Search

Rain streaks showing in images or videos would severely degrade the performance of computer vision applications. Thus, it is of vital importance to remove rain streaks and facilitate our vision systems. While recent convolutinal neural network based methods have shown promising results in single image rain removal (SIRR), they fail to effectively capture long-range location dependencies or aggregate convolutional channel information simultaneously. However, as SIRR is a highly illposed problem, these spatial and channel information are very important clues to solve SIRR. First, spatial information could help our model to understand the image context by gathering long-range dependency location information hidden in the image. Second, aggregating channels could help our model to concentrate on channels more related to image background instead of rain streaks. In this paper, we propose a non-local channel aggregation network (NCANet) to address the SIRR problem. NCANet models 2D rainy images as sequences of vectors in three directions, namely vertical direction, transverse direction and channel direction. Recurrently aggregating information from all three directions enables our model to capture the long-range dependencies in both channels and spaitials locations. Extensive experiments on both heavy and light rain image data sets demonstrate the effectiveness of the proposed NCANet model.

In this study, by employing the intra-pulse cross-correlation (IPCC) operation, an unambiguous velocity estimation method is proposed for narrow-band long-range radars with high carrier frequency and low pulse repetition frequency. This estimation algorithm is simple and could be easily implemented in existing radar systems without changing the radar hardware or the pulse transmitting scheme. Comparing with the slow time dimension correlation algorithm, the accuracy of the proposed intra-pulse frequency domain method is greatly improved, and the brute-force search for the unknown motion parameters is also unnecessary. By first setting a small frequency offset of the IPCC operation, the unambiguous velocity region could be significantly enlarged. Using the relatively coarse but unambiguous estimates and increasing the frequency offset step by step, the IPCC is repeatedly applied to obtain more accurate estimates. Note that the estimation results of the IPCC algorithm could be used in the maximum-likelihood estimator for ambiguity resolution. The Cramer-Rao bound for the proposed algorithm is derived, and the optimal frequency offset in the sense of estimation accuracy is also analysed. Through numerical simulations for both synthetic and real radar data, the effectiveness of the proposed estimation algorithm is verified.

For maneuvering targets, their motion during long observing time will deteriorate the integration results and degrade the performance of range and motion parameters estimation. To solve this problem, a computationally efficient method based on the time-reversal (TR) process and maximum likelihood (ML) principle, i.e., TR-MLE is proposed. The proposed method decouples the joint parameter estimation problem into two simpler problems, which not only increases the efficiency but also improves the estimation performance in low signal-noise-ratio. Furthermore, a fast method using Chirp-Z transform and Newton's method is developed for a more efficient implementation. The theoretical analysis of the noise properties after the TR process is carried out. Then, the corresponding Cramer–Rao lower bound that can evaluate the performance loss introduced by the TR process is discussed in detail. Simulated data and real data are used to assess the performance of the proposed method.

A coupling extended multiscale finite element and peridynamic method is developed for the quasi-static mechanical analysis of large-scale structures with crack propagation. Firstly, a novel incremental peridynamic (PD) formulation based on the ordinary state-based PD model is derived utilizing the Taylor expansion technique. To combine the high computational efficiency of the EMsFEM and advantages of dealing with discontinuous problems of the PD, a coupling strategy based on the numerical base function is proposed, in which the displacement constraint relationships between the coarse element nodes of the EMsFEM and the material points of the PD among the coupling domain are constructed by the numerical base functions and are represented by a coupling strain energy function using the Lagrange multiplier method. Then, a bilinear softening material model is adopted to describe the damage and failure of the bond, and the incremental-iterative algorithms are applied to obtain the steady-state solutions. Finally, several representative numerical examples are presented, and the results demonstrate the accuracy and efficiency of the proposed coupling method for the quasi-static mechanical analysis of large-scale structures with crack propagation. Comparing with the single EMsFEM and PD method, the present coupling method can reduce much computational cost and well deal with crack problems, simultaneously.

The progress brought by the deep learning technology over the last decade has inspired many research domains, such as radar signal processing, speech and audio recognition, etc., to apply it to their respective problems. Most of the prominent deep learning models exploit data representations acquired with either Lidar or camera sensors, leaving automotive radars rarely used. This is despite the vital potential of radars in adverse weather conditions, as well as their ability to simultaneously measure an object’s range and radial velocity seamlessly. As radar signals have not been exploited very much so far, there is a lack of available benchmark data. However, recently, there has been a lot of interest in applying radar data as input to various deep learning algorithms, as more datasets are being provided. To this end, this paper presents a survey of various deep learning approaches processing radar signals to accomplish some significant tasks in an autonomous driving application, such as detection and classification. We have itemized the review based on different radar signal representations, as it is one of the critical aspects while using radar data with deep learning models. Furthermore, we give an extensive review of the recent deep learning-based multi-sensor fusion models exploiting radar signals and camera images for object detection tasks. We then provide a summary of the available datasets containing radar data. Finally, we discuss the gaps and important innovations in the reviewed papers and highlight some possible future research prospects.

Millimeter-wave radar has proven to have a good range estimation accuracy and is less influenced by weather conditions. However, it is difficult for radar to recognize objects, and it is prone to cause a false alarm. In this paper, we present an object detection and classification by jointly using a radar and camera sensors for traffic surveillance applications. The proposed method fuses the Regions of Interest (ROIs) generated on each of the detection results obtained independently from radar and camera sensors. Reducing the high false alarm of a radar sensor is the main aim of the fusion method. Then, a Convolutional Neural Network (CNN) is used to classify the final fused detected objects into one of the six-vehicle categories; Sedan, Truck, Minivan, Bus, Microbus, and SUV. The proposed method was verified using real data. Results obtained demonstrate the good performance of the proposed fusion approach in traffic surveillance context.

Synthetic Aperture Radar (SAR) Imaging and Time of Flight (ToF) Imaging are two popular approaches for terahertz (THz) 3D imaging nowadays. Both systems above require broadband operations and coherent detection, which requires complicated and costly hardware. Generally, 3D imaging is difficult for those objects with large-curvature features in the case of both optical and THz systems. In this paper, we propose a simple but effective 3D imaging approach, which works in the point-cloud imaging mode. Our in-house THz lenses can focus the THz wave very well and it can gather reflected power within a large angle, thus making the decent 3D imaging feasible. By introducing the entropy concept for threshold optimization, decent figures can be obtained in case of poor raw data with a low signal-to-noise (SNR) ratio. In experiments, our system outperforms the commercial Terahertz Time-Domain Spectroscopy (THz-TDS) system. The proposed method could benefit industrial production, heritage studies, etc.

Fatigue is a significant degradation mode that affects nuclear power plants around the world.A fewself-designed computer codes have been developed forfatigue analysis. The Nuclear Power Institute of China has finalized an in-house computer code called FAC_NPIC for evaluating fatigue according to international standards. FAC_NPIC extracts stress data from finite element analysis results or manual input and is independent of specific finite element software. The “rubberband” peak-and-valley detection algorithm searches for extreme stress values automatically, and the modified rainflow-3D algorithm captures the features of secondary fluctuations. This paper presents transient combination schemes and how seismic loading effects are handled, as well asexamples of input and output files. A simple benchmark calculationwasconducted and compared with the literature. The results showed that all key features of FAC_NPIC work well.Despite differences in fatigue algorithms, good agreement with thestandards was achieved.

The sea clutter modeling is critical to the radar design and assessment of relevant detection algorithms. In this paper, we investigate a family of generalized autoregressive conditional heteroscedastic (GARCH) processes to model the sea clutter as a time series, in which the current variance is dependent on historical information. The most general model (so-called the ALLGARCH model) provides more flexible variance structures to model non-Gaussian, asymmetry, and nonlinear properties of the clutter. However, after going through the usage of the ALLGARCH model, we find that it is not very suitable because the coefficients of the model, which are numerous, would be difficult to estimate in a real-time operating environment. Meanwhile, we find that some of the coefficients are negligible under almost all kinds of sea environments and weather conditions. Motivated by these observations, we propose a novel GARCH model for sea clutter modeling, which is a generalization of the nonlinear-asymmetric GARCH (NAGARCH) model. Considering the correlation between adjacent clutter returns, autoregressive terms are also introduced. By systematically analyzing practical sea clutter data under different sea environments, we demonstrate that the proposed model achieves comparable fitting effect to some commonly used statistical models. Also, we develop the corresponding generalized likelihood ratio test (GLRT) algorithm for the new model. Numerical simulations exhibit that the proposed detector achieves higher probability of detection, comparing with the AR-GARCH detector.

This paper is dedicated to designing the overall structural topology for the lightweight design of an automobile wheel. A simplified two-dimensional finite element analysis (FEA) model for the wheel is established, in which the whole wheel structure is first defined as design domain during topology optimization. A rotationally periodic constraint is introduced to design the wheel into structural topology consisting of rotationally repetitive modules. Further, compliance-based topological design under multiple load cases within single module is carried out. In order to achieve a uniform deflection and stiffness distribution around the circumference of wheel, a weighted compliance under multiple load cases is taken as the objective function. In addition, some factors significantly affecting the structural topology are discussed.

In order to simulate the fracture process of particle-reinforced materials on the micro-scale, an image-based double-smoothing cohesive finite element framework is proposed in the present paper. Two separate smoothing processes are performed to reduce the noise in the digital image and eliminate the jagged elements in the finite element mesh. The main contribution of the present study is the proposed novel image-based cohesive finite element framework, and this method improved the quality of the meshes effectively. Meanwhile, the artificial resistance due to the jagged element is reduced with the double-smoothing cohesive finite element framework during the crack propagation. Therefore, the image-based double-smoothing cohesive finite element framework is significant for the simulation of fracture mechanics.

National Natural Science Foundation of China [61102135]; Natural Science Foundation of Fujian Province of China [2011J01372]; High Technology Research and Development Programme of Fujian Province of China [2010HZ0004-1, 2009HZ0003-1]; Fundamental Research Funds for the Central Universities, People's Republic of China [2010121063]

Traditional monopulse angle estimations are mainly based on phase comparison and amplitude comparison methods, which are commonly adopted in narrowband radars. In modern radar systems, wideband radars are becoming more and more important, while the angle estimation for wideband signals is little studied in previous works. As noise in wideband radars has larger bandwidth than narrowband radars, the challenge lies in the accumulation of energy from the high resolution range profile (HRRP) of monopulse. In wideband radars, linear frequency modulated (LFM) signals are frequently utilized. In this paper, we investigate the monopulse angle estimation problem for wideband LFM signals. To accumulate the energy of the received echo signals from different scatterers of a target, we propose utilizing a cross-correlation operation, which can achieve a good performance in low signal-to-noise ratio (SNR) conditions. In the proposed algorithm, the problem of angle estimation is converted to estimating the frequency of the cross-correlation function (CCF). Experimental results demonstrate the similar performance of the proposed algorithm compared with the traditional amplitude comparison method. It means that the proposed method for angle estimation can be adopted. When adopting the proposed method, future radars may only need wideband signals for both tracking and imaging, which can greatly increase the data rate and strengthen the capability of anti-jamming. More importantly, the estimated angle will not become ambiguous under an arbitrary angle, which can significantly extend the estimated angle range in wideband radars.

Hand gesture-based interaction provides a natural and powerful means for human-computer interaction. It is also a good interface for human-robot interaction. However, most of the existing proposals are likely to fail when they meet some skin-coloured objects, especially the face region. In this paper, we present a novel hand tracking method which can track the features of the hand based on the obstacle avoidance flocking behaviour model to overcome skin-coloured distractions. It allows features to be split into two groups under severe distractions and merge later. The experiment results show that our method can track the hand in a cluttered background or when passing the face, while the Flocking of Features (FoF) and the Mean Shift Embedded Particle Filter (MSEPF) methods may fail. These results suggest that our method has better performance in comparison with the previous methods. It may therefore be helpful to promote the use of the hand gesture-based human-robot interaction method.

As the innovative products of Office, Office2007 and Office2010 use Open XML language to describe and ZIP to pack, and choose the SHA1-AES encryption mechanism. The paper started from the document structure, extracted the encryption information flow, and then cracked respectively the password on CPU computing computer and GPU computing platform by password brute-force method, analyzed the security of Office at last.

Deblocking filter of H.264 is identified as an important technology for improving the subjective quality of video. However, it is difficult to apply to low-cost embedded terminals due to its high computational complexity. By analyzing the statistical correlations of boundary strength and true-edge detection checks among adjacent blocks and pixels, this paper proposes an enhanced method of deblocking filter that efficiently reduces the deblocking complexity. In the ‘statistics-based deblocking filter (SBDF)’ proposed in this paper, only one boundary strength (BS) derivation is required for each macroblock edge instead of original four BS derivations, and four true-edge detections for each macroblock edge instead of original 16 true-edge detections. Experimental results show that the proposed SBDF significantly reduce the computational complexity with a slight loss in video quality.