Search

Semantic segmentation using fine-resolution remotely sensed images plays a critical role in many practical applications, such as urban planning, environmental protection, natural and anthropogenic landscape monitoring, etc. However, the automation of semantic segmentation, i.e., automatic categorization/labeling and segmentation is still a challenging task, particularly for fine-resolution images with huge spatial and spectral complexity. Addressing such a problem represents an exciting research field, which paves the way for scene-level landscape pattern analysis and decision making. In this paper, we propose an approach for automatic land segmentation based on the Feature Pyramid Network (FPN). As a classic architecture, FPN can build a feature pyramid with high-level semantics throughout. However, intrinsic defects in feature extraction and fusion hinder FPN from further aggregating more discriminative features. Hence, we propose an Attention Aggregation Module (AAM) to enhance multi-scale feature learning through attention-guided feature aggregation. Based on FPN and AAM, a novel framework named Attention Aggregation Feature Pyramid Network (A2-FPN) is developed for semantic segmentation of fine-resolution remotely sensed images. Extensive experiments conducted on three datasets demonstrate the effectiveness of our A2 -FPN in segmentation accuracy. Code is available at https://github.com/lironui/A2-FPN.

Although the Convolutional Neural Network (CNN) has shown great potential for land cover classification, the frequently used single-scale convolution kernel limits the scope of information extraction. Therefore, we propose a Multi-Scale Fully Convolutional Network (MSFCN) with a multi-scale convolutional kernel as well as a Channel Attention Block (CAB) and a Global Pooling Module (GPM) in this paper to exploit discriminative representations from two-dimensional (2D) satellite images. Meanwhile, to explore the ability of the proposed MSFCN for spatio-temporal images, we expand our MSFCN to three-dimension using three-dimensional (3D) CNN, capable of harnessing each land cover category’s time series interaction from the reshaped spatio-temporal remote sensing images. To verify the effectiveness of the proposed MSFCN, we conduct experiments on two spatial datasets and two spatio-temporal datasets. The proposed MSFCN achieves 60.366% on the WHDLD dataset and 75.127% on the GID dataset in terms of mIoU index while the figures for two spatio-temporal datasets are 87.753% and 77.156%. Extensive comparative experiments and ablation studies demonstrate the effectiveness of the proposed MSFCN. Code will be available at https://github.com/lironui/MSFCN.

The highly dynamic nature of agro-ecosystems in space and time usually leads to high intra-class variance and low inter-class separability in the fine spatial resolution (FSR) remotely sensed imagery. This makes traditional classifiers essentially relying on spectral information for crop mapping from FSR imagery an extremely challenging task. To mine effectively the rich spectral and spatial information in FSR imagery, this paper proposed a Scale Sequence Object-based Convolutional Neural Network (SS-OCNN) that classifies images at the object level by taking segmented objects (crop parcels) as basic units of analysis, thus, ensuring that the boundaries between crop parcels are delineated precisely. These segmented objects were subsequently classified using a CNN model integrated with an automatically generated scale sequence of input patch sizes. This scale sequence can fuse effectively the features learned at different scales by transforming progressively the information extracted at small scales to larger scales. The effectiveness of the SS-OCNN was investigated using two heterogeneous agricultural areas with FSR SAR and optical imagery, respectively. Experimental results revealed that the SS-OCNN consistently achieved the most accurate classification results. The SS-OCNN, thus, provides a new paradigm for crop classification over heterogeneous areas using FSR imagery, and has a wide application prospect.

Microarc oxidation (MAO) coatings were produced on the Al–Cu alloys at different oxidation times. Surface morphologies, phase composition and residual stress, and wear resistance were analysed by S...

This study analyzes how exchange rate risk exposure and foreign industry competition affect Chinese industries. Using the data from 2005 to 2018, the authors find that industry competition from the...

Anhydrosafflor Yellow B (AHSYB), is an effective ingredient with a high content extracted from safflower. The effect of hydroxysafflor yellow A (HSYA), one of the main compounds in safflower, has b...

Cerebral ischemia can trigger the ERK1/2 signaling cascade that enables the brain to adapt to ischemic injury. However, the mechanism of ERK1/2 in ischemic brain injury remains unclear. The aim of this study was to examine the roles of the ERK1/2 signaling pathway and NMDA receptors in the apoptosis of CA1 pyramidal neurons after ischemia/reperfusion (I/R).Male Wistar rats were subjected to a sham or transient forebrain ischemia procedure. Animals received the intracerebroventricular injection of U0126 (5 μl, 0.2 μg/μl) or vehicle 30 min before ischemia. Homogenates of the hippocampal CA1 field were obtained from sham-operated and ischemic rats 6, 12 or 48 h after ischemia/reperfusion (n = 6 per group) and then subjected to Western blotting analysis and TUNEL staining. Caspase-3 activity was assayed with a colorimetric assay kit.We found that the phosphorylation level of ERK1/2 is increased in the CA1 region following transient I/R. Blocking the ERK1/2 signaling pathway by administration U0126 attenuated apoptotic neuronal cell death via inhibition of NMDA receptors.These findings suggest a novel mechanism by which the ERK1/2 signaling pathway affects the post-I/R apoptosis of CA1 pyramidal neurons, which will provide a therapeutic target for the treatment of stroke.

Background: Total bilirubin (TBIL) is known to be inversely associated with coronary heart disease (CHD) risk, however, whether this association is dose-response remains inconsistent and it is unclear which subtype of bilirubin is responsible for the potential protective effect. Methods: We included 12,097 participants who were free of CHD, stroke, cancer and potential liver, biliary and renal diseases at baseline from September 2008 to June 2010 and were followed-up until October 2013. Cox proportional hazards models were used to assess the hazard ratios (HR) and 95% confidence interval (95% CI) of bilirubin with incident CHD risk. Results: The adjusted HRs for incident CHD increased with increasing direct bilirubin (DBIL) (p for trend = .013). Participants within the highest quintile of DBIL had 30% higher risk of incident CHD compared to those in the lowest quintile (95% CI: 1.07, 1.58). In contrast, compared with subjects in the lowest quintile of TBIL, those in the third quintile had the lowest of 24% risk for CHD incidence (95% CI: 0.63, 0.92), which showed a U-shaped association (p for quadratic trend = .040). Conclusions: DBIL was associated with a dose-response increased risk for CHD incidence. However, a U-shaped association existed between TBIL, indirect bilirubin and incident CHD risk.Key messagesDirect bilirubin is independently associated with incident coronary heart disease (CHD) in a dose-response manner.A similarly consistent U-shaped association was found between total bilirubin, indirect bilirubin and incident CHD.The potential protective effect of total bilirubin within the normal range on incident CHD should be mainly attributed to mild-to moderate elevated levels of indirect bilirubin. Direct bilirubin is independently associated with incident coronary heart disease (CHD) in a dose-response manner. A similarly consistent U-shaped association was found between total bilirubin, indirect bilirubin and incident CHD. The potential protective effect of total bilirubin within the normal range on incident CHD should be mainly attributed to mild-to moderate elevated levels of indirect bilirubin.

In this paper, a real-time configurable intelligent property (IP) core is presented for image/video decoding process in compatibility with the standard MPEG-4 Visual and the standard H.264/AVC. The inverse quantised discrete cosine and integer transform can be used to perform inverse quantised discrete cosine transform and inverse quantised inverse integer transforms which only required shift and add operations. Meanwhile, COordinate Rotation DIgital Computer iterations and compensation steps are adjustable in order to compensate for the video compression quality regarding various data throughput. The implementations are embedded in publicly available software XVID Codes 1.2.2 for the standard MPEG-4 Visual and the H.264/AVC reference software JM 16.1, where the experimental results show that the balance between the computational complexity and video compression quality is retained. At the end, FPGA synthesised results show that the proposed IP core can bring advantages to low hardware costs and a...

—In conventional model predictive control, 27 switching states for three-level power converters are evaluated online, and the state in which the cost function is the smallest will be applied in the next sampling period. The amount of calculation, however, is too high, and it may cause too high of a jump for the phase voltage or line voltage in the AC side of the three-level converter, which means that each switching action causes voltage variation to exceed Udc/2. To solve these problems, an improved model predictive method is proposed in this article. The proposed method only needs a subset of all the three-level states, which is obtained by adding and subtracting two-level state groups in the current switching state, for prediction and optimization. The high jump of the AC-side voltage is effectively avoided, and computational effort is greatly reduced. In addition, reduced switching frequency is obtained. The experiments are carried out by the two methods, and the results verify the correctness...

This paper considers the imaging of objects located close to rough surfaces such as ocean or terrain. If transmitters and receivers are also located close to rough surfaces, incident wave is no longer a plane wave nor a spherical wave in free space and it is necessary to consider Green’s functions with the point source located close to the surface, similar to the Sommerfeld dipole problem. This paper considers the near-surface imaging by making use of time-reversal imaging and surface flattening transform. Surface flattening transform converts the rough surface problem into flat surface with inhomogeneous random medium. Mutual coherence function is obtained and used to obtain imaging of point target near rough surface, making use of the multi-static data matrix, time-reversal matrix, the eigenvectors, and the steering matrix. Numerical examples are given. An important point is that integration of stochastic wave propagation and signal processing is necessary to obtain imaging through complex clutter envir...

It was the aim of this study to develop a methodology for dissection in laparoscopic distal gastrectomy with D2 lymphadenectomy (D2 LDG) for gastric cancer. One-hundred and thirty-two patients with distal gastric cancer underwent D2 LDG with a novel sequence of lymph node dissection between August 2004 and June 2008. Live anatomy in each step was observed simultaneously to ensure and confirm the newly developed methodology. Dissections in LDG were standardized as sequential steps: Dividing the gastrocolic ligament and getting access to the prepancreatic space--lymph node dissection in the lower left area--lymph node dissection in the lower right area--lymph node dissection in the upper right area--lymph nodes dissection centrally--lymph node dissection between liver and stomach. All dissections were successfully performed in peripancreatic spaces and their extensions. Gastric vessels were located by special landmarks, traced along vascular trunks and bifurcations, and identified by fine dissection technique in vaginavasorum. Sequential dissection around the pancreas was an effective method for D2 LDG. It was ensured by anatomical knowledge in each step: Vessels and fascial spaces around a central landmark, the pancreas.