It is aware that big data has gathered tremendous attentions from academic research institutes, governments, and enterprises in all aspects of information sciences. With the development diversity marine acquisition techniques, grow exponentially last decade, which forms . As an innovation, a double-edged sword. On one hand, there are many potential highly useful values hidden huge volume data, widely used marine-related fields, such as tsunami red-tide warning, prevention, forecasting,...

Multivariate Time Series Classification (MTSC) is a challenging task in real-world applications. Current approaches emphasize modeling multiscale relationships over time. However, the (MTS) also exhibits cross-channel relationships. Furthermore, long-term temporal time series are difficult to capture. In this paper, we introduce MD-Former, Multiscale Dual-Branch Attention network leveraging Transformer architecture capture across and channels for MTSC. MTS embedded into 2D vectors using...

The recent advances in remote sensing and computer techniques give birth to the explosive growth of images. emergence cloud storage has brought new opportunities for management massive images with its large space, cost savings. However, openness brings challenges image data security. In this paper, we propose a weighted sharing scheme ensure security environment, which takes weights participants (i.e., service providers) into consideration. An extended Mignotte sequence is constructed...

The Sea Surface Temperature (SST) is one of the key factors affecting ocean climate change. Hence, Prediction (SSTP) great significance to study navigation and meteorology. However, SST data well-known suffer from high levels redundant information, which makes it very difficult realize accurate predictions, for instance when using time-series regression. This paper constructs a simple yet effective SSTP model, dubbed DSL (given its origination methods known as DTW, SVM LSPSO). based on...

Automatic detection of mesoscale oceanic eddies is in great demand to monitor their dynamics which play a significant role ocean current circulation and marine climate change. Traditional methods using remotely sensed data are usually based on physical parameters, geometrics, handcrafted features or expert knowledge, they face challenge accuracy efficiency due the high variability our limited understanding process, especially for rich large data. In this paper, we propose simple deep...

Anticyclonic eddies are widely distributed off Peru, producing obvious effects on climate-environment-sensitive jumbo flying squid Dosidicus gigas . However, how the habitat of D. changes within anticyclonic during El Niño events is still unknown. Therefore, this study constructed a model with different weights and screened optimal performance using fishery data from September to December 2015 2019, as well vertical water temperature (sea surface 50 m depth temperature) chlorophyll...

Accurately predicting the trajectories of mesoscale eddies is essential for comprehending distribution marine resources and multiscale energy cascade in ocean. Nevertheless, current approaches eddy frequently exhibit inadequate examination intrinsic temporal data, resulting diminished predictive precision. To address this challenge, our research introduces an enhanced transformer-based framework trajectories. Initially, a multivariate dataset constructed expanded, encompassing properties...

Speech enhancement aims to make noisy speech signals clearer. Traditional time-frequency domain methods struggle differentiate between and noise, leading a risk of distortion. This paper introduces an approach that combines the time using W-net module suppress noise at front end. The is improved version Wave-U-Net, called TTF-W-Net. We conducted experiments TIMIT NOISEX-92 datasets evaluate performance achieved by integrating preprocessing networks, specifically Wave-U-Net our TTF-W-Net,...

The problem of heterogeneous data, caused by the differences marine data acquisition equipments, information processing platforms, storage formats, makes integration, exchange and share difficult. Nowadays, semantics in integration is even more severe. Aim at solving this problem, we introduce an ontology-based automatic ETL (Extract, Transform Load) method. This method can be achieved automatically as follows:Firstly, extracting structures from source sets, secondly, achieving match sets to...

Automatic recognition of ocean eddies has become one the hotspots in field physical oceanography. Traditional methods based on either parameters or geometry features require manual parameter adjustment, and cannot adapt to dynamic changes caused by complicated environments. To address these issues, we propose a new eddy method SAR images with adaptive weighted multi-feature fusion. Specially, better characterize eddies, first extract texture, shape corner using global Gray Level...

With the development of remote sensing technology, content-based image retrieval has become a research hotspot. Remote datasets not only contain rich location, semantic and scale information but also have large intra-class differences. Therefore, key to improving performance is make full use limited sample extract more comprehensive class features. In this paper, we propose proxy-based deep metric learning method an adaptive multi-proxy framework. First, intra-cluster synthesis strategy with...

Due to the variations of aircraft types, sizes, orientations, and complexity remote sensing images, it is still difficult effectively obtain accurate position type by detection, which plays an important role in intelligent air transportation digital battlefield. Current detection methods often use horizontal detectors, produce significant redundancy, nesting, overlap areas negatively affect performance. To address these difficulties, a framework based on RetinaNet that combines multi-feature...

目的 海洋涡旋精准检测是揭示海洋涡旋演变规律及其与其他海洋现象相互作用的基础。然而，海洋涡旋在其活跃海域呈现小尺度目标、密集分布的特点，导致显著的检测精度低问题。传统方法受限于人工设计参数缺乏泛化能力，而深度学习模型的高采样率在检测小目标过程中底层细节和轮廓等信息损失严重，使得目标检测轮廓与目标真实轮廓相差甚远。针对海洋涡旋小目标特点导致检测精度低，高采样率深度模型检测轮廓不精确的问题，提出一种改进的U-Net网络。方法 该模型基于渐进式采样结构，为获取上下文信息提升不同极性海洋涡旋目标的检测精度，增加上下文特征融合模块；为增加该模块对海洋涡旋小目标的关注，在特征融合前对最底层特征嵌入残差注意力模块，使模型可以更多关注海洋涡旋的轮廓信息。最后引入数据扩充方法缓解模型存在的过拟合问题。结果 本文以南大西洋的卫星海表面高度数据集开展实验，结果表明，本文模型检测准确率达到了93.24%，同时在海洋涡旋的检测数量上与真实结果更加接近，验证了模型在小目标检测方面的性能更加优秀。结论...

With the establishment of marine monitoring network with an all-round marine, land, air and space, quality data showed a trend exponential growth from GB, TB to PB. How achieve high scalability, reliability, security low-cost storage massive has become one key issues restricting development "digital ocean" applications data. Based on characteristics data(mass, timeliness, sensitivity, regionalism dynamics), this paper designed hybrid cloud solution in view performance, private large capacity...

To discover the spatial–temporal patterns of sea surface temperature (SST) in South China Sea (SCS), this paper proposes a co-clustering algorithm optimized by information divergence. This method allows for clustering SST data simultaneously across temporal and spatial dimensions is adaptable to large volumes anomalous situations. First, are initially clustered using algorithm. Second, we use divergence as loss function refine results iteratively. During iterative optimization results, treat...

Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI

Oceanic trajectories frequently exhibit multiple periodic patterns across various time intervals, e.g., tidal variations, mesoscale eddies, and El Niño events correspond to diurnal, seasonal, interannual fluctuations in environmental factors. To explore hidden spatiotemporal behaviors noisy ocean data, we propose a novel trajectory clustering method, namely DTID-STFC. It first identifies dense intervals (DTIs) which occur frequently. Subsequently, within each DTI, it utilizes spectral...

Recently, multiple and heterogeneous remote sensing images have provided a new development opportunity for Earth observation research. utilizing deep learning to gain the shared representative information between different modalities is important resolve problem of geographical region classification. In this paper, CNN-based multi-modal framework low-sample-size data classification introduced. This method has three main stages. Firstly, features are extracted from high- low-resolution...

Abstract. Space syntax and distance measurement are significant theoretical methods for studying road accessibility. deals with the topological relationship between networks, but length of is ignored. While method only takes into account factors such as distance, time, economic costs when Both them defective in evaluating accessibility transportation there a few researches that consider interaction axes lines. In this paper traffic network evaluation proposed by combining space theory...