Band Selection-Based Dimensionality Reduction for Change Detection in Multi-Temporal Hyperspectral Images

Abstract

This paper proposes to use band selection-based dimensionality reduction (BS-DR) technique in addressing a challenging multi-temporal hyperspectral images change detection (HSI-CD) problem. The aim of this work is to analyze and evaluate in detail the CD performance by selecting the most informative band subset from the original high-dimensional data space. In particular, for cases where ground reference data are available or unavailable, either supervised or unsupervised CD approaches are designed. The following sub-problems in HSI-CD are investigated, including: (1) the estimated number of multi-class changes; (2) the binary CD; (3) the multiple CD; (4) the estimated optimal number of selected bands; and (5) computational efficiency. The main contribution of this paper is to provide for the first time a thorough analysis of the impacts of band selection on the HSI-CD problem, thus to fix the gap in the state-of-the-art techniques either by simply utilizing the full dimensionality of the data or exploring a complex hierarchical change analysis. It is applicable to CD problems in multispectral or PolSAR images when the feature space is expanded for discriminant feature extraction. Two real multi-temporal hyperspectral Hyperion datasets are used to validate the proposed approaches. Quantitative and qualitative experimental results demonstrated that by selecting a subset of the most informative and distinct spectral bands, the proposed approaches offered better CD performance than the state-of-the-art techniques using original full bands, without losing the change representative and discriminable capabilities of a detector.

Change Detection in Hyperspectral Images Using Recurrent 3D Fully Convolutional Networks

“…To conduct multi-class CD, we divided Ω c into Ω c1 , Ω c2 , Ω c3 , and Ω c4 depending on the pattern of changes from T 1 to T 2 . The four major land-cover changes are related to vegetation, bare soil, and water changes [33]. The ground truths with four classes of sites 1 and 2 are shown in Figure 7c,d, respectively.…”

“…These samples have high probabilities of being either changed or unchanged; thus, the accuracy of the samples affects the CD results. Many studies have verified the performance of such networks with samples extracted from the ground-truth map [30,31,33]. However, the available labeled HSIs are limited because it is difficult Remote Sens.…”

Methods and Challenges Using Multispectral and Hyperspectral Images for Practical Change Detection Applications

“…A paper by Liu et al [93] presents both unsupervised and supervised band selection-based, dimensional reduction techniques. The reduced dimension data are then used for change detection.…”

Progressive Sample Processing of Band Selection for Hyperspectral Image Transmission

“…In PBS, the bands are first prioritized by a certain criterion, and are transmitted by the order of prioritization. Du et al [15] proposed a BS-based dimensional reduction method for change detection in multi-temporal hyperspectral images. Except for the above-mentioned work, there were still various kinds of BS research published in the literature [16][17][18][19][20][21][22][23][24][25].…”

“…Many different types of BS algorithms [2][3][4][5][6][7][8][9][10][11][12][13][14][15] have been proposed in the past two decades. Most of them make an assumption that the BS problem is an optimization problem, which maximizes or minimizes a pre-defined objective function that can measure the amount of information or inter-band redundancy contained by the currently selected bands.…”