Due to advances in remote sensing satellite imaging and image processing technologies their wide applications, intelligent satellites are facing an opportunity for rapid development. The key technologies, standards, laws of also experiencing a series new challenges. Novel concepts the hyperspectral system have been proposed since 2011. aim these is provide real-time, accurate, personalized information services. This article reviews current developments new-generation systems, with focus on...

The purpose of hyperspectral anomaly detection is to distinguish abnormal objects from the surrounding background. In actual scenes, however, complexity ground objects, high-dimensionality data and non-linear correlation bands have high requirements for generalizability, feature extraction ability nonlinear expression algorithms. order address above problems, we propose a transferable network with Siamese architecture image (TSN-HAD). contribution TSN-HAD three-fold. First, problem through...

In the traditional signal model, is assumed to be deterministic, and noise random, additive uncorrelated component. A hyperspectral image has high spatial spectral correlation, a pixel can well predicted using its and/or neighbors; any prediction error considered from noise. Using this concept, several algorithms have been developed for estimation images. However, these not rigorously analyzed with unified scheme. paper, we conduct comparative study such linear regression-based simulated...

Monitoring chlorophyll a (CHLA) by remote sensing is particularly challenging for turbid productive waters. Although several empirical and semianalytical algorithms have been developed such waters, their accuracy varies significantly due to variability in optical properties. In this paper, we evaluated the performance of six CHLA concentration (C <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">chla</sub> ) estimation [e.g., two-band ratio...

Object detection is a focal point in remote sensing applications. Remote images typically contain large number of small objects and wide range orientations across objects. This results great challenges to object approaches based on images. Methods directly employ channel relations with equal weights construct information features leads inadequate feature representation complex image tasks. Multiscale methods improve the speed accuracy detection, while themselves limited information, are...

Target detection is a critical task in interpreting aerial images. Small target detection, such as vehicles, challenging. Different lighting conditions affect the accuracy of vehicle detection. For example, vehicles are difficult to distinguish from background RGB images under low illumination conditions. In contrast, high conditions, color and texture not significantly different thermal infrared (TIR) To improve various we propose an adaptive multi-modal feature fusion cross-modal index...

Onboard processing systems are becoming very important in remote sensing data processing. However, a main problem with specialized hardware architectures used for onboard is their high power consumption, which limits exploitation earth observation missions. In this paper, novel strategy approximate computing proposed reducing energy consumption remotely sensed tasks. As case study, the implementation of support vector machine (SVM) hyperspectral image classification considered by using...

Automatically extracting buildings with high precision from remote sensing images is crucial for various applications. Due to their distinct imaging modalities and complementary characteristics, optical synthetic aperture radar (SAR) serve as primary data sources this task. We propose a novel Boundary-Link Multimodal Fusion Network (BLMFNet) joint semantic segmentation leverage the information in these images. An initial building extraction result obtained multimodal fusion network, followed...

Object detection based on remote sensing imagery has become increasingly popular over the past few years. Unlike natural images taken by humans or surveillance cameras, scale of is large, which requires training and inference procedure to be a cutting image. However, objects appearing in are often sparsely distributed labels for each class imbalanced. This results unstable inference. In this paper, we analyze characteristics propose fusion aggregated-mosaic method, with assigned-stitch...

The nascent graph representation learning has shown superiority for resolving data. Compared to conventional convolutional neural networks, graph-based deep the advantages of illustrating class boundaries and modeling feature relationships. Faced with hyperspectral image (HSI) classification, priority problem might be how convert data into irregular domains from regular grids. In this regard, we present a novel method that performs localized filtering on HSIs based spectral theory. First,...

Accurate reorientation and segmentation of the left ventricular (LV) is essential for quantitative analysis myocardial perfusion imaging (MPI). This study proposes an end-to-end model, named as Multi-Scale Spatial Transformer UNet (MS-ST-UNet), which involves multi-scale spatial transformer network (MSSTN) (MSUNet) modules to perform simultaneous LV region from nuclear cardiac images. The sampler produces images with varying resolutions, while scale (ST) blocks are employed align scales...

Deep convolution networks have been widely used in remote sensing target detection for various applications recent years. Target models with many parameters provide better results but are not suitable resource-constrained devices due to their high computational cost and storage requirements. Furthermore, current lightweight imagery rarely the advantages of existing models. Knowledge distillation can improve learning ability a small student network from large teacher acceleration compression....

Target detection is a critical task in interpreting hyperspectral remote sensing images. Small target (such as airplanes) challenging, especially large-scale complex scenes with high spectral variability of different land cover types. In this paper, we propose transfer learning-based, large-scale, robust detector (TLH <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> TD) to improve the accuracy (HTD) scenes. TLH TD learns spatial features...

We present a parallel implementation of the optimized maximum noise fraction (G-OMNF) transform algorithm for feature extraction hyperspectral images on commodity graphics processing units (GPUs). The proposed approach explored data-level concurrency and computing flow. first defined three-dimensional grid, in which each thread calculates sub-block data to easily facilitate spatial spectral neighborhood searches estimation, is one most important steps involved OMNF. Then, we flow computed...

Dichotomous image segmentation (DIS) has a wide range of real-world applications and gained increasing research attention in recent years. In this paper, we propose to tackle DIS with informative frequency priors. Our model, called FP-DIS, stems from the fact that prior knowledge domain can provide valuable cues identify fine-grained object boundaries. Specifically, generator jointly utilize fixed filter learnable filters extract Before embedding priors into network, first harmonize...

Anterior cruciate ligament (ACL) tears are prevalent knee injures, particularly among active individuals. Accurate and timely diagnosis is essential for determining the optimal treatment strategy assessing patient prognosis. Various previous studies have demonstrated successful application of deep learning techniques in field medical image analysis. This study aimed to develop a model detecting ACL magnetic resonance Imaging (MRI) enhance diagnostic accuracy efficiency. The proposed consists...

Urban morphology and change their impacts on urban transportation have been studied extensively in planar space. The essential feature of space, however, is its three-dimensionality (3D), few studies conducted from a 3D perspective, overly limiting the accuracy relationships between transportation. aim this paper to simulate morphologies under Digital Earth framework. On basis principle that population distribution movement are largely confined by morphologies, which affect transportation,...

We propose a commodity graphics processing units (GPUs)–based massively parallel efficient computation for spectral-spatial classification of hyperspectral images. The framework is based on the marginal probability distribution which uses all information in data. In this framework, first, posterior class modeled with discriminative random field association potential linked multinomial logistic regression (MLR) classifier and interaction modeling spatial to Markov multilevel (MLL) prior....

AbstractRemote sensing techniques can offer powerful tools for measuring concentrations of chlorophyll-a (chl-a), which is an important proxy water quality. However, remote estimates chl-a be difficult in bodies that have high levels total suspended matter (TSM). In this study, we examined the applicability synthetic chlorophyll index (SCI) and a parameter relevant to pigments (Hchl) used conjunction with remote-sensing data predict (Cchl-a) Taihu Lake, highly turbid hypereutrophic lake...

Anomaly detection is one of the most important techniques for remotely sensed hyperspectral data interpretation. Developing fast processing anomaly has received considerable attention in recent years, especially analysis scenarios with real-time constraints. In this paper, we develop an embedded graphics units based parallel computation streaming background statistics algorithm. The method can simulate detection, which refer to that be performed at same time as are collected. algorithm...