Anomaly detection has become an important remote sensing application due to the abundant spectral and spatial information contained in hyperspectral images. Recently, anomaly methods based on collaborative representation model have attracted significant attention. Nevertheless, these face two main challenges: (1) all features (spectral signatures) are constrained share same coefficient, which ignores differences among features; (2) existing dictionaries for pixel-by-pixel usually not...

Hyperspectral unmixing is a vibrant research field that focuses on the task of decomposing mixed pixels into collection pure spectral signatures, known as endmembers, along with their corresponding fractional abundances. Conventional algorithms often need to combine two techniques, namely endmember extraction and abundance estimation, accomplish task. Recently, deep learning (DL) has succeeded in hyperspectral due its strong feature data-fitting capabilities. By extracting output weight...

Fast and accurate remote sensing image retrieval from large data archives has been an important research topic in the literature. Recently, hashing-based attracted extreme attention because of its efficient search capabilities. Especially, deep hashing algorithms have developed based on convolutional neural networks (CNNs) shown effective performance. However, implementing a network tends to be highly expensive terms storage space computing resources suitable for on-orbit retrieval, which...

The storage and processing of remotely sensed hyperspectral images (HSIs) is facing important challenges due to the computational requirements involved in analysis these images, characterized by continuous narrow spectral channels. Although HSIs offer many opportunities for accurately modeling mapping surface Earth a wide range applications, they comprise massive data cubes. These huge amounts impose from points view. support vector machine (SVM) has been one most powerful learning...

Convolutional neural networks (CNNs) have proven to be a powerful tool for the classification of hyperspectral images (HSIs). The CNN kernels are able naturally include spatial information smooth out spectral variability and noise present in HSI data. However, these composed large number learning parameters that must correctly adjusted achieve good performance. This forces model consume amount training data, being prone overfitting when limited labeled samples available. In addition,...

Hyperspectral unmixing extracts pure spectral constituents (endmembers) and their corresponding abundance fractions from remotely sensed scenes. Most traditional hyperspectral methods require the results of other endmember extraction algorithms to complete estimation step. Due impressive learning data fitting capabilities convolutional neural networks (CNNs), deep (DL)-based technologies have rapidly developed in literature. According procedure used combine different layers (i.e., fully...

To address the pressing challenges of quality and sustainability in agricultural product supply chains, this paper proposes a multi-stakeholder collaborative governance framework. Adopting perspective sustainability, develops an evolutionary game model Chinese chain. This involves four key stakeholders: enterprises, government, NGOs, consumers. It integrates principles to ensure that decisions each stakeholder contribute safety products while also promoting long-term environmental social...

Hyperspectral images (HSIs) comprise plenty of information in the spatial and spectral domain, which is highly beneficial for performing classification tasks a very accurate way. Recently, attention mechanisms have been widely used HSI due to their ability extract relevant features. Notwithstanding positive results, most attentional strategies usually introduce significant number parameters be trained, making models more complex increasing computational load. In this paper, we develop new...

Spectral unmixing plays a vital role in hyperspectral image analysis. It mainly consists of two procedures, i.e., endmember extraction and abundance estimation. Although most algorithms for each the procedures may exhibit good performance, few studies have been done considering both problems simultaneously. Therefore, accuracy is normally achieved by exploring all possible combinations types algorithms, which renders high computational overloads. We propose novel orthogonal projection...

Autoencoder-based hyperspectral anomaly detectors have received significant attention. The core of these is to reconstruct backgrounds by optimizing autoencoders so that anomalies can be detected reconstruction residuals. Nevertheless, existing methods are flawed in two aspects: 1) most them the background along with anomalies, resulting undesired performance for large target detection complex backgrounds; 2) they only focus on encoder optimization part, ignoring decoder quality background....

Automatically detecting macroalgae from optical remote sensing images has been a long-standing problem in ocean monitoring. Existing operational spectral-based techniques tend to assign threshold an indicator such as Normalized Difference Vegetation Index (NDVI) determine whether or not exists pixel. This thresholding scheme relies on individual pixel features and cannot address the environment variability among different even across one single image. To these limitations, we develop...

Most existing endmember extraction techniques require prior knowledge about the number of endmembers in a hyperspectral image. The is normally estimated by separate procedure, whose accuracy has large influence on performance. In order to bridge two seemingly independent but, fact, highly correlated procedures, we develop new estimation strategy that simultaneously counts and extracts endmembers. We consider image as pixel set define subset pixels are most different from one another...

With the recent development of remote sensing technology, large image repositories have been collected. In order to retrieve desired images massive data sets effectively and efficiently, we propose a novel central cohesion gradual hashing (CCGH) mechanism for retrieval. First, design deep model based on ResNet-18 which has shallow architecture extracts features imagery efficiently. Then, new training by minimizing loss guarantees that remote-sensing hash codes are as close their code centers...

Endmember estimation plays a key role in hyperspectral image unmixing, often requiring an of the number endmembers and extracting endmembers. However, most existing extraction algorithms require prior knowledge regarding endmembers, being critical process during unmixing. To bridge this, new maximum distance analysis (MDA) method is proposed that simultaneously estimates spectral signatures without any information on experimental data containing pure pixel no noise, based assumption form...

We present a cofactor-based endmember extraction strategy for estimating green algae area in geostationary ocean color imager multispectral images. Our improves the efficiency of widely used N-FINDR method from two aspects. First, our exploits cofactor matrix searching largest simplex volume, which just computes inverse and determinant small number times (or even once). This is more efficient than enumeration determinants all pixels N-FINDR. Second, empirically obtains optimal endmembers...

We explore the use of graphical generative adversarial networks (Graphical-GAN) for synthesizing remote sensing images. The model is probabilistic based (GAN). It pairs a network G with recognition R. Both them are adversarially trained discriminative D. Particularly, R employed to infer underlying causal relationships among both observed and latent variables from real advantages Graphical-GAN multiple categories images two fold. Firstly, it considers captures true data distribution...

Endmember estimation consists of two tasks, that is, determining the number pure spectral constituents (endmembers) and extracting their signatures. We present a new geometric distance-based method for endmember from hyperspectral images (HSIs), which does not need to know endmembers in advance. Our strategy optimizes widely used maximum distance analysis (MDA) viewpoints. First, traditional MDA performs by computing distances between any pixel one specific pixel, line, plane, or affine hull...

We present a novel endmember extraction method for estimating green algae areas in Geostationary Ocean Color Imager (GOCI) multispectral images. commence by conducting spectral dimension reduction via the locality preserving projection (LPP) that captures image spatial local characteristics. The LPP enables each projected pixel to encode information around original pixel. then exploit minimum noise fraction (MNF) reducing effects from pixels. apply N-FINDR strategy pixels extraction. compute...

Deep learning-based anomaly detectors are receiving increasing attention, with the black box represented by neural networks being an ever-present puzzle. Recently, interpretable deep network (LRR-Net) –the first model-driven in field of hyperspectral detection–provided inspiring endeavor. However, this work is incomplete, as it separates dictionary construction from detection, leading to a mismatch problem. To address issue, paper proposes unrolling active learning, aiming integrate into...

During the past decades, volume of big data available in remote sensing (RS) applications has grown significantly. In addition, a number related to monitoring human activity are being developed based on this kind data. This considerably increased demand for RS processing methods. sense, scientific community is facing challenge how maximize potential that produced fast and efficient way. particular, provision algorithms can be an easy way fundamental problem community, due large offered by...