In this article, a two-stream convolutional network-based target detector (denoted as TSCNTD) for hyperspectral images is proposed. The TSCNTD utilizes the networks to extract abundant spectral information in images. For background samples, finds enough typical pixels via hybrid sparse representation and classification-based pixel selection strategy entire image. To tackle problem under limited novel synthesis method proposed generate sufficient samples with priori some pixels. Once are...

Hyperspectral imagery (HSI) with a high spectral resolution contains hundreds and even thousands of bands, conveys abundant information, which provides unique advantage for target detection. A number classical detectors have been proposed based on the linear mixing model (LMM) sparsity-based model. Compared LMM, present better performance dealing variability. Despite great success in recent years, one problem all state-of-the-art models still exist: dictionary is formed via training samples...

Due to the limitation of target size and spatial resolution, targets interest in hyperspectral images (HSIs) often appear as subpixel targets, which makes detection still faces an important bottleneck, that is, detection. In this article, we propose a new detector by learning single spectral abundance for (denoted LSSA). Different from most existing detectors are designed based on match spectrum assisted information or focusing background, proposed LSSA addresses problem detecting directly....

Abstract To reduce the driving load and enhance heat exchange capacity elastocaloric refrigeration efficiency, increasing interests in porous structure design laser-based additive manufacturing (LAM) of NiTi materials with a large specific surface area have been emerging. As type characteristic unit components, we mainly focused on LAM process optimization effect NiTi-based thin-walled structures (TWSs) this work. Firstly, systemically studied influence laser processing parameter forming...

UAV-borne hyperspectral remote sensing has emerged as a promising approach for underwater target detection (UTD). However, its effectiveness is hindered by spectral distortions in nearshore environments, which compromise the accuracy of traditional UTD (HUTD) methods that rely on bathymetric model. These lead to significant uncertainty and background spectra, challenging process. To address this, we propose Hyperspectral Underwater Contrastive Learning Network (HUCLNet), novel framework...

Visible spectrum images capture limited information from just three discrete bands, often resulting in suboptimal performance underwater depth estimation (UDE) due to significant loss water absorption. In contrast, HSIs, which include hundreds of continuous provide abundant spectral that offers greater resilience against the adverse effects this paper, we conduct a comprehensive study investigate how can enhance remote sensing UDE through two key aspects: benchmark dataset and general...

Hyperspectral target detection refers to an approach that tries locate targets in a hyperspectral image (HSI) on the condition of given spectrum, which plays important role remote sensing processing. In this letter, we propose binary-class collaborative representation-based detector. The proposed algorithm uses concept each background pixel can be approximately represented by its adjacent pixels within sliding dual-window, and also some image; use represent it. Before estimating pixel,...

Calcium sulfate whiskers (CSWs) were hydroxylated with a sodium hydroxide (NaOH) solution and isolated for subsequent treatment an ethanolic 3-(methacryloxy)propyltrimethoxysilane (KH570) to introduce C=C double bonds on the CSWs' surfaces. Then, CSW-g-PMMA was prepared by grafting polymethyl methacrylate (PMMA) onto surface of modified CSW using in situ dispersion polymerization. The used as filler melt-blended polyvinyl chloride (PVC) prepare PVC-based composites. chemical structure, PMMA...

Elastocaloric refrigeration is the most promising green solid-state technology to replace conventional vapor compression refrigeration. The development direction of elastocaloric component that acts as a key part system contains large effect, low stress hysteresis, high heat exchange performance, and small driving loads. first two indices can be realized by material modification; however, last are more dependent on novel porous structure design. However, confronted with some critical...

The combined sparse and collaborative representation-based algorithm is one of the most effective methods among hyperspectral target detection based on representation dictionary learning. It encourages atoms to compete with each other background collaborate in representation. However, this method suffers from several drawbacks. In representation, an overcomplete necessary, whereas, non-negative coefficients are required. Besides, local dual window approach may result impure dictionaries...

In this article, a single-spectrum-driven binary-class sparse representation target detector (SSBSTD) via and background dictionary construction (BDC) is proposed. The SSBSTD leans upon the (BSR) model. Due to fact that spectrum usually consists in samples composed low-dimensional subspace also subspace, only should be used for sparsely representing test pixel under absent hypothesis from target-only present hypothesis. To alleviate problem there are insufficient available model, article...

In hyperspectral target detection, the conventional metric learning-based algorithms provide unique advantages in detecting targets as they do not require specific assumptions and adapt to condition of limited training samples. Nevertheless, usually learn a linear transformation for space, which is unable capture nonlinear mapping where imageries possess, especially occurs spectra variability mixing problems. To alleviate this limitation, study investigates new spatial-spectral adaptive...

Hyperspectral image (HSI) anomaly detection is an important task in remote sensing domain. In recent years, many scholars have been addicted to constructing deep network-based methods for hyperspectral and developed numerous related methods. Many of them are designed based on autoencoder, which aims reconstruct a stable background identify anomalies. However, these autoencoder-based suffer from some problems, such as ignoring the inter-band correlation HSI. That is, presents spectral...

In this paper, an encoder-decoder long short-term memory network-based anomaly detector (denoted as EDLAD) is proposed for hyperspectral images. The EDLAD aims to simultaneously alleviate contamination and build a stable background component detection. To reduce contamination, the first utilizes well-designed LSTM reconstruct image. Based on concept that pixels occupy extremely small fraction of image, network tends maintain during reconstruction process since whole image employed training...

In the era of increasingly advanced Earth Observation (EO) technologies, extracting pertinent information (such as water-bodies) from Earth's surface has become a crucial task. Deep Learning, especially via pre-trained models, currently offers highly promising approach for semantic segmentation Remote Sensing Imagery (RSI). However, effectively adapting these models to RSI tasks remains challenging. Typically, undergo fine-tuning specialized tasks, involving modifications their parameters or...

The limited training sample has become a great challenge for hyperspectral target detection with deep learning-based methods. In this paper, long short-term memory based detector is proposed. To handle the insufficient background samples, an endmember extraction pixel selection strategy proposed to select pixels from entire image. For we utilize synthesis method generate sufficient samples using given spectrum and extracted samples. Then obtained are fed into well-designed network learn...