Search

The rotation matching between synthetic aperture radar (SAR) and optical images is essential for their registration. In this article, we introduce a novel fast rotation matching method for SAR and optical images. In the method, the absolute directional derivative histograms (ADDHs) of SAR and optical images can be used to reflect the rotation difference between them, and then the rotation difference is removed by the cross correlation of the two images’ ADDHs. Subsequently, we generate the ADDHs by oriented Gaussian-Gamma-shaped operator. Then, the cross correlation based on fast Fourier transform is used to deal with the ADDHs of SAR and optical images in order to determine the rotation difference of the two images. The experiment results demonstrate that the proposed method can detect the rotation difference in various conditions and achieves rotation matching with high speed.

The coupling relationship between the web tension and velocity in unwinding system is an important factor which affects system dynamic performance, how to realize independently control between tension and velocity with time-varying radius is an inevitable task. The coupling model of the web tension and velocity is established in this paper, and the diagonal matrix decoupling compensator is designed to separate the coupling variables. To optimize the dynamic performance of the unwinding system decoupled, a hybrid sensitivity robust controller is applied to improve the disturbance rejection performance and stability under uncertainty. The simulation results show that the robust decoupling control strategy can effectively reduce the coupling degree of tension and velocity, and improve the robustness of the unwinding system.

According to the mathematical model and the conduction mode of the brushless DC motor (BLDCM), a method to estimate the rotor position is presented, which is based on the conduction state of the corresponding power tube obtained by the rate of change of the non-commutative phase current combined with the zero-crossing of line back EMF. It is also applied to the sensorless control of the two-three conduction BLDCM. The brushless DC motor can run in no-load or load, low speed or high speed with this method, and the system uses the current inner loop and the speed outer loop to form a double closed-loop control, so that the motor runs more smoothly. The rotor position information of BLDCM under different conduction states is located. The simulation results are consistent with the theoretical analysis, which verifies the correctness and feasibility of the method.

For the permanent magnet synchronous motor drive system of traditional field-oriented control, the dual closed-loop PI cascade control structure has some drawbacks such as slow dynamic response, and poor low speed characteristic, which inevitably produces high torque ripple. Aiming at the finite switching sequence of the inverter, the optimal duty cycle predictive current control strategy is proposed to optimize the tracking performance of the torque current in the current inner loop. In the cascade structure control system, the power reaching law with sliding mode control strategy is applied to the speed outer loop to optimize torque performance during motor startup. A hybrid control algorithm combined sliding mode control with optimal duty cycle finite control set MPC is proposed to optimize the torque current setting and tracking. The proposed scheme can improve the control performance of the stator current and the robustness of the system. A performance comparison of the proposed hybrid control scheme with traditional control schemes is provided in this paper. The simulation results verify the feasibility of the proposed combined algorithm and prove that the dynamic performance of the system is improved. The purpose of suppressing the torque ripple of the motor is achieved by optimizing the torque current.

The web tension disturbance is the main problem in a roll-to-roll (R2R) systems, which results in position error, telescoping of the wound roll, and poor quality. The roller eccentricity in the unwinding process usually causes a periodic disturbance of the web tension. Then a plug-in repetitive controller was presented to attenuate the periodic tension disturbance. The control scheme just added a repetitive control part in primary controller, with the benefits of following or attenuating a discretional periodic signal in high accuracy. The simulation was carried out with a plug-in repetitive controller and a traditional PID controller. The comparison results show that the proposed control scheme performs well in the periodic tension disturbance attenuation, and dramatically improves the system steady errors.

Prandtl-Ishlinskii (PI) model is used to describe the hysteresis nonlinear characteristics of piezoelectric actuator. It has many parameters which are difficult to identify. In order to overcome the drawback, an adaptive particle swarm optimization (PSO) algorithm is proposed in this paper. The algorithm combines the adaptive adjustment of the weighting factor and the non-linear change of the learning factor. As a result the algorithm has a good global searching ability in the early stage and a fast convergence speed. After the algorithm is iterated to the later stage, the local searching ability is enhanced and the model accuracy is improved. Simulation results show that the proposed adaptive PSO algorithm has faster convergence speed and higher accuracy compared with the PSO Scheme.

The coupling and control of the web tension and lateral displacement is the key problem in flexible high-precision roll-to-roll manufacturing process. The web tension and lateral displacement are modeling independently with distributed control in current research, which is hard to solve the coupling between the web tension, velocity and lateral displacement for the reason of the model uncertainties, such as time varying radius of the unwinding roll, velocity variations, and so on. To solve the above problems, this paper establishes a three inputs and three outputs coupling model of an extended unwinding system with three axes. A state feedback decoupling control method with pole placement is presented to decouple of the web tension, velocity, and lateral displacement. The transient response of the decoupled system is achieved with pole placement method based on integral of time multiplied by absolute error (ITAE) optimal principles. This approach achieves dynamic decoupling of the web tension, velocity and lateral displacement of the unwinding system. The simulation results show well performance of the decoupled system.

The unwinding system is the front part of the roll-to-roll manufacturing process, in which the control accuracy of the web tension has a huge effect on the manufacturing accuracy and product quality. Sliding mode control (SMC) method is widely used in web tension control to solve the disturbance caused by the radius variation of the unwinding roller. However, it is difficult to avoid the system chattering problem by using the general equivalent Sliding mode control (ESMC). In this paper, the mathematical model of the web tension in unwinding system is established by analyzing the longitudinal dynamics of the web. Then a new equivalent fuzzy sliding mode control (EFSMC) method is proposed to solve the chattering problems. The simulation experiment is carried out to compare the performance of the web tension control with ESMC and EFSMC method. The simulation results show that the EFSMC method has nice dynamic performance and tracking accuracy of the web tension control. The chattering problem caused by the web velocity variation can be effectively attenuated. The EFSMC method also shows good performance on the disturbance rejection of the web tension.

The task of Language-Based Image Editing (LBIE) aims at generating a target image by editing the source image based on the given language description. The main challenge of LBIE is to disentangle the semantics in image and text and then combine them to generate realistic images. Therefore, the editing performance is heavily dependent on the learned representation. In this work, conditional generative adversarial network (cGAN) is utilized for LBIE. We find that existing conditioning methods in cGAN lack of representation power as they cannot learn the second-order correlation between two conditioning vectors. To solve this problem, we propose an improved conditional layer named Bilinear Residual Layer (BRL) to learning more powerful representations for LBIE task. Qualitative and quantitative comparisons demonstrate that our method can generate images with higher quality when compared to previous LBIE techniques., Comment: To appear at ICASSP 2019. Implementation: https://github.com/vtddggg/BilinearGAN_for_LBIE

Account takeover (ATO) is a type of fraud where a fraudster gains unauthorized access of a legitimate user's account through phishing, malware, bought credentials from dark web etc. The sophisticated evasion of detection by fraudsters and the requirement of friction-free experience by customers call for a new detection technique. Rather than using statistical features from behavior sequence in most of existing solutions, we represent account/context with graph node embedding and extract inherent sequential patterns with Recurrent Neural Networks (RNNs). Instead of using plain RNN by state-of-art fraud detection models, we further propose a selective graph attention mechanism within sequence (GAS), attending to relevant steps only to assist learning longer dependencies. The proposed selective graph attention mechanism is applicable to general graph structures. Experiments on a real dataset from Alibaba Group are conducted to compare the proposed model to several state-of-the-art approaches. The results demonstrate the effectiveness of the proposed graph attention mechanism. A real case study is also presented to further explain how proposed GAS works.

One of the problems of real time compressed sensing system is the computational cost of the reconstruction algorithms. It is especially problematic for close loop sensory applications where the sensory parameters needs to be constantly adjust to adapt to a dynamic scene. Through a preliminary experiment with MNIST dataset, we showed that we can extract some scene information (object recognition, scene movement direction and speed) based on the compressed samples using a deep convolutional neural network. It achieves 100% percent accuracy in distinguishing moving velocity, 96.22% in recognizing the digit and 90.04% in detecting moving direction after the code images are re-centered. Even though the classification accuracy drops slightly compared to using original videos, the computational speed is two time faster than classification on videos directly. This method also eliminates the need for sparse reconstruction prior to classification.

We present a hardware-friendly spatiotemporal compressed sensing framework for video compression. The spatiotemporal compressed sensing incorporates random sampling in both spatial and temporal domain to encode the video scene into a single coded image. During decoding, the video is reconstructed using dictionary learning and sparse recovery. The evaluation results demonstrate the proposed approach can achieve high compression rate (10 : 1–30 : 1) and robustness reconstruction quality (> 20dB) on noisy database. Additionally, it also enables power efficient and real-time CMOS implementation (0.7 nJ/pixel).

Visual saliency models are difficult to implement in hardware for real time applications due to their computational complexity. The conventional digital implementation is not optimal because of the requirement of a large number of convolution operations for filtering on several feature channels across multiple image pyramids [1], [2]. Here, we propose an alternative approach to implement a neuromorphic visual saliency algorithm [3] in digital hardware using stochastic computation, which can achieve very low power and small area. We show the real time implementation of important building blocks of the system and compare the overall system with its software implementation. Our implementation will be useful for facilitating high-fidelity selective rendering in computer graphics applications using the output of the saliency model, and for communications, where the non-salient parts of an image can be compressed more heavily than the salient parts. Our implementation will find several applications as a frontend co-processor for information triaging, compression and analysis in computer vision tasks. Our proposed SC-based convolution circuit could be a potential building block for implanting deep convolutional neural networks (CNN) on hardware.

A compact all-CMOS spatiotemporal compressed sensing (CS) video camera is demonstrated. This CS-based framework [1], implemented on integrated circuits, is able to achieve 20-fold reduction in the readout speed and consumes only 14μW to provide 100 fps videos. Taking advantage of dictionary learning and sparse recovery, this prototype image sensor (127×90 pixels) can reconstruct 100 fps videos from the coded images sampled at 5 fps.

At present, the performance of image registration mainly depends on the extracted features in feature-based image registration. However, due to the speckle noise, synthetic aperture radar (SAR) image registration will have a lower accuracy and less robustness. For this purpose, we design a deep neural network (DNN) for SAR image registration, using the DNN to learn the image features, automatically. The deep learning could learn the more essential features of the images, which are make the image registration to achieve more robust features and accurate matching. Moreover, this paper proposed a new strategy to remove the wrong matching points based on the RANSAC. The experimental results on SAR image registration show that this image registration method based on DNN have a better performance, and the new RANSAC strategy could eliminate many wrong matching points and get a good transformational model.

This paper presents an efficient conflict-free memory address scheme for arbitrary 2k-point memory-based FFT processors. The address scheme can support in-place strategy, continuous-flow mode, and variable size for arbitrary long sized FFT. More over, the address scheme can adjust the throughput by changing the computation radix and the parallelism of the arithmetic processing units according to the specific implementation requirements. We have designed a configurable 128 — 2048 point FFT processor for LTE system. From the results, we can see the proposed address scheme has excellent point flexibility, hardware efficiency, and can support almost any 2k-point FFT implementations.

This paper presents an efficient memory-based fast Fourier transform processor including 35 different working sizes for LTE systems. A factorization method named high-radix-small-butterfly combined with a conflict-free address scheme for 2p3q5r point memory-based FFT processor is proposed. The processor can not only provide conflict-free concurrent data access from different memory banks but also continuous-flow working mode. Moreover, we exploit prime factor algorithm to decrease the multiplications and twiddle factor storage. In addition, a unified Winograd Fourier transform algorithm butterfly core was designed for the small 2, 3, 4, 5-point DFTs. The FFT processor was implemented in a SMIC 55nm CMOS process with core area 1.063mm2. The chip consumes 40 8mW at 122.88MHz operating frequency with 1.08V voltage supply.

Recently, CMOS image sensors have attracted more and more attention from the applications of navigation, monitoring and search-and-rescue operations. Specially, CMOS image sensors mounted on insects need to be fast, adaptive to the environment and power efficiency. To simultaneously satisfy both requirements of reconstruction quality and low power consumption, we propose a compressed sensing block-wise exposure control algorithm using optical flow estimation. This framework has been demonstrated to further improve recovery performance (> 25 dB) with high compression ratio (>= 10 : 1), which also provides a promising method for real-time CMOS implementation.

Due to the high cost of gold wire, copper wire and silver wire become the mainstream of wire materials in electronic package industry. Recently, taking the reliability of the IC components into consideration, Pd-coated copper wire and Ag alloy wire are the most popular wires using in wire bonding. However the study of reliability about Ag alloy wire bonding is very rare. The differences between Pd-coated copper wire and Ag alloy wire have not been studied systematically. In this paper, bond reliability under humid environment for Pd-coated Cu and Ag alloy wire bonding were studied. Shear tests were used to evaluate the quality and robustness of joints. The reliability test under humid environment used in this paper is unbiased-highly-accelerated temperature and humidity stress test (u-HAST). U-HAST is conducted at 130°C and 85% relative humidity. The bond-pad interface was analyzed by scanning electron microscopy (SEM) and electron-dispersive microscopy (EDS). The bond reliability of Cu wire under humid environment is better than Ag alloy wires.

K novel wearable Ultra-Wideband antenna with dual band-notched characteristics is proposed. Slots are etched on both microstrip feed-line and ground plane of the antenna to produce two notching bands, aiming to reduce the interference from WLAN and WiMAX systems. Effects of different bending condition on antenna have been investigated, showing that the antenna bas a high degree of practicability.

To meet the growing demand of wireless and power efficient neural recordings systems, we demonstrate an unsupervised dictionary learning algorithm in Compressed Sensing (CS) framework which can be implemented in VLSI systems. Without prior label information of neural spikes, we extend our previous work to unsupervised learning and construct a dictionary with discriminative structures for spike sorting. To further improve the reconstruction and classification performance, we proposed a joint prediction to determine the class of neural spikes in dictionary learning. When the neural spikes is compressed 50 times, our approach can achieve an average gain of 2 dB and 15 percentage units over state-of-the-art of CS approaches in terms of the reconstruction quality and classification accuracy respectively.

Nonnegative Matrix Factorization (NMF), defined as factorizing a nonnegative matrix into two nonnegative factor matrices, is a particularly important problem in machine learning. Unfortunately, it is also ill-posed and NP-hard. We propose a fast, robust, and provably correct algorithm, namely Gradient Vertex Pursuit (GVP), for solving a well-defined instance of the problem which results in a unique solution: there exists a polytope, whose vertices consist of a few columns of the original matrix, covering the entire set of remaining columns. Our algorithm is greedy: it detects, at each iteration, a correct vertex until the entire polytope is identified. We evaluate the proposed algorithm on both synthetic and real hyperspectral data, and show its superior performance compared with other state-of-the-art greedy pursuit algorithms.

The explosive growth of mobile applications causes the mobile traffic to easily exceed the capacity of the cloud service due to the bandwidth limits of last mile connections to the cloud and legacy backhauls to macrocells' base stations. It degrades mobile applications' quality of service since the mobile devices have to spend more time and thus consume more battery power for data transmissions. It also enforces the cloud provider to put a huge investment to update its infrastructure and the induced cost is inevitably borne by all mobile users. To resolve this issue, in this paper we propose a so-called community clinic solution, which embeds the cloudlet group between the cloud and mobile users, to cut down the cost introduced by the massive deployment of the cloud's data centers and save the battery power consumed by the mobile devices. We firstly show that the mobile devices can consume less energy by choosing the service provided by the cloudlet group. We then model the system with and without the cloudlet group as two types of supply chain and prove that the cloudlet group can increase the cloud's profit without putting additional cost on mobile users. We also propose the real-time group-buying auction for the cloudlet group to promote its service to its nearby mobile users with a lower price and maximize its profit. The community clinic can result in a win-win-win outcome among the cloud, cloudlet group and mobile users. Numerical experiments are further conducted to demonstrate the effectiveness of our scheme.

The main target of equipment layout in manned spacecraft is to build a safe and comfortable circumstance for astronauts. By discussing the effects of space environment to maintainability design of manned spacecraft, the evaluation index was proposed based on the difference between on-orbit maintenance and ground maintenance. According to the maintainability design principles and engineering practice, the evaluation index system was completed for the manned spacecraft, which was constituted by the evaluation index and the method of evaluation.

In conventional terrain classification for the polarimetric SAR (POLSAR) images, color features are rarely involved unless in one recent supervised work. Unlike that work, the color features are exploited for the unsupervised classification in this paper. Firstly, based on the polarimetric decomposition of the POLSAR data, the common color spaces, such as RGB, HSI, and CIELab are calculated. The color feature is quantitatively selected from these color spaces by introducing the color entropy. Then together with the spatial information, extended scattering power entropy and the copolarized ratio, the adaptive Mean-shift algorithm is used to segment the POLSAR image. Finally, the segments are merged according to the Wishart distance measurement. The experiments using AIRSAR L-band POLSAR data indicate that the proposed method has better discriminative ability for urban areas and for boundary preservation compared with existing works.

—Opposition-based Learning (OBL) has been reported with an increased performance in enhancing various optimization approaches. Instead of investigating the opposite point of a candidate in OBL, this study proposed a partial opposition-based learning (POBL) schema that focuses a set of partial opposite points (or partial opposite population) of an estimate. Furthermore, a POBL-based adaptive differential evolution algorithm (POBL-ADE) is proposed to improve the effectiveness of ADE. The proposed algorithm is evaluated on the CEC2014’s test suite in the special session and competition for real parameter single objective optimization in IEEE CEC 2014. Simulation results over the benchmark functions demonstrate the effectiveness and improvement of the POBL-ADE compared with ADE. Keywords—opposition-based learning; differential evolution; real parameter; optimization. I. INTRODUCTION Opposition-based learning (OBL), originally introduced by Tizhoosh [1], tries to find a better candidate solution by simultaneously considering an estimate point and its corresponding opposite estimate. It has been proved that an opposite candidate solution can provide a higher chance of finding solutions that are closer to the global optimum one [2][3]. The concept of OBL has been applied to improve the performance meta-heuristic algorithms and machine learning algorithms [4], [5]. In [6], the convergence speed of evolutionary algorithm is accelerated by replacing the random initialization with the opposition-based population initialization. Further, [2] mathematically and experimentally proves this advantage when there is no prior knowledge about the solution. The benefits of the opposite of a candidate solution over random solutions is also shown intuitively based on an Euclidean distance-to-optimal solution proof in [7]. Recently, by considering the opposite individuals in the population initialization stage and generation jumping stage, OBL has been recently applied to accelerate various meta-heuristic algorithms, such as Differential Evolution (DE)[8]–[10], Particle Swarm Optimization (PSO) [11], [12], Biogeography-Based Optimization (BBO) [13]–[15], teaching and learning algorithm [16], gravitational search algorithm [17], Harmony Search (HS) [18], and Artificial Bee Colony (ABC)[19]. OBL are also employed to accelerate machine learning algorithms including reinforcement learning and backpropagation learning in neural networks, and Estimation of Distribution Algorithm (EDA). Opposition-based reinforcement learning (ORL) were proposed by considering opposite states and opposite actions[20]–[22]. The results demonstrate that ORL outperforms the reinforcement learning. Similarly, by considering the opposite transfer functions and opposite weights, the opposition-based neural networks were also proposed to improve their learning speed and accuracy [23]–[25]. Motivated by the idea of OBL, this study presents an improved OBL, namely partial opposition-based learning (POBL). Rather than only examining the opposite point of a candidate, the POBL is devised to compute partial opposite points (or partial opposite population) of an estimate. Further, a POBL-based adaptive DE algorithm is proposed to solve the numerical optimization problems in “CEC2014 Special Session and Competition on real parameter single objective optimization”[26]. In the proposed POBL-based adaptive DE, the Adaptive Differential Evolution (ADE) [27] that needs no parameters to be tuned is improved by POBL during the population initialization and generation jumping. Experimental simulations on benchmark functions show that POBL-ADE obtains better performance on the majority of the test problems compared with basic ADE and OBL-ADE. The rest of the paper is structured as follows. Section II provides the overview of the ADE and the Opposition-based learning. Section III describes the proposed Partial opposition based learning and the POBL-based ADE. The experimental setting is depicted in Section IV. Section V reports the experimental results with discussions. Finally, Section V concludes this study with comments toward future research directions.

Energy efficiency is known to be a critical issue of wireless devices. Our analysis reveals that the energy cost on overhearing unnecessary packet transmissions accounts for the majority of unnecessary energy consumption of wireless devices in high traffic wireless LANs. In this paper, we propose a novel SASD mechanism to eliminate the energy inefficiency of the packet overhearing problem. By adding a new header that carries some information in front of the normal packet frame at the PHY layer, SASD enables the wireless devices to discern the required information under the energy-saving downclocking mode so that the devices, whenever not addressed, can switch to the sleeping mode to save energy. Our simulation results show that SASD can effectively save the energy cost by avoiding the packet overhearing in high traffic wireless LANs.