- 标题
- 摘要
- 关键词
- 实验方案
- 产品
过滤筛选
- 2018
- Conditional Random Fields (CRF)
- Convolutional Neural Network (CNN)
- Fine Classification
- Airborne hyperspectral
- green tide
- Elegant End-to-End Fully Convolutional Network (E3FCN)
- deep learning
- remote sensing
- Moderate Resolution Imaging Spectroradiometer (MODIS)
- Optoelectronic Information Science and Engineering
- Ocean University of China
- Wuhan University
- Central South University
- Hubei University
-
A CNN With Multiscale Convolution and Diversified Metric for Hyperspectral Image Classification
摘要： Recently, researchers have shown the powerful ability of deep methods with multilayers to extract high-level features and to obtain better performance for hyperspectral image classification. However, a common problem of traditional deep models is that the learned deep models might be suboptimal because of the limited number of training samples, especially for the image with large intraclass variance and low interclass variance. In this paper, novel convolutional neural networks (CNNs) with multiscale convolution (MS-CNNs) are proposed to address this problem by extracting deep multiscale features from the hyperspectral image. Moreover, deep metrics usually accompany with MS-CNNs to improve the representational ability for the hyperspectral image. However, the usual metric learning would make the metric parameters in the learned model tend to behave similarly. This similarity leads to obvious model’s redundancy and, thus, shows negative effects on the description ability of the deep metrics. Traditionally, determinantal point process (DPP) priors, which encourage the learned factors to repulse from one another, can be imposed over these factors to diversify them. Taking advantage of both the MS-CNNs and DPP-based diversity-promoting deep metrics, this paper develops a CNN with multiscale convolution and diversified metric to obtain discriminative features for hyperspectral image classification. Experiments are conducted over four real-world hyperspectral image data sets to show the effectiveness and applicability of the proposed method. Experimental results show that our method is better than original deep models and can produce comparable or even better classification performance in different hyperspectral image data sets with respect to spectral and spectral–spatial features.
关键词： deep metric learning,determinantal point process (DPP),image classification,multiscale features,Convolutional neural network (CNN),hyperspectral image
更新于2025-09-23 15:23:52
-
FPGA-Based Implementation of an Artificial Neural Network for Measurement Acceleration in BOTDA Sensors
摘要： In recent years, using distributed fiber-optic sensors based on Brillouin scattering, for monitoring pipelines, tunnels, and other constructional structures have gained huge popularity. However, these sensors have a low signal-to-noise ratio (SNR), which usually increases their measurement error. To alleviate this issue, ensemble averaging is used which improves the SNR but in return increases the measurement time. Reducing the noise by averaging requires hundreds or thousands of scans of the optical fiber; hence averaging is usually responsible for a large percent of the entire system latency. In this paper, we propose a novel method based on artificial neural network for SNR enhancement and measurement acceleration in distributed fiber-optic sensors based on the Brillouin scattering. Our method takes the noisy Brillouin spectrums and improves their SNR by 20 dB, which reduces the measurement time significantly. It also improves the accuracy of the Brillouin frequency shift estimation process and its latency by more than 50% in comparison with the state-of-the-art software and hardware solutions.
关键词： Artificial neural network (ANN),digital signal processing,optical fibers,curve fitting,field-programmable gate arrays (FPGAs)
更新于2025-09-23 15:23:52
-
[IEEE 2018 IEEE Conference on Antenna Measurements & Applications (CAMA) - Va?ster?s (2018.9.3-2018.9.6)] 2018 IEEE Conference on Antenna Measurements & Applications (CAMA) - Design of Low-profile Antenna with Multi-directional Beam
摘要： An planar antenna having a low-pro?le structure using loop elements is proposed. The proposed antenna is vertically polarized and has four directional beams for sensor nodes. The thickness of the antenna is only 3.2 mm (0.11 effective wavelength) resulting in a low-pro?le structure. Simulated and measured average gain of 8.0 dBi and 5.6 dBi are obtained, respectively.
关键词： wireless sensor network,sensor node,low-pro?le antenna,vertical polarization,multi-directional beam
更新于2025-09-23 15:23:52
-
Estimation of spatiotemporal PM1.0 distributions in China by combining PM2.5 observations with satellite aerosol optical depth
摘要： Particulates smaller than 1.0 μm (PM1.0) have strong associations with public health and environment, and considerable exposure data should be obtained to understand the actual environmental burden. This study presented a PM1.0 estimation strategy based on the generalised regression neural network model. The proposed strategy combined ground-based observations of PM2.5 and satellite-derived aerosol optical depth (AOD) to estimate PM1.0 concentrations in China from July 2015 to June 2017. Results indicated that the PM1.0 estimates agreed well with the ground-based measurements with an R2 of 0.74, root mean square error of 19.0 μg/m3 and mean absolute error of 11.4 μg/m3 as calculated with the tenfold cross-validation method. The diurnal estimation performance displayed remarkable single-peak variation with the highest R2 of 0.80 at noon, and the seasonal estimation performance showed that the proposed method could effectively capture high-pollution events of PM1.0 in winter. Spatially, the most polluted areas were clustered in the North China Plain, where the average estimates presented a bimodal distribution during daytime. In addition, the quality of satellite-derived AOD, the robustness of the interpolation algorithm and the proportion of PM1.0 in PM2.5 were confirmed to affect the estimation accuracy of the proposed model.
关键词： Himawari-8,PM1.0,Neural network,Air pollution,Aerosol optical depth
更新于2025-09-23 15:23:52
-
DeepLens
摘要： We aim to generate high resolution shallow depth-of-field (DoF) images from a single all-in-focus image with controllable focal distance and aperture size. To achieve this, we propose a novel neural network model comprised of a depth prediction module, a lens blur module, and a guided upsampling module. All modules are differentiable and are learned from data. To train our depth prediction module, we collect a dataset of 2462 RGB-D images captured by mobile phones with a dual-lens camera, and use existing segmentation datasets to improve border prediction. We further leverage a synthetic dataset with known depth to supervise the lens blur and guided upsampling modules. The effectiveness of our system and training strategies are verified in the experiments. Our method can generate high-quality shallow DoF images at high resolution, and produces significantly fewer artifacts than the baselines and existing solutions for single image shallow DoF synthesis. Compared with the iPhone portrait mode, which is a state-of-the-art shallow DoF solution based on a dual-lens depth camera, our method generates comparable results, while allowing for greater flexibility to choose focal points and aperture size, and is not limited to one capture setup.
关键词： Neural Network,Shallow Depth of Field
更新于2025-09-23 15:23:52
-
Learning to reconstruct shape and spatially-varying reflectance from a single image
摘要： Reconstructing shape and reflectance properties from images is a highly under-constrained problem, and has previously been addressed by using specialized hardware to capture calibrated data or by assuming known (or highly constrained) shape or reflectance. In contrast, we demonstrate that we can recover non-Lambertian, spatially-varying BRDFs and complex geometry belonging to any arbitrary shape class, from a single RGB image captured under a combination of unknown environment illumination and flash lighting. We achieve this by training a deep neural network to regress shape and reflectance from the image. Our network is able to address this problem because of three novel contributions: first, we build a large-scale dataset of procedurally generated shapes and real-world complex SVBRDFs that approximate real world appearance well. Second, single image inverse rendering requires reasoning at multiple scales, and we propose a cascade network structure that allows this in a tractable manner. Finally, we incorporate an in-network rendering layer that aids the reconstruction task by handling global illumination effects that are important for real-world scenes. Together, these contributions allow us to tackle the entire inverse rendering problem in a holistic manner and produce state-of-the-art results on both synthetic and real data.
关键词： rendering layer,global illumination,deep learning,SVBRDF,single image,flash light,cascade network
更新于2025-09-23 15:23:52
-
Relighting humans
摘要： Relighting of human images has various applications in image synthesis. For relighting, we must infer albedo, shape, and illumination from a human portrait. Previous techniques rely on human faces for this inference, based on spherical harmonics (SH) lighting. However, because they often ignore light occlusion, inferred shapes are biased and relit images are unnaturally bright particularly at hollowed regions such as armpits, crotches, or garment wrinkles. This paper introduces the first attempt to infer light occlusion in the SH formulation directly. Based on supervised learning using convolutional neural networks (CNNs), we infer not only an albedo map, illumination but also a light transport map that encodes occlusion as nine SH coefficients per pixel. The main difficulty in this inference is the lack of training datasets compared to unlimited variations of human portraits. Surprisingly, geometric information including occlusion can be inferred plausibly even with a small dataset of synthesized human figures, by carefully preparing the dataset so that the CNNs can exploit the data coherency. Our method accomplishes more realistic relighting than the occlusion-ignored formulation.
关键词： convolutional neural network,light transport,inverse rendering
更新于2025-09-23 15:23:52
-
Influence of Current Density on Orientation-Controllable Growth and Characteristics of Electrochemically Deposited Au Films
摘要： This paper is concerned with the stability analysis of time varying delayed stochastic Hopfield neural networks in numerical simulation. To achieve our expected conclusions, we will reform the classical contractive mapping principle in functional analysis, with some modifications, to adapt to our conditions and both the continuous and the discrete delayed models. Under the reasonable conditions, it is shown that, the Euler–Maruyama numerical scheme is mean square exponentially stable of exact solution dependent of step size. Further more, it is also shown that the backward Euler–Maruyama numerical scheme can share the mean square exponential stability of the exact solution independent of step size under the same conditions.
关键词： Numerical simulation,Time delay,Stochastic differential equation,Hopfield neural network,Stability
更新于2025-09-23 15:23:52
-
Characteristics of Orbit Determination with Short-Arc Observation by an Optical Tracking Network, OWL-Net
摘要： An optical tracking network, the Optical Wide-field patroL Network (OWL-Net), has been developed to maintain the orbital ephemeris of 11 domestic low Earth orbit satellites. The schedule overlapped events were occurred in the scheduling of the OWL-Net with reduction of the optical observation chances. A short-arc observation strategy for the OWL-Net was tested to reduce schedule overlapped events with the optical observation simulation and the orbit determination. In the full-scale optical observation simulation from January 2014 to December 2016, the most frequent overlapped events were occurred 127, 132, and 116 times in the 4th, 34th, and 18th weeks of 2014, 2015, and 2016, respectively. The average number of overlapped event for three years was over 10% for the whole observation chances of five stations. Consequently, the short-arc observation strategy reduced the schedule overlapped events for every observation target of the OWL-Net. In case of the 5 s and 10 s cases, the most schedule overlapped events were removed. The test results of the orbit determination results show that the most maximum orbit prediction errors after seven days are maintained at <10 km in the in-track direction for the short-arc observation simulations. The results demonstrate that the short-arc optical observation strategy is more optimal to maintaining the accuracy of orbital ephemeris with more observation chances.
关键词： short-arc observation,optical tracking network,OWL-Net,orbit determination,LEO satellites
更新于2025-09-23 15:23:52
-
Multiple deep-belief-network-based spectral-spatial classification of hyperspectral images
摘要： A deep-learning-based feature extraction has recently been proposed for HyperSpectral Images (HSI) classification. A Deep Belief Network (DBN), as part of deep learning, has been used in HSI classification for deep and abstract feature extraction. However, DBN has to simultaneously deal with hundreds of features from the HSI hyper-cube, which results into complexity and leads to limited feature abstraction and performance in the presence of limited training data. Moreover, a dimensional-reduction-based solution to this issue results in the loss of valuable spectral information, thereby affecting classification performance. To address the issue, this paper presents a Spectral-Adaptive Segmented DBN (SAS-DBN) for spectral-spatial HSI classification that exploits the deep abstract features by segmenting the original spectral bands into small sets/groups of related spectral bands and processing each group separately by using local DBNs. Furthermore, spatial features are also incorporated by first applying hyper-segmentation on the HSI. These results improved data abstraction with reduced complexity and enhanced the performance of HSI classification. Local application of DBN-based feature extraction to each group of bands reduces the computational complexity and results in better feature extraction improving classification accuracy. In general, exploiting spectral features effectively through a segmented-DBN process and spatial features through hyper-segmentation and integration of spectral and spatial features for HSI classification has a major effect on the performance of HSI classification. Experimental evaluation of the proposed technique on well-known HSI standard data sets with different contexts and resolutions establishes the efficacy of the proposed techniques, wherein the results are comparable to several recently proposed HSI classification techniques.
关键词： hyperspectral image classification,support vector machine,deep belief network,segmentation
更新于2025-09-23 15:23:52