant hyperspectral image bands bring rich spectral information, but they result in the Sensors 2019, 19, 479; doi:10.3390/s19030479 www.mdpi.com/journal/sensors “Hughes phenomenon” [11,12] that reduces the accuracy and efficiency of the classification task. By reducing the dimensionality of the data, more compact low-dimensional hyperspectral images can be obtained. Dimensionality reduction mainly includes feature selection and feature extraction [13]. Feature selection methods reduce the dimensions of the original data by selecting the most representative and distinguishing features [14]. Feature extraction methods combine multiple features linearly and non-linearly on the basis of maintaining the data structure information [15].

Objectives

Methods

Results

Conclusion