Browsing by Author "Chang, Chein-I"
Now showing 1 - 20 of 23
Results Per Page
Sort Options
Item Background Annihilated Target-Constrained Interference-Minimized Filter (TCIMF) for Hyperspectral Target Detection(IEEE, 2022-09-21) Chen, Jie; Chang, Chein-ITarget-constrained interference-minimized filter (TCIMF) has been widely used in various target detection applications for hyperspectral data exploitation. However, like other classic target detection algorithms, the complex background (BKG) of a scene significantly impacts its performance. To better cope with BKG this paper develops a BKG-annihilated TCIMF (BA-TCIMF) which can be implemented in two stages with BKG annihilation in the first stage followed by target detectability enhancement and target BKG suppression performed by TCIMF in the second stage. In particular, the second stage extracts additional BKG signatures from the BKG-annihilated data as unwanted signatures to enhance target detectability via orthogonal subspace projection (OSP), while suppressing target BKG in the BKG-annihilated data by constrained energy minimization (CEM). Depending upon how these two stages are carried out, three versions of BATCIMF, data sphered BA-TCIMF (DS-BA-TCIMF), low rank and sparse matrix decomposition (LRaSMD) BA-TCIMF (LRaSMD-BA-TCIMF) and component decomposition analysisBA-TCIMF (CDA-BA-TCIMF) are derived. Experimental results demonstrate that BA-TCIMF performs as it is designed and better than many existing target detection algorithmsItem Band Sampling of Hyperspectral Anomaly Detection in Effective Anomaly Space(IEEE, 2023-12-25) Chang, Chein-I; Lin, Chien-Yu; Hu, PeterThis article investigates four issues, background (BKG) suppression (BS), anomaly detectability, noise effect, and interband correlation reduction (IBCR), which have significant impacts on its performance. Despite that a recently developed effective anomaly space (EAS) was designed to use data sphering (DS) to remove the second-order data statistics characterized by BKG, enhance anomaly detectability, and reduce noise effect, it does not address the IBCR issue. To cope with this issue, this article introduces band sampling (BSam) into EAS to reduce IBCR and further suppress BKG more effectively. By implementing EAS in conjunction with BSam (EAS-BSam), these four issues can be resolved altogether for any arbitrary anomaly detector. It first modifies iterative spectral–spatial hyperspectral anomaly detection (ISSHAD) to develop a new variant of ISSHAD, called iterative spectral–spatial maximal map (ISSMax), and then generalizes ISSMax to EAS-ISSMax, which further enhances anomaly detectability and noise removal. Finally, EAS-BSam is implemented to reduce IBCR. As a result, combining EAS, BSam, and ISSMax yields four versions: EAS-BSam, EAS-ISSMax, BSam-SSMAX, and EAS-BSam-SSMax. Such integration presents a great challenge because all these four versions are derived from different aspects, iterative spectral–spatial feedback process, compressive sensing, and low-rank and sparse matrix decomposition. Experiments demonstrate that EAS-BSam and EAS-BSam-SSMax show their superiority to ISSHAD and many current existing hyperspectral anomaly detection (HAD) methods.Item Band Subset Selection for Hyperspectral Image Classification(MDPI, 2018) Yu, Chunyan; Song, Meiping; Chang, Chein-IThis paper develops a new approach to band subset selection (BSS) for hyperspectral image classification (HSIC) which selects multiple bands simultaneously as a band subset, referred to as simultaneous multiple band selection (SMMBS), rather than one band at a time sequentially, referred to as sequential multiple band selection (SQMBS), as most traditional band selection methods do. In doing so, a criterion is particularly developed for BSS that can be used for HSIC. It is a linearly constrained minimum variance (LCMV) derived from adaptive beamforming in array signal processing which can be used to model misclassification errors as the minimum variance. To avoid an exhaustive search for all possible band subsets, two numerical algorithms, referred to as sequential (SQ) and successive (SC) algorithms are also developed for LCMV-based SMMBS, called SQ LCMV-BSS and SC LCMV-BSS. Experimental results demonstrate that LCMV-based BSS has advantages over SQMBS.Item A compressed sensing approach to hyperspectral classification(SPIE, 2019-05-13) Porta, C. J. Della; Lampe, Bernard; Bekit, Adam; Chang, Chein-IAlthough hyperspectral technology has continued to improve over the years, its use is often still limited due to size, weight and power (SWaP) constraints. One of the more taxing requirements, is the need to sample a large number of very fine spectral bands. The prohibitively large size of hyperpsectral data creates challenges in both archival and processing. Compressive sensing is an enabling technology for reducing the overall processing and SWaP requirements. This paper explores the viability of performing classification on sparsely sampled hyperspectral data without the need of performing sparse reconstruction. In particular, a spatial-spectral classifier based on a Support Vector Machine (SVM) and edgepreserving filters (EPFs) is applied directly in the compressed domain. The well-known Restricted Isometry Property (RIP) and a random spectral sampling strategy are used to evaluate analytically, the error between the compressed classifier and the full band classifier. The mathematical analysis presented shows that the classification error can be expressed in terms of the Restricted Isometry Constant (RIC) and that it is indeed possible to achieve full classification performance in the compressed domain, given that sufficient sampling conditions are met. A set of experiments are performed to empirically demonstrate compressed classification. Images from both the Airborne Visible / Infrared Imaging Spectrometer (AVIRIS) and Reflective Optics System Imaging Spectrometer (ROSIS) are examined to draw inferences on the impact of scene complexity. The results presented clearly demonstrate the possibility of compressed classification and lead to several open research questions to be addressed in future work.Item Edge-Inferring Graph Neural Network with Dynamic Task-Guided Self-Diagnosis for Fewshot Hyperspectral Image Classification(IEEE, 2022-08-04) Yu, Chunyan; Huang, Jiahui; Song, Meiping; Wang, Yulei; Chang, Chein-IThe current hyperspectral image classification (HSIC) model based on the convolutional neural network for feature extraction and softmax classifier has been prone to the barrier of label prediction with limited samples. Substituting for the enormously complicated work of terrain labeling, few-shot learning provides a popular option for HSIC with very few annotated samples. In this article, we proposed a novel edge-inferring framework with the metalearning paradigm for hyperspectral few-shot classification (HSFSC), in which a graph neural network for similarity measurement is first presented to iteratively infer edge labels with the exploitation of instance-level similarity and the distribution-level similarity. Besides, in the metatraining stage, the pixel prediction model and the patch prediction model based on edge-inferring architecture are concretized jointly to improve the classification accuracy of the test samples. Expressly, at the metatesting phase, the dynamic task-guided self-diagnosis strategy is developed for the first time to diagnose the samples separability of the current classification task, which is responsible for dynamically assigning the most reliable results based on the generated reliability grade of the sample. The extensive experimental results and analysis of three hyperspectral image datasets demonstrate the superiority of the proposed HSFSC architecture compared with other advanced methods.Item Estimating Optimal Number of Compressively Sensed Bands for Hyperspectral Classification via Feature Selection(IEEE, 2021-11-16) Porta, Charles Della; Chang, Chein-ICompressive sensing (CS) has received considerable interest in hyperspectral sensing. Recent articles have also exploited the benefits of CS in hyperspectral image classification (HSIC) in the compressively sensed band domain (CSBD). However, on many occasions, the requirement of full bands is not necessary for HSIC to perform well. So, a great challenge arises in determining the minimum number of compressively sensed bands (CSBs), n CSB , needed to achieve full-band performance. Practically, the value of n CSB varies with the complexity of an imaged scene. Although virtual dimensionality (VD) has been used to estimate the number of bands to be selected, n BS , it is not applicable to CSBD because a CSB is actually a mixture of n CSB bands sensed by a random sensing matrix, while VD is used to estimate n BS which is the number of single bands to be selected. As expected, n CSB will be generally smaller than n BS . To estimate an optimal value of n CSB , two feature selection approaches, filter and wrapper methods, are proposed to extract scene features that can be used to estimate the minimum value of n CSB required to maximize performance with minimum redundancy. Specifically, these methods are fully automated by leveraging optimal partitioning schemes which enable classification to further reduce storage requirements in CSBD. Finally, a set of experiments are conducted using real-world hyperspectral images to demonstrate the viability of the proposed approach.Item Exploration of Data Scene Characterization and 3D ROC Evaluation for Hyperspectral Anomaly Detection(MDPI, 2023-12-28) Chang, Chein-I; Chen, Shuhan; Zhong, Shengwei; Shi , YidanWhether or not a hyperspectral anomaly detector is effective is determined by two crucial issues, anomaly detectability and background suppressibility (BS), both of which are very closely related to two factors, the datasets used for a selected hyperspectral anomaly detector and detection measures used for its performance evaluation. This paper explores how anomaly detectability and BS play key roles in hyperspectral anomaly detection (HAD). To address these two issues, we investigate three key elements attributed to HAD. One is a selected hyperspectral anomaly detector, and another is the datasets used for experiments. The third one is the detection measures used to evaluate the effectiveness of a hyperspectral anomaly detector. As for hyperspectral anomaly detectors, twelve commonly used anomaly detectors were evaluated and compared. To address the appropriate use of datasets for HAD, seven popular and widely used datasets were studied for HAD. As for the third issue, the traditional area under a receiver operating characteristic (ROC) curve of detection probability—PD versus false alarm probability, PF, (AUC(D,F))—was extended to 3D ROC analysis where a 3D ROC curve was developed to generate three 2D ROC curves from which eight detection measures could be derived to evaluate HAD in all round aspects, including anomaly detectability, BS and joint anomaly detectability and BS. Qualitative analysis showed that many works reported in the literature which claimed that their developed hyperspectral anomaly detectors performed better than other anomaly detectors are actually not true because they overlooked these two issues. Specifically, a comprehensive study via extensive experiments demonstrated that these 3D ROC curve-derived detection measures can be further used to address the various characterizations of different data scenes and also to provide explanations as to why certain data scenes are not suitable for HAD.Item Hyperspectral Band Selection based on Improved Affinity Propagation(IEEE, 2021) Zhu, Qingyu; Wang, Yulei; Wang, Fengchao; Song, Meiping; Chang, Chein-IDimensionality reduction is a common method to reduce the computational complexity of hyperspectral images and improve the classification performance. Band selection is one of the most commonly used methods for dimensionality reduction. Affinity propagation (AP) is a clustering algorithm that has better performance than traditional clustering methods. This paper proposes an improved AP algorithm (IAP), which divides each intrinsic cluster into several subsets, and combines the information entropy to change the initial availability matrix to obtain a suitable number of clustering results with arbitrary shapes. The experimental results on the public hyperspectral data set show that the band combination selected by IAP has a better classification accuracy compared with all bands data set and band subset by traditional AP algorithm.Item Iterative constrained energy minimization convolutional neural network for hyperspectral image classification(SPIE, 2019-05-14) Xue, Bai; Shang, Xiaodi; Zhong, Shengwei; Hu, Peter F.; Chang, Chein-IIn hyperspectral image classification, how to jointly take care of spectral and spatial information received considerable interest lately, and many spectral-spatial classification approaches have been proposed. Unlike spectral-spatial classifications which are developed from traditional aspect, iterative constrained energy minimization (ICEM) and iterative target-constrained interference-minimized classifier (ITCIMC) approaches are developed from subpixel detection and mixed pixel classification point of view, and generally performs better than existing spectral-spatial approaches in terms of several measurements, such as accuracy rate and precision rate. Recently, convolutional neural networks (CNNs) have been successfully applied to visual imagery classification and have received great attention in hyperspectral image classification, due to the outstanding ability of CNN to capture spatial information. This paper extends ICEM to iterative constrained energy minimization convolution neural network approach for hyperspectral image classification. In order to capture spatial information, instead of Gaussian filter, CNN is utilized to generate binary pixelwise classification map for constrained energy minimization (CEM) detection results, and CNN classification map is feedbacked into hyperspectral bands, and then CEM detection is reprocessed in an iteration manner. Since CNN can reduce the performance of precision rate, a background recovery procedure is designed, to recover background detection map from CEM detection map and add it into CEM result as a new detection map.Item Iterative Random Training Sampling Convolutional Neural Network for Hyperspectral Image Classification(IEEE, 2023-05-26) Chang, Chein-I; Liang, Chia-Chen; Hu, PeterConvolutional neural network (CNN) has received considerable interest in hyperspectral image classification (HSIC) lately due to its excellent spectral–spatial feature extraction capability. To improve CNN, many approaches have been directed at exploring the infrastructure of its network by introducing different paradigms. This article takes a rather different approach by developing an iterative CNN that extends a CNN by including a feedback system to repeatedly process the same CNN in an iterative manner. Its idea is to take advantage of a recently developed iterative training sampling spectral–spatial classification (IRTS-SSC) that allows CNN to update its spatial information of classification maps through a feedback spatial fil- tering system via IRTS. The resulting CNN is called iterative random training sampling CNN (IRTS-CNN) with several unique features. First, IRTS-CNN combines CNN and IRTS-SSC into one paradigm, an architecture that has never been investigated in the past. Second, it implements a series of spatial filters to capture spatial information of classified data samples and further feeds this information back via an iterative process to expand the current input data cube for the next iteration. Third, it utilizes the expanded data cube to randomly reselect training samples and then to reimplement CNN iteratively. Last but not least, IRTS-CNN provides a general framework that can implement any arbitrary CNN as an initial classifier to improve its performance through an iterative process. Extensive experiments are conducted to demonstrate that IRTS-CNN indeed significantly improves CNN, specifically when only a small size of limited training samples is used.Item Iterative Random Training Sampling Spectral Spatial Classification for Hyperspectral Images(IEEE, 2020-07-09) Chang, Chein-I; Ma, Kenneth Yeonkong; Liang, Chia-Chen; Kuo, Yi-Mei; Chen, Shuhan; Zhong, ShengweiHyperspectral image classification (HSIC) has generated considerable interests over the past years. However, one of challenging issues arising in HSIC is inconsistent classification, which is mainly caused by random training sampling (RTS) of selecting training data. This is because a different set of training samples may produce a different classification result. A general approach to addressing this problem is the so-called K-fold method which implements RTS K times and takes the average of overall accuracy with respect to standard deviation to describe a confidence level of classification performance. To deal with this issue, this article develops an iterative RTS (IRTS) method as an alternative to the K-fold method to reduce the uncertainty caused by RTS. Its idea is to add the spatial filtered classification maps to the image cube that is currently being processed via feedback loops to augment image cubes iteratively. Then, the training samples will be reselected randomly from the new augmented image cubes iteration-by-iteration. As a result, the training samples selected from each iteration will be updated by new added spatial information captured by spatial filters implemented at the iteration. The experimental results clearly demonstrate that IRTS successfully improves classification accuracy as well as reduces inconsistency in results.Item Iterative Scale-Invariant Feature Transform for Remote Sensing Image Registration(IEEE, 2020-07-22) Chen, Shuhan; Zhong, Shengwei; Xue, Bai; Li, Xiaorun; Zhao, Liaoying; Chang, Chein-IDue to significant geometric distortions and illumination differences, developing techniques for high precision and robust multisource remote sensing image registration poses a great challenge. This article presents an iterative image registration approach, called iterative scale-invariant feature transform (ISIFT) for remote sensing images, which extends the traditional scale-invariant feature transform (SIFT)-based registration system to a close-feedback SIFT system that includes a rectification feedback loop to update rectified parameters in an iterative manner. Its key idea uses consistent feature point sets obtained by maximum similarity to calculate new alignment parameters to rectify the current sensed image and the resulting rectified sensed image is then fed back to update and replace the current sensed image as a new sensed image to reimplement SIFT for next iteration. The same process is repeated iteratively until an automatic stopping rule is satisfied. To evaluate the performance of ISIFT, both the simulated and real images are used for experiments for the validation of ISIFT. In addition, several data sets are particularly designed to conduct a comparative study and analysis with existing state-of-the-art methods. Furthermore, experiments with different rotation are also performed to verify the adaptability of ISIFT under different rotation distortions. The experimental results demonstrate that ISIFT improves performance and produces better registration accuracy than traditional SIFT-based methods and existing state-of-the-art methods.Item A Novel FPGA-Based Architecture for Fast Automatic Target Detection in Hyperspectral Images(MDPI, 2019-01-14) Lei, Jie; Wu, Lingyun; Li, Yunsong; Xie, Weiying; Chang, Chein-I; Zhang, Jintao; Huang, BiyingOnboard target detection of hyperspectral imagery (HSI), considered as a significant remote sensing application, has gained increasing attention in the latest years. It usually requires processing huge volumes of HSI data in real-time under constraints of low computational complexity and high detection accuracy. Automatic target generation process based on an orthogonal subspace projector (ATGP-OSP) is a well-known automatic target detection algorithm, which is widely used owing to its competitive performance. However, ATGP-OSP has an issue to be deployed onboard in real-time target detection due to its iteratively calculating the inversion of growing matrices and increasing matrix multiplications. To resolve this dilemma, we propose a novel fast implementation of ATGP (Fast-ATGP) while maintaining target detection accuracy of ATGP-OSP. Fast-ATGP takes advantage of simple regular matrix add/multiply operations instead of increasingly complicated matrix inversions to update growing orthogonal projection operator matrices. Furthermore, the updated orthogonal projection operator matrix is replaced by a normalized vector to perform the inner-product operations with each pixel for finding a target per iteration. With these two major optimizations, the computational complexity of ATGP-OSP is substantially reduced. What is more, an FPGA-based implementation of the proposed Fast-ATGP using high-level synthesis (HLS) is developed. Specifically, an efficient architecture containing a bunch of pipelines being executed in parallel is further designed and evaluated on a Xilinx XC7VX690T FPGA. The experimental results demonstrate that our proposed FPGA-based Fast-ATGP is able to automatically detect multiple targets on a commonly used dataset (AVIRIS Cuprite Data) at a high-speed rate of 200 MHz with a significant speedup of nearly 34.3 times that of ATGP-OSP, while retaining nearly the same high detection accuracyItem Optical Remote Sensing Image Registration Using Spatial-Consistency and Average Regional Information Divergence Minimization via Quantum-Behaved Particle Swarm Optimization(MDPI, 2020-09-19) Chen, Shuhan; Xue, Bai; Yang, Han; Li, Xiaorun; Zhao, Liaoying; Chang, Chein-IDue to invariance to significant intensity differences, similarity metrics have been widely used as criteria for an area-based method for registering optical remote sensing image. However, for images with large scale and rotation difference, the robustness of similarity metrics can greatly determine the registration accuracy. In addition, area-based methods usually require appropriately selected initial values for registration parameters. This paper presents a registration approach using spatial consistency (SC) and average regional information divergence (ARID), called spatial-consistency and average regional information divergence minimization via quantum-behaved particle swarm optimization (SC-ARID-QPSO) for optical remote sensing images registration. Its key idea minimizes ARID with SC to select an ARID-minimized spatial consistent feature point set. Then, the selected consistent feature set is tuned randomly to generate a set of M registration parameters, which provide initial particle warms to implement QPSO to obtain final optimal registration parameters. The proposed ARID is used as a criterion for the selection of consistent feature set, the generation of initial parameter sets, and fitness functions used by QPSO. The iterative process of QPSO is terminated based on a custom-designed automatic stopping rule. To evaluate the performance of SC-ARID-QPSO, both simulated and real images are used for experiments for validation. In addition, two data sets are particularly designed to conduct a comparative study and analysis with existing state-of-the-art methods. The experimental results demonstrate that SC-ARID-QPSO produces better registration accuracy and robustness than compared methods.Item Orthogonal Subspace Projection Target Detector for Hyperspectral Anomaly Detection(IEEE, 2021-03-25) Chang, Chein-I; Cao, Hongju; Song, MeipingOrthogonal subspace projection (OSP) is a versatile hyperspectral imaging technique which has shown great potential in dimensionality reduction, target detection, spectral unmixing, etc. However, due to its inherent requirement of prior target knowledge, OSP has not been explored in anomaly detection. This article takes advantage of an unsupervised OSP-based algorithm, automatic target generation process (ATGP), and a recently developed OSP-go decomposition (OSP-GoDec) along with data sphering (DS) to make OSP applicable to anomaly detection. Its idea is to implement ATGP on the background (BKG) and target subspaces constructed from the low-rank matrix L and sparse matrix S generated by OSP-GoDec to derive an OSP-based anomaly detector (OSP-AD). In particular, OSP-AD also includes DS to remove BKG interference from the target subspace so as to enhance anomaly detection. Surprisingly, operating data samples on different constructions of the BKG subspace and the target subspace yields various versions of OSP-AD. Experiments show that given an appropriate construction of the BKG subspace and the target subspace, OSP-AD can be shown to outperform existing anomaly detectors including Reed-Xiaoli anomaly detector and collaborative representation-based anomaly detector (CRD).Item Orthogonal Subspace Projection-Based Go-Decomposition Approach to Finding Low-Rank and Sparsity Matrices for Hyperspectral Anomaly Detection(IEEE, 2020-07-14) Chang, Chein-I; Cao, Hongju; Chen, Shuhan; Shang, Xiaodi; Yu, Chunyan; Song, MeipingLow-rank and sparsity-matrix decomposition (LRaSMD) has received considerable interests lately. One of effective methods for LRaSMD is called go decomposition (GoDec), which finds low-rank and sparse matrices iteratively subject to the predetermined low-rank matrix order m and sparsity cardinality k. This article presents an orthogonal subspace-projection (OSP) version of GoDec to be called OSPGoDec, which implements GoDec in an iterative process by a sequence of OSPs to find desired low-rank and sparse matrices. In order to resolve the issues of empirically determining p = m + j and k, the well-known virtual dimensionality (VD) is used to estimate p in conjunction with the Kuybeda et al. developed minimax-singular value decomposition (MX-SVD) in the maximum orthogonal complement algorithm (MOCA) to estimate k. Consequently, LRaSMD can be realized by implementing OSP-GoDec using p and k determined by VD and MX-SVD, respectively. Its application to anomaly detection demonstrates that the proposed OSP-GoDec coupled with VD and MX-SVD performs very effectively and better than the commonly used LRaSMD-based anomaly detectors.Item A Simplified 2D-3D CNN Architecture for Hyperspectral Image Classification Based on Spatial–Spectral Fusion(IEEE, 2020-04-27) Yu, Chunyan; Han, Rui; Song, Meiping; Liu, Caiyu; Chang, Chein-IConvolutional neural networks (CNN) have led to a successful breakthrough for hyperspectral image classification (HSIC). Due to the intrinsic spatial-spectral specificities of a hyperspectral cube, feature extraction with 3-D convolution operation is a straightforward way for HSIC. However, the overwhelming features obtained from the original 3-D CNN network suffers from the overfitting and more training cost problem. To address this issue, in this article, a novel HSIC framework based on a simplified 2D-3D CNN is implemented by the cooperation between a 2-D CNN and a 3-D convolution layer. First, the 2-D convolution block aims to extract the spatial features abundantly involved spectral information as a training channel. Then, the 3-D CNN approach primarily concentrates on exploiting band co-relation data by using a reduced kernel. The proposed architecture achieves the spatial and spectral features simultaneously based on a joint 2D-3D pattern to achieve superior fused feature for the subsequent classification. Furthermore, a deconvolution layer intends to enhance the robustness of the deep features is utilized in the proposed CNN network. The results and analysis of extensive real HSIC experiments demonstrate that the proposed light-weighted 2D-3D CNN network can effectively extract refined features and improve the classification accuracy.Item Target-to-Anomaly Conversion for Hyperspectral Anomaly Detection(IEEE, 2022-10-03) Chang, Chein-I— A known target detection assumes that the target to be detected is provided a priori, while anomaly detection is an unknown target detection without any prior knowledge. As a result, known target detection generally performs searchbefore-detect detection in an active mode, referred to as active target detection as opposed to anomaly detection, which performs throw-before-detect detection in a passive mode, referred to as passive target detection. Accordingly, techniques designed for these two types of detection are completely different. This article shows that there is indeed a mechanism, called target-to-anomaly conversion, which can convert hyperspectral target detection (HTD) to hyperspectral anomaly detection (HAD) via a novel idea, called dummy variable trick (DVT). By virtue of such target-to-anomaly conversion many well-known target detection techniques, such as likelihood ratio test (LRT), constrained energy minimization (CEM), and orthogonal subspace projection (OSP), the spectral angle mapper (SAM) and the adaptive cosine estimator (ACE) can be converted to their corresponding anomaly detectors, referred to as target-to-anomaly conversion-derived anomaly detectors (TAC-ADs). Since a target detector requires target knowledge while TAC-AD does not, a direct use of TACAD is not effective. To make TAC-AD work, a newly developed approach to effective anomaly space (EAS) is implemented in conjunction with TAC-AD so that anomalies can be retained in EAS and interference, and noise including background (BKG) can be removed from EAS. The experiments demonstrate that TAC-AD operating in EAS performs better than many existing anomaly detection approaches, including model-based methods.Item Underwater Hyperspectral Target Detection with Band Selection(MDPI, 2020-03-25) Fu, Xianping; Shang, Xiaodi; Sun, Xudong; Yu, Haoyang; Song, Meiping; Chang, Chein-ICompared to multi-spectral imagery, hyperspectral imagery has very high spectral resolution with abundant spectral information. In underwater target detection, hyperspectral technology can be advantageous in the sense of a poor underwater imaging environment, complex background, or protective mechanism of aquatic organisms. Due to high data redundancy, slow imaging speed, and long processing of hyperspectral imagery, a direct use of hyperspectral images in detecting targets cannot meet the needs of rapid detection of underwater targets. To resolve this issue, a fast, hyperspectral underwater target detection approach using band selection (BS) is proposed. It first develops a constrained-target optimal index factor (OIF) band selection (CTOIFBS) to select a band subset with spectral wavelengths specifically responding to the targets of interest. Then, an underwater spectral imaging system integrated with the best-selected band subset is constructed for underwater target image acquisition. Finally, a constrained energy minimization (CEM) target detection algorithm is used to detect the desired underwater targets. Experimental results demonstrate that the band subset selected by CTOIFBS is more effective in detecting underwater targets compared to the other three existing BS methods, uniform band selection (UBS), minimum variance band priority (MinV-BP), and minimum variance band priority with OIF (MinV-BP-OIF). In addition, the results also show that the acquisition and detection speed of the designed underwater spectral acquisition system using CTOIFBS can be significantly improved over the original underwater hyperspectral image system without BS.Item Unsupervised Domain Adaptation With Dense-Based Compaction for Hyperspectral Imagery(IEEE, 2021-11-18) Yu, Chunyan; Liu, Caiyu; Yu, Haoyang; Song, Meiping; Chang, Chein-IEnormously hard work of label obtaining leads to the lack of enough annotated samples in the hyperspectral imagery (HSI). The mentioned reality inferred the unsupervised classification performance barely satisfactorily. Unsupervised domain adaptation is exploited for knowledge delivery from a labeled source domain to boost the performance on an unlabeled target domain. In this article, we propose an unsupervised domain adaptation architecture with dense-based compaction (UDAD) for HSI classification (HSIC). The processes of spectral–spatial feature compaction, unsupervised domain adaptation, and classifier training are incorporated with an integrated framework to complete the HSI cross-scene classification. The core of the proposed framework is to utilize adversarial domain learning to reduce the domain discrepancy. To this end, the classifier trained in the source domain would accomplish well in the target domain for the unsupervised HSIC. Besides, to extract the discriminative spectral–spatial feature for the HSI domains, a dense-based compaction network is applied to complete the semisymmetric mapping. Our experiments illustrate that the UDAD model yields more effective classification performance than other state-of-the-art unsupervised HSIC methods.