Search

A deep convolutional neural network (CNN) becomes a widely used tool for object detection. Many previous works have achieved excellent performance on object detection benchmarks. However, these works present generic detectors whose performance will drop rapidly when they are applied to a surveillance scene. In this paper, we propose an efficient method to construct a scene-specific regression model based on a generic CNN-based classifier. Our regression model is an adaptive deep CNN (ADCNN), which can predict object locations in the surveillance scene. First, we transfer the generic CNN-based classifier to the surveillance scene by selecting useful kernels. Second, we learn the context information of the surveillance scene in our regression model for accurate location prediction. Our main contributions are: 1) a transfer learning method that selects useful kernels in the generic CNN-based classifier; 2) a special architecture that can effectively learn the local and global context information in the surveillance scene; and 3) a new objective function to effectively train parameters in ADCNN. Compared with some state-of-the-art models, ADCNN achieves the best performance on three surveillance data sets for pedestrian detection and one surveillance data set for vehicle detection. [ABSTRACT FROM AUTHOR]

We present a practical and inexpensive method to reconstruct 3D scenes that include transparent and mirror objects. Our work is motivated by the need for automatically generating 3D models of interior scenes, which commonly include glass. These large structures are often invisible to cameras or even to our human visual system. Existing 3D reconstruction methods for transparent objects are usually not applicable in such a room-sized reconstruction setting. Our simple hardware setup augments a regular depth camera (e.g., the Microsoft Kinect camera) with a single ultrasonic sensor, which is able to measure the distance to any object, including transparent surfaces. The key technical challenge is the sparse sampling rate from the acoustic sensor, which only takes one point measurement per frame. To address this challenge, we take advantage of the fact that the large scale glass structures in indoor environments are usually either piece-wise planar or a simple parametric surface. Based on these assumptions, we have developed a novel sensor fusion algorithm that first segments the (hybrid) depth map into different categories such as opaque/transparent/infinity (e.g., too far to measure) and then updates the depth map based on the segmentation outcome. We validated our algorithms with a number of challenging cases, including multiple panes of glass, mirrors, and even a curved glass cabinet. [ABSTRACT FROM AUTHOR]

A single-layer circularly polarized (CP) antenna with omnidirectional radiation in the azimuth plane has been designed and evaluated. The concept is the combination of multiple identical endfire CP antenna elements in a circular array configuration. The design of each element is based on the superposition of two closely spaced complementary current sources, which are, respectively, realized by a sectorial parallel-plate cavity and arc-shaped strip arms. A meandered double-sided parallel strip line is introduced to adjust the phase relationship between two orthogonal linear polarized components. All antenna elements can be simultaneously excited by adopting a single coaxial probe in the center of the overall structure. Take 5.8-GHz Industrial Scientific Medical (ISM) band applications (5.725–5.875 GHz), for example, a prototype with a diameter of 0.62\lambda 0 and a low profile of 0.029\lambda 0 was fabricated. Experimental results show that the designed antenna has a usable bandwidth from 5.68 to 5.91 GHz (the common overlapped bandwidth limited by the −10 dB \vert S11\vert and the 3-dB axial ratio). Within the operation band, good omnidirectivity and a wide coverage can be obtained. [ABSTRACT FROM PUBLISHER]

In this paper we present a novel real-time algorithm for simultaneous pose and shape estimation for articulated objects, such as human beings and animals. The key of our pose estimation component is to embed the articulated deformation model with exponential-maps-based parametrization into a Gaussian Mixture Model. Benefiting from this probabilistic measurement model, our algorithm requires no explicit point correspondences as opposed to most existing methods. Consequently, our approach is less sensitive to local minimum and handles fast and complex motions well. Moreover, our novel shape adaptation algorithm based on the same probabilistic model automatically captures the shape of the subjects during the dynamic pose estimation process. The personalized shape model in turn improves the tracking accuracy. Furthermore, we propose novel approaches to use either a mesh model or a sphere-set model as the template for both pose and shape estimation under this unified framework. Extensive evaluations on publicly available data sets demonstrate that our method outperforms most state-of-the-art pose estimation algorithms with large margin, especially in the case of challenging motions. Furthermore, our shape estimation method achieves comparable accuracy with state of the arts, yet requires neither statistical shape model nor extra calibration procedure. Our algorithm is not only accurate but also fast, we have implemented the entire processing pipeline on GPU. It can achieve up to 60 frames per second on a middle-range graphics card. [ABSTRACT FROM PUBLISHER]

In this paper, we introduce the notion of sufficient set and necessary set for distributed processing of probabilistic top-k queries in cluster-based wireless sensor networks. These two concepts have very nice properties that can facilitate localized data pruning in clusters. Accordingly, we develop a suite of algorithms, namely, sufficient set-based (SSB), necessary set-based (NSB), and boundary-based (BB), for intercluster query processing with bounded rounds of communications. Moreover, in responding to dynamic changes of data distribution in the network, we develop an adaptive algorithm that dynamically switches among the three proposed algorithms to minimize the transmission cost. We show the applicability of sufficient set and necessary set to wireless sensor networks with both two-tier hierarchical and tree-structured network topologies. Experimental results show that the proposed algorithms reduce data transmissions significantly and incur only small constant rounds of data communications. The experimental results also demonstrate the superiority of the adaptive algorithm, which achieves a near-optimal performance under various conditions. [ABSTRACT FROM AUTHOR]

In this paper, we introduce two types of probabilistic aggregation queries, namely, Probabilistic Minimum Value Queries (PMVQ)s and Probabilistic Minimum Node Queries (PMNQ)s. A PMVQ determines possible minimum values among all imprecise sensed data, while a PMNQ identifies sensor nodes that possibly provide minimum values. However, centralized approaches incur a lot of energy from battery-powered sensor nodes and well-studied in-network aggregation techniques that presume precise sensed data are not practical to inherently imprecise sensed data. Thus, to answer PMVQs and PMNQs energy-efficiently, we devised suites of in-network algorithms. For PMVQs, our in-network minimum value screening algorithm (MVS) filters candidate minimum values; and our in-network minimum value aggregation algorithm (MVA) conducts in-network probability calculation. PMNQs requires possible minimum values to be determined a prior, inevitably consuming more energy to evaluate than PMVQs. Accordingly, our one-phase and two-phase in-network algorithms are devised. We also extend the algorithms to answer PMNQ variants. We evaluate all our proposed approaches through cost analysis and simulations. [ABSTRACT FROM PUBLISHER]

When the independent sources are known to be nonnegative and well-grounded, which means that they have a nonzero pdf in the region of zero, Oja and Plumbley have proposed a "Nonnegative principal component analysis (PCA)" algorithm to separate these positive sources. Generally, it is very difficult to prove the convergence of a discrete-time independent component analysis (ICA) learning algorithm. However, by using the skew-symmetry property of this discrete-time "Nonnegative PCA" algorithm, if the learning rate satisfies suitable condition, the global convergence of this discrete-time algorithm can be proven. Simulation results are employed to further illustrate the advantages of this theory. [ABSTRACT FROM AUTHOR]

The narrow duty cycle in the buck converter limits its application for high-step-down dc-dc conversion. With a simple structure, the tapped-inductor buck converter shows promise for extending the duty cycle. However, the leakage inductance causes a huge turn-off voltage spike across the top switch. Also, the gate drive for the top switch is not simple due to its floating source connection. This paper solves all these problems by modifying the tapped-inductor structure. A simple lossless clamp circuit can effectively clamp the switch turn-off voltage spike and totally recover the leakage energy. Experimental results for 12 V-to-1.5 V and 48 V-to-6 V dc-dc conversions show significant improvements in efficiency. [ABSTRACT FROM AUTHOR]

In recent years, significant developments have occurred in the field of optically pumped magnetometers (OPMs), which are widely used in biomagnetic measurements. In this study, we propose a temperature optimization model for a single-beam miniaturized atomic magnetometer. Based on this model, we find that there is an optimal operating temperature for each individual vapor cell, under which the OPM achieves optimal performance. We construct a single-beam zero-field modulation OPM to prove this model. The sensitivity of the magnetometer reaches 18 fT/Hz1/2 (close to the theoretical sensitivity) when we adjust the temperature of the vapor cell through our model, and a detection bandwidth of more than 160 Hz is achieved. It is also discovered that the system bandwidth is reduced when the temperature is increased, whereas the sensitivity of the system does not show notable variation during this process. Our model enables a quick definition of the best working temperature for each individual magnetometer, ensuring the best performance at the lowest operating temperature. This model is especially advantageous for miniaturized OPMs and can also be further applied to arrayed integration for biomagnetic measurement in the future work. [ABSTRACT FROM AUTHOR]

Graph regularized non-negative matrix factorization (NMF) algorithms can be applied to information retrieval, image processing, and pattern recognition. However, challenge that still remains is to prove the convergence of this class of learning algorithms since the geometrical structure of the data space is considered. This paper presents the convergence properties of the graph regularized NMF learning algorithms. In the analysis, we focus on the study of Euclidian distance based algorithms. The structures of the fixed points are presented. The non-divergence of the learning algorithms is analyzed by constructing invariant sets for update rules. Based on Lyapunov indirect method, the stability of the algorithms is discussed in detail. The analysis shows that this class of NMF algorithms can converge to their fixed points under some given conditions. In the simulations, theoretical results presented in the paper are confirmed. For different initializations and data sets, variations of cost functions and decomposition data in the learning are presented to show the convergence features of the discussed NMF update rules, and the convergence speed of the algorithms is also investigated. [ABSTRACT FROM AUTHOR]

The skyline query, aiming at identifying a set of skyline tuples that are not dominated by any other tuple, is particularly useful for multicriteria data analysis and decision making. For uncertain databases, a probabilistic skyline query, called P-Skyline, has been developed to return skyline tuples by specifying a probability threshold. However, the answer obtained via a P-Skyline query usually includes skyline tuples undesirably dominating each other when a small threshold is specified; or it may contain much fewer skyline tuples if a larger threshold is employed. To address this concern, we propose a new uncertain skyline query, called U-Skyline query, in this paper. Instead of setting a probabilistic threshold to qualify each skyline tuple independently, the U-Skyline query searches for a set of tuples that has the highest probability (aggregated from all possible scenarios) as the skyline answer. In order to answer U-Skyline queries efficiently, we propose a number of optimization techniques for query processing, including 1) computational simplification of U-Skyline probability, 2) pruning of unqualified candidate skylines and early termination of query processing, 3) reduction of the input data set, and 4) partition and conquest of the reduced data set. We perform a comprehensive performance evaluation on our algorithm and an alternative approach that formulates the U-Skyline processing problem by integer programming. Experimental results demonstrate that our algorithm is 10-100 times faster than using CPLEX, a parallel integer programming solver, to answer the U-Skyline query. [ABSTRACT FROM AUTHOR]

This article presents investigation of 3-D electrical capacitance tomography (ECT) sensors with different excitation strategies, including dual-electrode excitation in the same plane, in adjacent plane, in planes separated by one plane or two planes, and in the adjacent position or opposite position in the same or different planes. The results show that a 3-D ECT sensor with dual-electrode excitation in the opposite position provides higher mean capacitance and lower standard deviation than in the adjacent position. Dual-electrode excitation in different planes can increase the capacitance signal and decrease the dynamic range of capacitance measurements, and thus, decrease random errors and improve the quality of 3-D reconstructed images. A 3-D ECT sensor with dual-electrode excitation in planes separated by one plane and opposite position can provide small measurement uncertainty and high image quality. [ABSTRACT FROM AUTHOR]

A conventional electrical capacitance tomography (ECT) sensor consists of multiple rectangular electrodes in a single plane. To generate 3-D images, some researchers use an ECT sensor with multiplane rectangular electrodes and in-line arrangement of different electrode planes and measure the capacitance between the electrode pairs not only within the same plane but also between different planes. Because those electrode pairs in different planes are far away from each other, the measured capacitance signals between them are weak and, hence, are easily affected by noise, resulting in the obvious disturbance on the measured signals and, hence, low image quality. To address those problems, the 3-D ECT sensors with different arrangements and different shapes of electrodes are studied to investigate the effect of different arrangements and different shapes of electrodes on the measured capacitance signals and the image quality. Four 3-D ECT sensors are made with different arrangements or shapes of electrodes: rectangular electrodes are in line arrangement, whereas diamond, circular, and hexagon electrodes are staggered arrangement. The performance of the four 3-D ECT sensors is compared. The results show that 3-D ECT sensors with staggered electrodes can decrease the disturbance on the measured signals and improve the image quality of static models compared with that with conventional rectangular electrodes. Among the three 3-D ECT sensors with staggered electrodes, the 3-D ECT sensor with hexagonal electrodes has superior performance to the other two 3-D ECT sensors with diamond and circular electrodes. It was used to measure moving objects and a fluidized bed. [ABSTRACT FROM AUTHOR]

Image reconstruction plays a key role in the application of electrical capacitance tomography (ECT). Although many different algorithms have been developed in the past, it is often difficult to obtain satisfactory images in all imaging regions by the use of a single algorithm due to the soft-field nature of ECT. This motivated us to develop a novel ECT image reconstruction strategy, in which a high-quality image can be obtained by combining the images reconstructed by two different algorithms via a graph cut method. By doing so, it is possible to retain the advantage of each algorithm for specified imaging regions and improve the quality of the whole image. This strategy was verified both numerically and experimentally for two widely used ECT reconstruction algorithms, i.e., linear backprojection and Tikhonov regularization. The results for stationary objects as well as gas–solid fluidized beds demonstrated that this graph-cut-based combination strategy presents a promising approach for ECT image reconstruction. [ABSTRACT FROM AUTHOR]

Sensitivity improvement and measurable pressure limit extension of Pirani vacuum gauge are urgent needed for meeting the rapid growing requirements in ultra-low and wide-range pressure applications. However, the adjustable range of measureable pressure of the Pirani vacuum gauge is limited to only several orders of magnitude of vacuum range. In this work, a fusion method was proposed to improve the sensitivity and extend the measurable pressure limit of the wafer-level packaged Pirani gauge for meeting the wider applications. An analytical temperature model was presented for estimating the electrothermal behavior of the suspended micro-heater with multi heat dissipation paths. The results calculated by the model are validated by the test results of Pirani gauges with both three and four heat sinks. The results indicate that the model for Pirani gauge with one heat sink structure will no longer make sense for devices with multi heat sinks. The electrical fusion of the Pirani gauges with these structure parameters extends dynamic range from 1 ~ 104 Pa to that of $6 \times 10^{-2} \sim 10^{4}$ Pa, which generates one more sensitivity peak compared to one single device. The sensitivity peaks are $5.7 \times 10^{3}$ K/W/decade and $3.4 \times 10^{3}$ K/W/decade around the upper and lower limits, respectively. [2019-0201] [ABSTRACT FROM AUTHOR]

This paper presents an experimental study on the failure mechanism of high-current gallium arsenide photoconductive semiconductor switches (GaAs PCSSs) triggered by the laser diode. We propose a Blumlein pulse forming line to test the GaAs PCSS. The result shows that the damage of GaAs PCSS almost occurs at the anode under the circumstances that either anode or cathode is illuminated, since the highest amplitude of Gunn oscillation always occurs on the anode. We speculate that the failure of high-current GaAs PCSS is attributed to the Gunn effect according to the statistics about failure factors of GaAs PCSS. We use a scanning electron microscope to semiquantitatively analyze the compositions of the erosions microball or granules, which are gallium and arsenium. The result proves our speculation. [ABSTRACT FROM AUTHOR]

The distributed file system, HDFS, is widely deployed as the bedrock for many parallel big data analysis. However, when running multiple parallel applications over the shared file system, the data requests from different processes/executors will unfortunately be served in a surprisingly imbalanced fashion on the distributed storage servers. These imbalanced access patterns among storage nodes are caused because a). unlike conventional parallel file system using striping policies to evenly distribute data among storage nodes, data-intensive file system such as HDFS store each data unit, referred to as chunk file, with several copies based on a relative random policy, which can result in an uneven data distribution among storage nodes; b). based on the data retrieval policy in HDFS, the more data a storage node contains, the higher probability the storage node could be selected to serve the data. Therefore, on the nodes serving multiple chunk files, the data requests from different processes/executors will compete for shared resources such as  hard disk head and network bandwidth, resulting in a degraded I/O performance. In this paper, we first conduct a complete analysis on how remote and imbalanced read/write patterns occur and how they are affected by the size of the cluster. We then propose novel methods, referred to as Opass, to optimize parallel data reads, as well as to reduce the imbalance of parallel writes on distributed file systems. Our proposed methods can benefit parallel data-intensive analysis with various parallel data access strategies. Opass adopts new matching-based algorithms to match processes to data so as to compute the maximum degree of data locality and balanced data access. Furthermore, to reduce the imbalance of parallel writes, Opass employs a heatmap for monitoring the I/O statuses of storage nodes and performs HM-LRU policy to select a local optimal storage node for serving write requests. Experiments are conducted on PRObE’s Marmot 128-node cluster testbed and the results from both benchmark and well-known parallel applications show the performance benefits and scalability of Opass. [ABSTRACT FROM PUBLISHER]

Moving object detection is about foreground and background separation based on motion detection. Detecting moving objects from similarly colored background (known as camouflage problem) has been a long-standing open question in this field. Discriminative modeling (DM), which focuses on enhancing the performance to distinguish foreground from background with discriminative features and well-designed classifiers, has been widely used for moving object detection. However, DM may tend to fail when encountering the camouflage problem, as the class separability in camouflaged areas is generally poor. In this paper, we propose a new strategy, camouflage modeling (CM), to identify camouflaged foreground pixels. In view of the fact that camouflage involves both foreground and background, we need to model both the background and the foreground, and compare them in a well-designed way in camouflage detection. Specifically, we develop a global model for the background, and an integration of global and local models for the foreground, respectively. Based on both background and foreground models, we introduce a factor to measure the degree of camouflage, and further identify truly camouflaged areas. In view of the fact that a moving object is usually composed of both camouflaged and noncamouflaged areas, CM and DM are fused in a Bayesian framework to perform complete object detection. Experiments are conducted on testing sequences to demonstrate the effectiveness of the proposed algorithm. [ABSTRACT FROM PUBLISHER]

Multiperson decision making (MPDM) problems with different formats of preference information are one of the emerging research areas in decision analysis. Existing approaches for dealing with different preference formats tend to be unwieldy. This paper proposes a new method to solve the problem, in which the preference information on alternatives provided by experts can be represented in four different formats, namely: 1) utility values; 2) preference orderings; 3) multiplicative preference relations; and 4) fuzzy preference relations. An optimization model is constructed to integrate the four formats of preference and to assess ranking values of alternatives. The model is shown to be theoretically sound and complete via a series of theorems, and then a corresponding algorithm is developed. A numerical example is given to illustrate the procedure. The proposed approach is more efficient and simpler than existing approaches because it does not need to unify different formats of preferences or to aggregate individual preferences into a collective one. Therefore, it overcomes a major shortcoming of existing approaches that lose or distort the original preference information in the process of unifying the formats. [ABSTRACT FROM AUTHOR]