Anomaly detection is of great importance among hyperspectral applications, which aims at locating targets that are spectrally different from their surrounding background. A variety anomaly methods have been proposed in the past. However, most them fail to take high spectral correlations all pixels into consideration. Low-rank representation (LRR) has drawn a deal interest recent years, as promising model exploit intrinsic low-rank property images. Nevertheless, original LRR only analyzes...

Learning semantic correspondence between image and text is significant as it bridges the gap vision language. The key challenge to accurately find correlate shared semantics in text. Most existing methods achieve this goal by representing a weighted combination of all fragments (image regions or words), where relevant obtain more attention, otherwise less. However, despite ones contribute semantic, irrelevant will less disturb it, thus lead misalignment correlation phase. To address issue,...

Nonnegative matrix factorization (NMF) has been recently applied to solve the hyperspectral unmixing problem because it ensures nonnegativity and needs no assumption for presence of pure pixels. However, algorithm a large amount local minima due obvious nonconvexity objective function. In order improve its performance, auxiliary constraints can be introduced into algorithm. this paper, we propose new approach named abundance separation smoothness constrained NMF by introducing two...

Soft sensor technology is widely used in industries to handle highly nonlinear, dynamic, time-dependent sequence data of industrial processes for predicting the key variables associated with auxiliary process variables. Many existing soft algorithms based on deep learning are able build complex nonlinear models but ignore dynamic characteristics processes. The long short-term memory (LSTM) neural network exploited solve modeling issue, which related strong time-varying features. In this...

In recent years, dimensionality reduction (DR) and classification have become important issues of hyperspectral image analysis. this paper, we propose a new spatial–spectral similarity measure, which maps the distances between two patches in images. Including spatial information by using neighbors, proposed measure is based on fact that observed pixels images are spatially related, meaningful features can be extracted from both spectral domains. First, effectively exploit rich structures...

The geometrical features of airport line segments are seldom used by traditional methods for detection in panchromatic remote sensing images. This letter presents a novel method based on both bottom-up (BU) saliency and top-down saliency. Noticing that runways have vicinity parallelity their lengths among certain range, we introduce the concept near first time treat it as prior knowledge can fully exploit relationship runways. Meanwhile, simplified graph-based visual model is to extract BU...

In the field of connectomics, neuroscientists seek to identify cortical connectivity comprehensively. Neuronal boundary detection from Electron Microscopy (EM) images is often done assist automatic reconstruction neuronal circuit. But segmentation EM a challenging problem, as it requires detector be able detect both filament-like thin and blob-like thick membrane, while suppressing ambiguous intracellular structure. this paper, we propose multi-stage multi-recursiveinput fully convolutional...

For hyperspectral images (HSIs), the imbalance between high dimensionality and limited labeled samples has been a main obstacle to classification task. As solution, semisupervised learning utilizing both unlabeled shown its potential. In this letter, novel framework based on graph attention networks (GATs) for HSIs is proposed. Spatial-spectral joint measurement designed model construction make full use of spatial information. convolution process, different weights are assigned neighboring...

In this paper, a novel hyperspectral anomaly detector based on low-rank representation (LRR) and learned dictionary (LD) has been proposed. This method assumes that two-dimensional matrix transformed from three-dimensional imagery can be decomposed into two parts: low rank representing the background sparse standing for anomalies. The direct application of LRR model is sensitive to tradeoff parameter balances parts. To mitigate problem, introduced decomposition process. whole image with...

Although domain adaptation has been extensively studied in natural image-based segmentation tasks, the research on cross-domain for very-high-resolution (VHR) remote sensing images (RSIs) still remains underexplored. The VHR RSI-based mainly faces two critical challenges: 1) large area land covers with many diverse object categories bring severe local patch-level data distribution deviations, thus yielding different difficulties patches and 2) sensor types or dynamically changing modes cause...

Global channel pruning (GCP) aims to remove a subset of channels (filters) across different layers from deep model without hurting the performance. Previous works focus on either single task or simply adapting it multitask scenario, and still face following problems when handling pruning: 1) Due mismatch, well-pruned backbone for classification focuses preserving filters that can extract category-sensitive information, causing may be useful other tasks pruned during stage; 2) For...

To mitigate the impact of mixed pixels in hyperspectral images (HSIs), substantial progress has been made both model- and deep learning-based unmixing methods. However, issues such as complex computational processes limited interpretability, hinder improvement their performance. Particularly, unsupervised nonlinear (HU) remains a great challenge. In this paper, we propose an extended multilinear mixing (EMLM) model-inspired dual-stream network for HU. Firstly, alternating direction method...

Recently, the autoencoder (AE) has received significant attention in hyperspectral anomaly detection task. However, all existing AE-based detectors operate under linear mixing model, which cannot accurately model nonlinear phenomenon practical images (HSIs). Moreover, these rarely consider spatial information between pixels, is crucial to obtain accurate results of detection. To address above issues, this paper proposes a transformer-based AE framework (TAEF) for Specifically, proposed...

The scarcity of labeled samples has been the main obstacle to development scene classification for remote sensing images. To alleviate this problem, efforts have dedicated semisupervised which exploits both and unlabeled training classifiers. In letter, we propose a novel method that utilizes effective residual convolutional neural network (ResNet) extract preliminary image features. Moreover, strategy ensemble learning (EL) is adopted establish discriminative representations by exploring...

Anomaly detection in hyperspectral imagery has been an active topic among the remote sensing applications. It aims at identifying anomalous targets with different spectra from their surrounding background. Therefore, effective detector should be able to distinguish anomalies, especially for weak ones, background and noise. In this article, we propose a novel method anomaly based on total variation (TV) sparsity regularized decomposition model. This model decomposes into three components:...

The ability to detect strip steel surface defects is crucial for ensuring the quality of items made from that material. A YOLOv5-CD algorithm proposed overcome problems insufficient feature extraction ability, low detection accuracy, and slow convergence rate defect technique in an industrial scenario. First, a coordination attention mechanism added backbone network. This strategy embeds positional information into channel attention, addressing issue loss caused by global pooling effectively...

Hyperspectral target detection (HTD) aims to identifying targets within a hyperspectral image (HSI) based on provided spectra. In the current HTD field, representation-based detectors have attracted much attention. However, there are two prominent challenges that particularly noteworthy. First, background class encompasses diverse land covers, making its accurate representation challenging. Second, ability can be significantly influenced by abnormalities and noise in HSI. To tackle these...

Remote sensing (RS) scene classification plays an important role in the intelligent interpretation of RS data. Recently, CNN-based and attention-based methods have become mainstream with impressive results. However, existing do not utilize long-range information, fully exploit multi-scale although both aspects information are essential for a comprehensive understanding images. To overcome above limitations, we propose progressive feature fusion (PFF) framework based on graph convolutional...

The effectiveness of spectral-spatial feature learning is crucial for the hyperspectral image (HSI) classification task. Diffusion models, as a new class groundbreaking generative have ability to learn both contextual semantics and textual details from distinct timestep dimension, enabling modeling complex relations in HSIs. However, existing diffusion-based HSI methods only utilize manually selected single-timestep single-stage features, limiting full exploration exploitation rich...

Although remote sensing (RS) data with multiple modalities can be used to significantly improve the accuracy of semantic segmentation in RS data, how effectively extract multimodal information through feature fusion remains a challenging task. Specifically, existing methods for still face two major challenges: 1) Due diverse imaging mechanisms boundaries same foreground may vary across different modalities, leading inclusion unwanted background semantics fused features; 2) from exhibit...

Large Language Models (LLMs) demonstrate remarkable capabilities in text generation, yet their emotional consistency and semantic coherence social media contexts remain insufficiently understood. This study investigates how LLMs handle content maintain relationships through continuation response tasks using two open-source models: Gemma Llama. By analyzing climate change discussions from Twitter Reddit, we examine transitions, intensity patterns, similarity between human-authored...

Percutaneous transthoracic puncture of small pulmonary nodules is technically challenging. We developed a novel electromagnetic navigation system for the sub-centimeter lung by combining multiple deep learning models with and spatial localization technologies. compared performance DL-EMNS conventional CT-guided methods in percutaneous punctures using phantom animal models. In study, group showed higher technical success rate (95.6% vs. 77.8%, p = 0.027), smaller error (1.47 ± 1.62 mm 3.98...

Change detection (CD) is a critical task in analyzing the geographic information changes remote sensing images (RSIs), yet it still faces challenges such as complex background interference, multi-scale varying objects, and class imbalance between positive negative samples. Recently, with development of pre-training fine-tuning techniques, transferring general knowledge embedded large-scale pre-trained visual foundation models (PVFMs) to various downstream tasks has attracted significant...