The classical vector median filter (VMF) has been widely used to remove impulse noise from color images. However, since the VMF cannot identify thin lines during the denoising process, many thin lines may be removed out as noise. This serious problem can be solved by a newly proposed filter that uses a noise detector to find these thin lines and then keep them unchanged. In this new approach, the noise detection scheme applied on a current processed pixel is realized through counting the close pixels in its eight neighbor positions and the expanded window to see whether the current pixel is corrupted by impulse noise. Based on the previous outputs, our algorithm can increase the performance in detecting and canceling the impulse noise. Extensive experiments indicate that this approach can be used to remove the impulse noise from a color image without distorting the useful information.

Vector graphic , as a kind of geometric representation of raster images, has many advantages, e.g., definition independence and editing facility. A popular way to convert raster images into vector graphics is {image meshing}, the aim of which is to find a mesh to represent an image as faithfully as possible. For traditional meshing algorithms, the crux of the problem resides mainly in the high non-linearity and non-smoothness of the objective, which makes it difficult to find a desirable optimal solution. To ameliorate this situation, we present a hierarchical optimization algorithm solving the problem from coarser levels to finer ones, providing initialization for each level with its coarser ascent. To further simplify the problem, the original non-convex problem is converted to a linear least squares one, and thus becomes convex, which makes the problem much easier to solve. A dictionary learning framework is used to combine geometry and topology elegantly. Then an alternating scheme is employed to solve both parts. Experiments show that our algorithm runs fast and achieves better results than existing ones for most images.

Automatic pectoral muscle removal on medio-lateral oblique (MLO) view of mammogram is an essential step for many mammographic processing algorithms. However, it is still a very difficult task since the sizes, the shapes and the intensity contrasts of pectoral muscles change greatly from one MLO view to another. In this paper, we propose a novel method based on a discrete time Markov chain (DTMC) and an active contour model to automatically detect the pectoral muscle boundary. DTMC is used to model two important characteristics of the pectoral muscle edge, i.e., continuity and uncertainty. After obtaining a rough boundary, an active contour model is applied to refine the detection results. The experimental results on images from the Digital Database for Screening Mammography (DDSM) showed that our method can overcome many limitations of existing algorithms. The false positive (FP) and false negative (FN) pixel percentages are less than 5% in 77.5% mammograms. The detection precision of 91% meets the clinical requirement.

Carrier frequency offset (CFO) estimation is critical for orthogonal frequency-division multiplexing (OFDM) based transmissions. In this paper, we present a low-complexity, blind CFO estimator for OFDM systems with constant modulus (CM) signaling. Both single-input single-output (SISO) and multiple-input multiple-output (MIMO) systems are considered. Based on the assumption that the channel keeps constant during estimation, we prove that the CFO can be estimated uniquely and exactly through minimizing the power difference of received data on the same subcarriers between two consecutive OFDM symbols; thus, the identifiability problem is assured. Inspired by the sinusoid-like cost function, curve fitting is utilized to simplify our algorithm. Performance analysis reveals that the proposed estimator is asymptotically unbiased and the mean square error (MSE) exhibits no error floor. We show that this blind scheme can also be applied to a MIMO system. Numerical simulations show that the proposed estimator provides excellent performance compared with existing blind methods.

This paper presents a robust lossless data hiding scheme. The original cover image can be recovered without any distortion after data extraction if the stego-image remains intact, and conversely, the hidden data can still be extracted correctly if the stego-image goes through JPEG compression to some extent. A cover image is divided into a number of non-overlapping blocks, and the arithmetic difference of each block is calculated. By shifting the arithmetic difference value, we can embed bits into the blocks. The shift quantity and shifting rule are fixed for all blocks, and reversibility is achieved. Furthermore, because the bit-0- and bit-1-zones are separated and the particularity of the arithmetic differences, minor changes applied to the stego-image generated by non-malicious attacks such as JPEG compression will not cause the bit-0- and bit-1-zones to overlap, and robustness is achieved. The new embedding mechanism can enhance embedding capacity and the addition of a threshold can make the algorithm more robust. Experimental results showed that, compared with previous schemes, the performance of the proposed scheme is significantly improved.

We investigate a problem of object-oriented (OO) software quality estimation from a multi-instance (MI) perspective. In detail, each set of classes that have an inheritance relation, named ‘class hierarchy’, is regarded as a bag, while each class in the set is regarded as an instance. The learning task in this study is to estimate the label of unseen bags, i.e., the fault-proneness of untested class hierarchies. A fault-prone class hierarchy contains at least one fault-prone (negative) class, while a non-fault-prone (positive) one has no negative class. Based on the modification records (MRs) of the previous project releases and OO software metrics, the fault-proneness of an untested class hierarchy can be predicted. Several selected MI learning algorithms were evaluated on five datasets collected from an industrial software project. Among the MI learning algorithms investigated in the experiments, the kernel method using a dedicated MI-kernel was better than the others in accurately and correctly predicting the fault-proneness of the class hierarchies. In addition, when compared to a supervised support vector machine (SVM) algorithm, the MI-kernel method still had a competitive performance with much less cost.

This paper presents a velocity controller for the cutting system of a trench cutter (TC). The cutting velocity of a cutting system is affected by the unknown load characteristics of rock and soil. In addition, geological conditions vary with time. Due to the complex load characteristics of rock and soil, the cutting load torque of a cutter is related to the geological conditions and the feeding velocity of the cutter. Moreover, a cutter’s dynamic model is subjected to uncertainties with unknown effects on its function. In this study, to deal with the particular characteristics of a cutting system, a novel adaptive fuzzy integral sliding mode control (AFISMC) is designed for controlling cutting velocity. The model combines the robust characteristics of an integral sliding mode controller with the adaptive adjusting characteristics of an adaptive fuzzy controller. The AFISMC cutting velocity controller is synthesized using the backstepping technique. The stability of the whole system including the fuzzy inference system, integral sliding mode controller, and the cutting system is proven using the Lyapunov theory. Experiments have been conducted on a TC test bench with the AFISMC under different operating conditions. The experimental results demonstrate that the proposed AFISMC cutting velocity controller gives a superior and robust velocity tracking performance.

Thin metal sheets are often located in the coupling paths of magnetic coupling energy transfer (MCET) systems. Eddy currents in the metals reduce the energy transfer efficiency and can even present safety risks. This paper describes the use of etched fractal patterns in the metals to suppress the eddy currents and improve the efficiency. Simulation and experimental results show that this approach is very effective. The fractal patterns should satisfy three features, namely, breaking the metal edge, etching in the high-intensity magnetic field region, and etching through the metal in the thickness direction. Different fractal patterns lead to different results. By altering the eddy current distribution, the fractal pattern slots reduce the eddy current losses when the metals show resistance effects and suppress the induced magnetic field in the metals when the metals show inductance effects. Fractal pattern slots in multilayer high conductivity metals (e.g., Cu) reduce the induced magnetic field intensity significantly. Furthermore, transfer power, transfer efficiency, receiving efficiency, and eddy current losses all increase with the increase of the number of etched layers. These results can benefit MCET by efficient energy transfer and safe use in metal shielded equipment.

In this Exa byte scale era, data increases at an exponential rate. This is in turn generating a massive amount of metadata in the file system. Hadoop is the most widely used framework to deal with big data. Due to this growth of huge amount of metadata, however, the efficiency of Hadoop is questioned numerous times by many researchers. Therefore, it is essential to create an efficient and scalable metadata management for Hadoop. Hash-based mapping and subtree partitioning are suitable in distributed metadata management schemes. Subtree partitioning does not uniformly distribute workload among the metadata servers, and metadata needs to be migrated to keep the load roughly balanced. Hash-based mapping suffers from a constraint on the locality of metadata, though it uniformly distributes the load among NameNodes, which are the metadata servers of Hadoop. In this paper, we present a circular metadata management mechanism named dynamic circular metadata splitting (DCMS). DCMS preserves metadata locality using consistent hashing and locality-preserving hashing, keeps replicated metadata for excellent reliability, and dynamically distributes metadata among the NameNodes to keep load balancing. NameNode is a centralized heart of the Hadoop. Keeping the directory tree of all files, failure of which causes the single point of failure (SPOF). DCMS removes Hadoop’s SPOF and provides an efficient and scalable metadata management. The new framework is named ‘Dr. Hadoop’ after the name of the authors.

The widespread availability of portable computing power and inexpensive digital cameras are opening up new possibilities for retailers in some markets. One example is in optical shops, where a number of systems exist that facilitate eyeglasses selection. These systems are now more necessary as the market is saturated with an increasingly complex array of lenses, frames, coatings, tints, photochromic and polarizing treatments, etc. Research challenges encompass Computer Vision, Multimedia and Human-Computer Interaction. Cost factors are also of importance for widespread product acceptance. This paper describes a low-cost system that allows the user to visualize different glasses models in live video. The user can also move the glasses to adjust its position on the face. The system, which runs at 9.5 frames/s on general-purpose hardware, has a homeostatic module that keeps image parameters controlled. This is achieved by using a camera with motorized zoom, iris, white balance, etc. This feature can be specially useful in environments with changing illumination and shadows, like in an optical shop. The system also includes a face and eye detection module and a glasses management module.

Congestion in wireless sensor networks (WSNs) not only causes severe information loss but also leads to excessive energy consumption. To address this problem, a novel scheme for congestion avoidance, detection and alleviation (CADA) in WSNs is proposed in this paper. By exploiting data characteristics, a small number of representative nodes are chosen from those in the event area as data sources, so that the source traffic can be suppressed proactively to avoid potential congestion. Once congestion occurs inevitably due to traffic mergence, it will be detected in a timely way by the hotspot node based on a combination of buffer occupancy and channel utilization. Congestion is then alleviated reactively by either dynamic traffic multiplexing or source rate regulation in accordance with the specific hotspot scenarios. Extensive simulation results under typical congestion scenarios are presented to illuminate the distinguished performance of the proposed scheme.

Software reverse engineering techniques are applied most often to reconstruct the architecture of a program with respect to quality constraints, or non-functional requirements such as maintainability or reusability. In this paper, AOPR, a novel actor-oriented program reverse engineering approach, is proposed to reconstruct an object-oriented program architecture based on a high performance model such as an actor model. Reconstructing the program architecture based on this model results in the concurrent execution of the program invocations and consequently increases the overall performance of the program provided enough processors are available. The proposed reverse engineering approach applies a hill climbing clustering algorithm to find actors.

In this paper, we propose an image driven shape deformation approach for stylizing a 3D mesh using styles learned from existing 2D illustrations. Our approach models a 2D illustration as a planar mesh and represents the shape styles with four components: the object contour, the context curves, user-specified features and local shape details. After the correspondence between the input model and the 2D illustration is established, shape stylization is formulated as a style-constrained differential mesh editing problem. A distinguishing feature of our approach is that it allows users to directly transfer styles from hand-drawn 2D illustrations with individual perception and cognition, which are difficult to identify and create with 3D modeling and editing approaches. We present a sequence of challenging examples including unrealistic and exaggerated paintings to illustrate the effectiveness of our approach.

We investigate the bandwidth allocation and power control schemes in orthogonal frequency division multiplexing (OFDM) based multi-hop cognitive radio networks, and the color-sensitive graph coloring (CSGC) model is viewed as an efficient solution to the spectrum assignment problem. We extend the model by taking into account the power control strategy to avoid interference among secondary users and adapt dynamic topology. We formulate the optimization problem encompassing the channel allocation, power control with the interference constrained below a tolerable limit. The optimization objective with two different optimization strategies focuses on the routes rather than the links as in traditional approaches. A heuristic solution to this nondeterministic polynomial (NP)-hard problem is presented, which performs iterative channel allocation according to the lowest transmission power that guarantees the link connection and makes channel reuse as much as possible, and then the transmission power of each link is maximized to improve the channel capacity by gradually adding power level from the lowest transmission power until all co-channel links cannot satisfy the interference constraints. Numerical results show that our proposed strategies outperform the existing spectrum assignment algorithms in the performance of both the total network bandwidth and minimum route bandwidth of all routes, meanwhile, saving the transmission power.

In this paper, a novel Bayesian-Gaussian neural network (BGNN) is proposed and applied to on-line modeling of a hydraulic turbine system (HTS). The new BGNN takes account of the complex nonlinear characteristics of HTS. Two redefined training procedures of the BGNN include the off-line training of the threshold matrix parameters, optimized by swarm optimization algorithms, and the on-line BGNN predictive application driven by the sliding window data method. The characteristics models of an HTS are identified using the new BGNN method and simulation results are presented which show the effectiveness of the BGNN in addressing modeling problems of HTS.

Many important developments in video compression technologies have occurred during the past two decades. The block-based discrete cosine transform with motion compensation hybrid coding scheme has been widely employed by most available video coding standards, notably the ITU-T H.26xand ISO/IEC MPEG-xfamilies and video part of China audio video coding standard (AVS). The objective of this paper is to provide a review of the developments of the four basic building blocks of hybrid coding scheme, namely predictive coding, transform coding, quantization and entropy coding, and give theoretical analyses and summaries of the technological advancements. We further analyze the development trends and perspectives of video compression, highlighting problems and research directions.

It is universally acknowledged by network security experts that proactive peer-to-peer (P2P) worms may soon engender serious threats to the Internet infrastructures. These latent threats stimulate activities of modeling and analysis of the proactive P2P worm propagation. Based on the classical two-factor model, in this paper, we propose a novel proactive worm propagation model in unstructured P2P networks (called the four-factor model) by considering four factors: (1) network topology, (2) countermeasures taken by Internet service providers (ISPs) and users, (3) configuration diversity of nodes in the P2P network, and (4) attack and defense strategies. Simulations and experiments show that proactive P2P worms can be slowed down by two ways: improvement of the configuration diversity of the P2P network and using powerful rules to reinforce the most connected nodes from being compromised. The four-factor model provides a better description and prediction of the proactive P2P worm propagation.

Recently, Sun et al. (2005) highlighted the essential property of perfect forward secrecy (PFS) for e-mail protocols when a higher security level is desirable. Furthermore, Sun et al. (2005)’s protocols take only a single e-mail server into account. Actually, it is much more common that the sender and the recipient register at different e-mail servers. Compared to existing protocols, the protocol proposed in this paper takes into account the scenario that the sender and the recipient register at different servers. The proposed protocol is skillfully designed to achieve PFS and end-to-end security as well as to satisfy the requirements of confidentiality, origin, integrity and easy key management. The comparison in terms of functionality and computational efficiency demonstrates the superiority of the present scheme.

We present an antenna-in-package system integrated with a meander line antenna based on low temperature co-fired ceramic (LTCC) technology. The proposed system employs a meander line patch antenna, a packaging layer, and a laminated multi-chip module (MCM) for integration of integrated circuit (IC) bare chips. A microstrip feed line is used to reduce the interaction between patch and package. To decrease electromagnetic coupling, a via hole structure is designed and analyzed. The meander line antenna achieved a bandwidth of 220 MHz with the center frequency at 2.4 GHz, a maximum gain of 2.2 dB, and a radiation efficiency about 90% over its operational frequency. The whole system, with a small size of 20.2 mm × 6.1 mm × 2.6 mm, can be easily realized by a standard LTCC process. This antenna-in-package system integrated with a meander line antenna was fabricated and the experimental results agreed with simulations well.

We propose a framework of hand articulation detection from a monocular depth image using curvature scale space (CSS) descriptors. We extract the hand contour from an input depth image, and obtain the fingertips and finger-valleys of the contour using the local extrema of a modified CSS map of the contour. Then we recover the undetected fingertips according to the local change of depths of points in the interior of the contour. Compared with traditional appearance-based approaches using either angle detectors or convex hull detectors, the modified CSS descriptor extracts the fingertips and finger-valleys more precisely since it is more robust to noisy or corrupted data; moreover, the local extrema of depths recover the fingertips of bending fingers well while traditional appearance-based approaches hardly work without matching models of hands. Experimental results show that our method captures the hand articulations more precisely compared with three state-of-the-art appearance-based approaches.

Inverted index traversal techniques have been studied in addressing the query processing performance challenges of web search engines, but still leave much room for improvement. In this paper, we focus on the inverted index traversal on document-sorted indexes and the optimization technique called dynamic pruning, which can efficiently reduce the hardware computational resources required. We propose another novel exhaustive index traversal scheme called largest scores first (LSF) retrieval, in which the candidates are first selected in the posting list of important query terms with the largest upper bound scores and then fully scored with the contribution of the remaining query terms. The scheme can effectively reduce the memory consumption of existing term-at-a-time (TAAT) and the candidate selection cost of existing document-at-a-time (DAAT) retrieval at the expense of revisiting the posting lists of the remaining query terms. Preliminary analysis and implementation show comparable performance between LSF and the two well-known baselines. To further reduce the number of postings that need to be revisited, we present efficient rank safe dynamic pruning techniques based on LSF, including two important optimizations called list omitting (LSF_LO) and partial scoring (LSF_PS) that make full use of query term importance. Finally, experimental results with the TREC GOV2 collection show that our new index traversal approaches reduce the query latency by almost 27% over the WAND baseline and produce slightly better results compared with the MaxScore baseline, while returning the same results as exhaustive evaluation.

This paper reviews recent studies in understanding neural-network representations and learning neu-
ral networks with interpretable/disentangled middle-layer representations. Although deep neural networks have
exhibited superior performance in various tasks, interpretability is always Achilles’ heel of deep neural networks.
At present, deep neural networks obtain high discrimination power at the cost of a low interpretability of their
black-box representations. We believe that high model interpretability may help people break several bottlenecks of
deep learning, e.g., learning from a few annotations, learning via human–computer communications at the semantic
level, and semantically debugging network representations. We focus on convolutional neural networks (CNNs),
and revisit the visualization of CNN representations, methods of diagnosing representations of pre-trained CNNs,
approaches for disentangling pre-trained CNN representations, learning of CNNs with disentangled representations,
and middle-to-end learning based on model interpretability. Finally, we discuss prospective trends in explainable
artificial intelligence.

概要：为了针对不同的应用和流量提供服务质量保障和差异化服务，负载均衡和多优先级队列技术被广泛地应用于网络中。在传统网络中，网络管理员经常使用”traceroute”和”ping”工具来检测负载均衡机制或者服务质量策略是否正常工作。然而，由于这些工具并不被现有的OpenFlow交换机所支持，所以还不能够应用于软件定义网络中。此外，traceroute和ping依靠主动发送探测包来探测路径。然而，当负载均衡机制把探测包和所需追踪流的数据包均衡到不同路径时，这些工具将无法探测出流的真实转发路径，更无法测量出真实的路径时延。因此，为了准确的测量链路时延，测量工具必须能够提前找出数据包的真实转发路径。基于此发现，我们提出了一套新的软件定义网络中的流追踪机制”FlowTrace”，利用它来追踪任意流量的转发路径以及测量数据流所经历的链路时延。该工具通过收集交换机的流表来计算流的转发路径。然而，如果直接从交换机中查询流表会产生很大的数据平面流量，从而带来巨大的开销。因此，我们提出了一种被动的零开销的流表收集方法来解决该问题。在获得流的真实转发路径后，我们提出了一种新的测量方法来测量不同流的网络时延。最后，实验结果显示我们设计的方法可以准确的找出流的真实转发路径并测量出不同种类流所经历的时延。

We present an algorithm for image compression based on an image inpainting method. First the image regions that can be accurately recovered are located. Then, to reduce the data, information of such regions is removed. The remaining data besides essential details for recovering the removed regions are encoded to produce output data. At the decoder, an inpainting method is applied to retrieve removed regions using information extracted at the encoder. The image inpainting technique utilizes partial differential equations (PDEs) for recovering information. It is designed to achieve high performance in terms of image compression criteria. This algorithm was examined for various images. A high compression ratio of 1:40 was achieved at an acceptable quality. Experimental results showed attainable visible quality improvement at a high compression ratio compared with JPEG.

目的：针对认知异构网络中频谱的动态性和干扰的复杂性，利用随机几何理论进行干扰分析并提出网络容量的闭式表达式，为认知异构网络的性能分析提供理论基础。
创新：利用随机几何理论分析认知异构网络中复杂的干扰问题，并提出干扰和网络容量的闭式表达式。
方法：采用\"分而治之\"的方法，一方面，针对频谱使用的动态性，利用离散时间马尔可夫链对主用户（宏基站用户）进行建模，得到主用户离开和到达概率，并考虑理想/非理想感知情况，分别计算出两种情况下主用户和次用户的数量；另一方面，针对用户位置的随机性，利用齐次泊松点过程对基站和用户进行建模，再采用随机几何理论对干扰进行分析，提出计算干扰的闭式表达式。最后，利用以上两部分结果分别求出理想/非理想感知情况下网络容量的闭式表达式。
结论：利用随机几何理论分析认知异构网络中的干扰和容量具有可行性和准确性；理想感知情况下得到的网络容量要大于非理想感知的情况。

This paper presents the design, fabrication, and characterization of a quartz tuning fork temperature sensor based on a new ZY-cut quartz crystal bulk acoustic wave resonator vibrating in a flexural mode. Design and performance analysis of the quartz tuning fork temperature sensor were conducted and the thermal sensing characteristics were examined by measuring the resonance frequency shift of this sensor caused by an external temperature. Finite element method is used to analyze the vibratory modes and optimize the structure of the sensor. The sensor prototype was successfully fabricated and calibrated in operation from 0 to 100 °C with the thermo-sensitivity of 70×10^?6/°C. Experimental results show that the sensor has high thermo-sensitivity, good stability, and good reproducibility. This work presents a high-precision low-power temperature sensor using the comprehensive thermal characterization of the ZY-cut quartz tuning fork resonator.

We introduce a kind of shape-adjustable spline curves defined over a non-uniform knot sequence. These curves not only have the many valued properties of the usual non-uniform B-spline curves, but also are shape adjustable under fixed control polygons. Our method is based on the degree elevation of B-spline curves, where maximum degrees of freedom are added to a curve parameterized in terms of a non-uniform B-spline. We also discuss the geometric effect of the adjustment of shape parameters and propose practical shape modification algorithms, which are indispensable from the user’s perspective.

This paper deals with a reinforced cumulative probability distribution approach (CPDA) based method for extracting classification rules. The method includes two phases: (1) automatic generation of the membership function, and (2) use of the corresponding linguistic data to extract classification rules. The proposed method can determine suitable interval boundaries for any given dataset based on its own characteristics, and generate the fuzzy membership functions automatically. Experimental results show that the proposed method surpasses traditional methods in accuracy.

Optical proximity correction (OPC) is a key step in modern integrated circuit (IC) manufacturing. The quality of model-based OPC (MB-OPC) is directly determined by segment offsets after OPC processing. However, in conventional MB-OPC, the intensity of a control site is adjusted only by the movement of its corresponding segment; this scheme is no longer accurate enough as the lithography process advances. On the other hand, matrix MB-OPC is too time-consuming to become practical. In this paper, we propose a new sparse matrix MB-OPC algorithm with model-based mapping between segments and control sites. We put forward the concept of ‘sensitive area’. When the Jacobian matrix used in the matrix MB-OPC is evaluated, only the elements that correspond to the segments in the sensitive area of every control site need to be calculated, while the others can be set to 0. The new algorithm can effectively improve the sparsity of the Jacobian matrix, and hence reduce the computations. Both theoretical analysis and experiments show that the sparse matrix MB-OPC with model-based mapping is more accurate than conventional MB-OPC, and much faster than matrix MB-OPC while maintaining high accuracy.

Number estimation of controllers is a fundamental question in pinning synchronization of complex networks. This paper studies the problem of controller number in synchronizing a complex network of coupled dynamical systems by means of pinning. For a complex network with a symmetric coupling matrix and full coupling between the nodes, we formulate network synchronization via pinning as a linear matrix inequality criterion, and provide a lower bound and an upper bound of the controller number for a given complex network with fixed architecture. Several numerical examples with Barabási-Albert network topologies are provided to verify our theoretical results.