Computer Science, Technology and Application

The following sections are included:

• Preface
• CSTA2016 Committees
• KEYNOTE SPEECH
• Contents

In this paper, a new continuous thresholding function is proposed. This new thresholding function does not only retain the advantages of conventional standard thresholding functions, but also overcome their shortcomings. The larger (than λ) coefficients are often taken as effective signal while the smaller (than λ) are noise. The proposed method of removing noise from the disturbed signal is to keep the larger coefficients and suppress the smaller coefficients. The simulation results also show that the proposed function can achieve better denoising performance than traditional methods.

Factual recognition in association with age estimation has been studied widely. In this experiment, age estimation is performed by extracting and analyzing the image features such as contour and wrinkle of the human faces using Local Binary Pattern (LBP) algorithm and Canny operator for image detection. In this study, we deployed uniform LBP algorithm to extract the feature information of face wrinkles, and used the Canny operator to enhance the face contour and wrinkles information before integrating the two features of face images together to estimate the age of the subject. In this paper, we used the FG-NET Face Age libraries which are widely used for age experiments on groups of subjects using every ten years as the classification. The experiments showed that uniform-LBP algorithm and Canny algorithm for feature extraction and fusion have achieved results of 72% to 83% of correct classification based on recognition of contour and the wrinkle information on the human face, as the means to estimate the age of the candidates.

As the fast development of computer science and the wide application of multimedia technology, a mass of digital videos have become an indispensable part of our lives. In this paper, we first introduce the research background, development progress and current researches of video abstraction and videos intelligent recognition. Then, we proposed demand analysis of monitoring video intelligent recognition system and described function of all modules. According to result of demand analysis, we proposed the all design of system, introduced work flow of all functional modules. At last, with Visual C++ and DirectShow software, we implemented the system.

Due to the increasing number of accidents happening when flying target landing in the weapon testing field, a smart video surveillance system based on moving target recognition was designed. The system adopts the capturing front-server-decision model. In our paper, a method for detecting moving targets images using background difference method and frame difference method is first introduced. Secondly, target recognition is studied by the technology of contour extraction and edge detection. Finally, characteristic parameters of target are extracted by feature algorithm. Based on it, key technologies involved in the system are described in detail. Additionally, corresponding algorithm is designed using OpenCV in Visual C++ 6.0, and part of the key codes are given. Simulation results shows that the system designed can meet the needs of monitor and control of flying target, and also verify the effectiveness of the algorithm.

Targeting the application of intelligence reconnaissance of MUAV in model airplanes, the reconnaissance intelligence automatic identification system is designed to generate fast and exact interpretations of targets of MUAV. The system contains two main modules: the stable platform control module and the target recognition. In this paper, we used BRISK and hamming distance match keypoints and introduced a new way to detect keypoints with adaptive threshold vale method, as well as screened keypoints with the RANSAC algorithm. The experimental results show that it has a good robustness to scale transformation, lean transformation and is able to prevent problems of MUAV. Besides, the BRISK algorithm is nearly 20 times faster than SURF algorithm.

Anemometer is widely used as an important measuring instrument in modern ground, aerial and aquatic vehicles. The paper presents a novel type of anemometer based on time difference. There are three key contributions as follows. Firstly, the key principle for designing an ultrasonic anemometer is described as an novel approach by using time difference. Secondly, the architecture design, electro circuit design and the filter arithmetic are given in detail to show the feasibility of our design. Thirdly, the implementation is measured in a real scenario to determine the accuracy of this approach. Final results show that our system can measure the wind speed and direction with higher accuracy and reliability, and with stronger anti-jamming properties than other known instruments.

Recommendation in APP becomes increasingly popular as mobile Internet develops rapidly nowadays. Different from music recommendation, tariff package recommendation in telecoms operators’ APP is able to use customers’ recorded information for better performance. Consequently, multi-label learning can be applied for recommendation in that case. During the past decade, multi-label learning has a wide application in real work, such as automatic annotation for multimedia contents, web mining, tag recommendation, etc, and many famous frameworks, such as Binary Relevance(BR). It can raise and solve large amounts of multi-label classification problems successfully. In this paper, a new framework is presented which aims at solving the problem of correlated multi-label classification (CMC), which can be applied in tariff package recommendation. The new framework has three layers. On the first layer, all labels are transformed into several new-generated groups using adapted Principal Components Analysis. Each group is scored on the middle layer and the score will be disperse to original labels on the last layer. Furthermore, we analyze the performance of the framework on tariff package data set. The outcome shows that our framework can obtain more profits with a low time complexity compared to traditional methods.

Fluorescence microscopic image restoration is of great importance, which can be applied to many fields, such as astronomical imaging, electronic microscopy, single particle emission computed tomography (SPECT) and positron emission tomography (PET). Traditional image restoration based on split Bregman (SB) algorithm can preserve sharp edges, and save the image texture. However, serious staircase effect phenomena usually occurs with traditional image restoration. Therefore, an improved image restoration algorithm is proposed in this paper, which considers two aspects. One is that the total variation (TV) regularization is used, which is an effective tool to recover blurred images. The other is that the weight function of the total variation is involved, which can not only suppress the staircase effect, but also preserve the image texture information. The experimental results show superior performance in terms of both objective criteria and subjective human vision via processing simulated and real fluorescence microscopic degraded images, compared with the traditional restoration image methods.

A blurring image enhancement method based on high-dimensional space information geometry is proposed in this paper. In this novel image enhancement method, every image is considered as a point in the high-dimensional space. With the “blurring-blurringdeblurring” thought, this method can get ideal enhanced images by comparing the corresponding locations of various images with simple geometry analysis. The algorithm is implemented using the MATLAB tool in this paper, and the experiment results show that the effect of this algorithm is good. Also, compared to traditional image enhancement methods, this method is simple and efficient.

A visibility-filtering-based algorithm for rendering approximate soft shadows is proposed, which is well compatible with the framework of ray tracing. The luminaire in a 3D scene is treated as an ideal point light source and a binary light-visibility map for the visible region of the 3D scene is subsequently created by casting shadow rays. The fractional light visibility for visible points in penumbrae is computed by smoothing out the abrupt changes in the binary light-visibility map with a box filter. The light leaving visible points into the eye resulting from direct illumination is obtained by multiplying the one without considering possible occluders by the corresponding light-visibility value achieved by filtering the binary light-visibility map. Our algorithm can render visually pleasing soft shadows with only a small amount of overhead.

The current Communication-based Train Operation Control (CBTC) system adopts the adhesion, closed system architecture and autonomous safety model. It only makes the brakes respond to the train operation speed, train position, the input of the front-back train relationship and other input states. With the rapid development of urban rail transit, train operations are becoming more and more complex, and operators need to respond to more external events. However, since the current data acquisition equipment of CBTC system only has basic functions but has neither data acquisition nor data interface for other emergencies, the system has no response stored or is unable to respond in time for exogenous emergencies. Therefore, based on the current CBTC system, this article presents a new generation of global cooperative signaling for train control. Through increasing acquisition of external environment data acquisition equipment and interface, the safety cooperative function of wayside subsystem, carborne subsystem and other subsystem can be improved. Also, the new generation of global cooperative signaling for train control is able to have a timely and efficient response to fail-safe information. Improving the safety and reliability of train operation system brings about the fail-safe of the whole rail transit system.

A liver is divided into 8 segments. Sometimes patients only need a small portion of the liver, not the whole segment, thus we brought up a new idea that surgeons could make a more accurate operation plan according to freedom grouping of vessels. We develop a new nonstandard liver segments method to ensure the maximum retention quality of affected healthy tissues and patients function. Clinicians could make a more accurate operation plan according to freedom grouping of vessels. The resection portion based on conventional segments method is more than our nonstandard resection volume. Therefore, it could ensure the maximum quality of affected healthy tissues and patients function.

In this paper, we propose a novel content-based image semi-fragile watermarking algorithm for authentication in finite ridgelet transform (FRIT) domain. The authentication watermark was generated according to the image feature, which was the mean value of an image extracted from FRIT. It was embedded in the FRIT coefficients, which were selected by the minimizing Bayesian risk estimate. Experimental results show the algorithm is simple and effective, providing very good classification from malicious tampering to incidental modification. Besides, it has the ability to localize the malicious tampering at a high resolution.

This paper analyses the hyperoloidal parameters impact on the spatial resolution. The most important part of an omnidirectional vision system is the catadioptric mirrors. It is important to note that single-viewpoint property is suitable for image analysis and processing because we can easily generate any desired image through projecting the original circle image on any designated image plane. Thus, we briefly introduced the single-viewpoint general constraint equations. We then analyzed the hyperbola parameters influence to the system structure because perfect curvature of the hyperboloid mirror is the only option for a clear picture. The catadioptric system's spatial resolution distance impact on the vertical and horizontal object was subsequently discussed. Lastly, we concluded that we can exploit a flexibility hyperboloidal mirror by choosing the curvature of the mirror for different objects identification.

The information capacity of the black and white QR code with the logo is small and the logo will cover the encoding module that leads to a problem of high bit error rate. Thus, this paper designs a colored QR code which adapts to the size of the logo. Firstly, we do so by enhancing the information capacity of the color QR code by choosing a variety of colors such as encoding color, and selecting the greatest contrast color. Secondly, we extract the contour of the logo and place coding modules around the logo contour extracted. MATLAB simulation experiments and the bar code capacity analysis show that the scheme of the colored QR code can improve the capacity of the two-dimensional code and solve the problem of decreasing ability to correct errors caused by the covered encoding module.

For an electroencephalograph (EEG)-based brain—computer interface (BCI) system, the detection of mental states from spontaneous EEG signals is very important. In this paper, an approach based on artificial neural network (ANN) and discrete wavelet transform (DWT) is developed. Computer simulation results on multiple subjects show that the proposed method can successfully classify imagined and actual hand movements using EEG signals.

In this paper, a multicore DSP parallel implementation strategy for infrared target detection and tracking is proposed. Tracking-Learning-Detection is considered a highly efficient algorithm for tracking a single target. Although this algorithm can re-track a target when the target is occluded by other targets, many shortcomings still exist. This paper deals with the issue of target tracking by fusing local spectrum suppression based on human visual attention mechanism with tracking-learning-detection algorithm. Specifically, the area of the target is estimated to reduce the detection region and to increase the processing speed. Experimental results conducted on VIVID benchmark video library demonstrate that the proposed method can detect properly and track accurately infrared target in complex scenes.

To speed up the application of chaos, the dynamic properties of two fractional-order hyperchaotic systems were investigated. It was found that both of them remain chaotic when their orders are in the range (0, 1]. Based on the stability condition of fractionalorder differential systems, a controller was designed to realize modified projective synchronization of the two systems. Numerical simulation showed the effectiveness of the proposed controller.

Safety is a priority for aviation companies. To ensure aviation safety, consistent maintenance is needed. The maintenance of airplanes requires good technology, and scientific management of maintenance tools help to improve maintenance efficiency and reliability. In this paper, we designed and implemented a Radio Frequency Identification (RFID) based intelligent tools management system for the scientific management of maintenance tools. The intelligent tool car that contains the tools needed for maintenance of airplanes has an enclosed metallic environment. Thus, its structure requires optimization and adjustments to be made before the RFID system can be embedded. RFID tags are then pasted on each tool. Due to the unique signals of each RFID tag, missing tools can be identified. Our experimental results show that the operation of the system is stable and reliable in the complex electromagnetic enclosed metallic environment.

sDOMO is a communication protocol developed for home automation and the building of robotic systems. Being optimized for small devices, sDOMO allows certain 8 bit microcontrollers to be a full featured, independent node in the domotic network, yet the protocol is powerful enough to provide soft-real-time communication for our computervision distributed system for domestic robots. Being designed with security and privacy concerns in mind, sDOMO has unique features to protect the residents.

Selectional preference is an important concept widely used for syntactic and semantic analysis of natural language. However, works published so far on the automatic acquisition of Chinese selectional preference are somewhat lacking in terms of accuracy. In this paper, a Chinese selectional preference model based on online resource Balanced Corpus of Modern Chinese developed by State Language Work Committee is introduced. Test results have indicated that the proposed model is effective, and demonstrate its fault tolerance ability in handling online materials with less accurate annotation.

In Mobile Ad-Hoc Networks (MANETS), data messages are transmitted along a wireless multihop transmission route consisting of multiple intermediate wireless nodes. Hence, intermediate wireless node and their neighbor nodes can receive the data messages. Due to the high performance of computers and advancement of various secure communication technologies, encrypted data messages can be decrypted by gathering a number of data messages. Works have been published on multi-route avoidance routing, which avoids sending data messages to malicious wireless nodes. Nevertheless, conventional avoidance routing operates under the assumption that the locations of malicious wireless nodes are already known by source wireless nodes. In order to solve this problem, this paper proposes a novel neighbor-disjoint multi-route ad-hoc routing, which could detect multiple wireless multihop transmission routes from the source node to the destination node where no other wireless node receives all the data messages. The simulation results show that with enough coverage of wireless networks, the wireless multihop transmission routes can be detected even at low ratio of the critical area. This paves the way for future studies of the route detection ratio and the size of the secure zones in critical areas caused by various disjoint route.

By analyzing the computerized flat knitter’s control system, we realized that data processing rate and storage speed are the main linchpins in improving the working speed of the machine. Since PCs have large capacity, high-frequency and good transmission characters, we propose a computerized flat knitting machine with network control suggestion which connects the PC and the flat knitter. To test the effectiveness of the proposed control system, a network control system model is created and the Truetime toolbox is used to simulate the sequence of control data timing. The results show that the flat knitters’ network control system transmission format is able to meet the industry’s requirements as well as increase the working speed of the flat knitter.

Based on the multi-layer characteristics of IOT security elements, this paper attempts to fine-tune dependability elements in network layer from the "internal cause" of the dependability situation, trying to make the optimization of system dependability indicators. Firstly, linear programming theory is used to analyze the change rule between the dependability elements and dependability indicators in the network layer, meanwhile, combined with the features and dependability properties, the change regulation function of the dependability is established from all details. At last, the optimization effect of network layer is simulated, and the results show that, the optimization method proposed in this paper can optimize the dependability of network layer in IOT to a large degree.

This study was designed to investigate hybrid teaching, which is a combination of traditional teaching with Moodle e-learning platform for classroom curriculum. Here, we will discuss the learning effectiveness and the learning attitude on special education students. This study analyzes the learning efficiency, motivation and satisfaction of 20 high school students before and after using Moodle online teaching modules. We evaluate the learning effectiveness on the hybrid teaching manner using a paper exam and a questionnaire to determine the learning attitudes of students. Our results show that the hybrid teaching model improves the learning effectiveness, and the students hold positive attitude towards learning.

Based on Elliptic Curve Cryptosystem, a threshold authenticated encryption and signature scheme characterized by (t, n) shared verification for (k, l) signature was put forward. Information m is transmitted to a particular verifier company, and the signature is then verified through the cooperation of t ones from the verifier company with n members after being signed by a signer company employing (k, l) threshold signature scheme. Similarly, by integrating (k, l) threshold signature scheme with the message recovery technique and after being encrypted singed by any k ones from a group with l members, information m was transmitted to a particular verifier group with n members, and then recovered through the cooperation of any t ones from the verifier group. Application examples of the company encryption communication system and the generation of the polynomial of the company private key and public key were given. The security of this scheme is based on Shamir threshold scheme and the safety of Elliptic Curve Cryptosystem. The scheme realized a directional transmission between different companies, and due to the advantages of Elliptic Curve, it enjoys wider applications in practice.

Since structured interleavers perform better than random interleavers for short block, a structured interleaver used in turbo codes for services requiring less latency time is preferred. Thus, we propose a new method to design interleavers for turbo codes[1] with short block. The designed interleaver is generated through ascertaining adjacent information bits distances and position of the first bit, which guarantees large minimal self interleaving distance and minimal correlative interleaving distance. Compared with other interleavers, the interleaver is able to minimize the negative impact of noise bursts and is easy to implement.

In order to solve the problems of accuracy and availability in traditional network trust evaluation, a new trusted network management model based on clustering analysis was built. The model evaluates the trust value in users’ behavior based on behavior expectation. Through clustering analysis of users’ past behavior data, the network manager can determine the user behavior expectation, which is then used to evaluate the trust in users’ behavior. Our model can manage the network based on users’ trust value. Experimental results have shown that our model can rapidly detect and isolate malicious users and accurately differentiate trusted users from malicious ones. In conclusion, our model has achieved the purpose of improving network security.

For uniform consumption of network energy and enhancement of network lifetime, clustering has gained much attention in wireless sensor networks. Clustering can potentially decrease energy usage and improve the network life time by power control and data aggregation. In order to make clustering more energy efficient, various researchers have proposed putting idle nodes in sleep mode, allowing significant amounts of energy to be conserved. In this paper, we proposed a multi-level sleep scheduling technique which is based on exponentially smoothed average in clustered heterogeneous wireless sensor networks. The duration of sleeping is calculated by the monitoring the inter arrival time between the two packets transmitted by node to cluster head. The experimental simulation has shown that the proposed method significantly enhances the network life time by conserving the energy of nodes otherwise wasted in idle mode.

The wireless energy transfer technology has evolved as a reliable approach to solve the energy limitation problem for Wireless Sensor Networks. Major work has been carried out to recharge only the static sensor node through wireless charging vehicle. Some of the wireless sensor networks applications require complete mobility and the proposed techniques cannot be applied to mobile sensor networks since the positions of the nodes are changing at regular intervals. This paper extends the prior approach in mobile sensor networks through which the lifetime of the mobile nodes have been significantly enhanced. The charging schedule of the nodes ensure that the nodes never run out energy.

Combined with the technology of network communication, embedded computer, data processing, device diagnosis and maintenance, Intelligent Maintenance Unit (IMU) is designed for remote monitoring, diagnosing, maintenance and controlling devices where it is embedded into the equipment to be monitored. To enhance the information process and maintenance ability of agile manufacturing systems, we proposed the IMU. The primary architecture and key technique of the Internet-based IMU is based on the research of new local and overseas products. We also constructed remote monitoring and maintenance system with IMU. Key technologies and Internet-based maintenance system integrated with multi–sensors fusion are also analyzed so as to meet the high standards of reliability, networking and intelligence of a monitoring system. Our experimental results have shown that our proposed embedded smart agent provided a sound and reliable emaintenance mechanism for computer-integrated manufacture (CIM), and sustained the value of products/equipment during its life-cycle. Finally, the application field and future tendency of IMU-based maintenance technology are presented.

We put forward a novel quantum secret sharing (QSS) protocol based on round-robin differential-phase-shift scheme (RRDPS), which achieves the distribution of a full secret key to one party and two partial keys to two parties that neither of the two partials alone can decipher the cryptographic message by full key. Making use of the advantages of round-robin differential-phase-shift, the protocol has a high tolerance of bit errors and enjoys a constant amount of privacy amplification and inherent randomness without precise signal disturbance estimation, especially when a finite key effect is considered. In terms of configuration, it is a simple modification, a variable-delay interferometer replacing the one-pulse-delay counterpart, of differential-phase-shift quantum secret sharing(DPS-QSS).

In this paper, we designed a system of scientific and technological project based on classification. Our system uses Check Weight Algorithm to extract keywords and Solr distributed index database of Lucene engine as the core to provide full text indexing and search open source enterprise platforms. This design enables our system to effectively improve the query speed and reduce computational complexity of the traditional query. Based on semantic technology project similarity calculation, it can also hasten the speed of the traditional rechecking system and improve the accuracy of rechecking the effect.

In this paper, an improved approximation formula is proposed to calculate the symmetric capacity of the 2-dimentional constellations under additive white Gaussian noise (AWGN) channel with the algorithm complexity reduced significantly. The parameters of the formula are evaluated by using the Levenberg-Marquardt (LM) and the Universal Global Optimization (UGO) algorithms. The simulation results show that the analytical solution calculated by the improved formula corresponds with the complex symmetry capacity evaluated by the Gauss-Hermit integration.

In this paper, we study the opportunistic relay selection for secrecy enhancement in dualhop network which uses decode-and-forward protocol in the presence of a passive eavesdropper. To deal with this challenge, we consider the selection of relay that has the best instantaneous signal-to-noise ratio to the main receiver as the forwarding node, while another relay with the lowest instantaneous signal-to-noise ratio to the main receiver is considered as the jamming node. The proposed scheme is analyzed in terms of average throughput and secrecy outage probability. Our simulation results show that there is a significant improvement in both secrecy throughput and secrecy outage probability with the proposed scheme.

This article proposes designing algorithms and setting up a mathematical model of stocks’ running with classical physical laws. By applying laws of kinetics and dynamics to the motion of stocks, a new model is put forward to predict how a stock is going to rise and fall. Experiments and a large amount of practical facts showed that the model and the algorithms are more accurate and reliable than the present ones, which can be beneficial to investors. This article is valuable for the designing of program software to analyze stocks.

Traditional recommendation algorithms not only cannot provide satisfactory real-time recommendation service for Web consumers, but also ignore the shift of user dynamic interests. Thus, this paper introduces the behavior characteristics of user dynamic interests, from which we established the dynamic interest preference extraction model of Web users. The corresponding recommendation algorithm considers the dynamic user interest. Our experimental results show that this method can significantly improve the recommendation efficiency, accuracy and satisfaction of Web consumers.

When exploiting indoor localization, it is known that there are two most important factors: how accurate and how fast users can obtain the result. Quad-tree algorithm is one of the most creative and novel method that aims to improve the localizing speed and accuracy and is deemed to be more superior to traditional n*n indoor area gridding method. It divides the region into 2*2 grids, and determines one target to be divided into 2*2 grids again and again. With a level to level advancement method, the accuracy will significantly improve. However, the regularly and evenly segmented geometry grids for the Quad-tree algorithm cause particular problems. Experiments show that a boundary error with a low probability of occurring but one that inflicts a lot of damage cannot be ignored. In this paper, we managed to come up with a solution, which is to use the irregular segmenting method based on Multi Logistic Regression to address this error. Also, we further our research to explore how we can determine the threshold of gridding based on the tradeoffs of accuracy and precision, so as to maintain the highest accuracy. Our experimental results have shown that our revised algorithm reveals at least 52% more results that leads to errors within 1 meter, causing the percentages to be up to 80%.

The current battlefield situation visualization field of military intelligence is the focus of the study, the visual effects to the commander's decision. According to the physical model of the physical process of battlefield explosion, we proposed to improved simulation model, using Eulerian simulation of instantaneous explosion fireball model, using the explosion of fluid physics volume as a function of spatial location and time, then using the particle system method to simulate the explosion after the spatter model and explosive fluid splash material point in the movement process of various physical quantities with the particle to the variation of the position and time recording, tracking all the fluid particles. The separate explosion simulation process, which can simulate the instantaneous explosion of the overall effect, and could clearly simulate sputtering spatter particles. Experiments show that this method is realistic, can satisfy the real-time simulation of the explosion.

From the perspective of agile learning, the cognitive student model is an important means to recognize students’ knowledge level and make an instructive learning. Due to the existence of several cognitive factors in this model, which can be viewed with strong uncertainty for properties with gray, fuzzy and difficulty quantifying, this paper presents a gray partial correlation method to quantify this model. Firstly, the mutual influence is discussed between any two factors to get the weight vector by calculating the partial correlation coefficient. Then we define the gray classes and white nization weight functions to output gray evaluation weights so as to build comprehensive evaluation model that combines with the weight vector. The model can give a guidance on learning catering exactly to the agile learning.

Due to continuous improvement of medical information construction, the scale of medical data is expanding, and its value will continue to increase. However, the traditional storage architecture cannot store such large amounts of medical data. Not only that, the common mining algorithm cannot analyze the value of potential information, largely hindering the development of medical wisdom. According to the concept and characteristics of smart medical, this paper proposes an architecture scheme of a medical cloud platform. Based on MapReduce, our Apriori medical data mining algorithm is conceived by introducing a degree of interest to improve the classic Apriori algorithm and combining with the knowledge of the cloud. The algorithm can ease the development bottleneck encountered by current smart medical. Overall, the proposed technology is more efficient and is able to apply itself more extensively in the field of medical research.

Based on k-means clustering, we designed and implemented an improved genetic algorithm(IGA) to find the best solution for the traveling salesman problem. IGA includes three characteristics of classical genetic algorithm(CGA). The first characteristic is the use of k-means clustering method on all cities to reduce the complexity of the problem by dividing the cities into several groups. The second characteristic is the adoption of genetic algorithm in each group for optimizing the sub-path. The third characteristic is the use of evolution operation on the inter-group for integral optimization. We performed many experiments to test the ability of IGA. Our experimental results show that IGA is more effective than CGA, especially for large-scale traveling salesman problems.

To solve the problem of extraction algorithm of high complexity for large-size images feature, sketch retrieval and compressed sensing theory is fused in this paper. The theory on a small amount of measurements can accurately reconstruct the original signal, thus, HOG algorithm based on compressed sensing is presented. Firstly, the algorithm is needed to preprocess and block in sketch retrieval. Then by means of the compressed sensing principle of block image feature extraction, a small amount of compressed image features are obtained. Finally, we used the weighted distance method to calculate measurement features of similarity, which is then used to implement the image retrieval. Our experimental evaluatio which uses two basic benchmarks shows that our proposed algorithm is more accurate in image retrieval than other algorithms.

In collaborative task systems, resource service sequences can complete more complex tasks and provide more efficient and better services than single resource services. During a business process execution, resource services are used in a certain sequence, which is called Resource-Service Sequence (RSS). A RSS mining and optimization approach based on Genetic Algorithm (RSS-GA) is proposed. In our approach, the degrees of temporal relationship between two resource services are first resolved via historical data. Initial RSSs, which are represented as a temporal constraint matrix, are then obtained. Next, a genetic algorithm is used to optimize the initial RSSs. Lastly, a simulated data set is used to test our algorithm and the experimental results prove that our approach is stable, consistent and effective.

The performance of key-value handling in database is not fast enough in situations with large-scale data sets and queries. Database searching is usually a type of data-thick application, which operates on relatively simple logic when handling requests and can be accelerated on the Graphics Processing Unit (GPU). In this paper, we designed a new data structure named as GSkiplist that can be implemented on GPU effectively. We also described algorithms used in database operations such as inserting, deleting and searching. We chose OpenCL as our programming platform and the experimental results proved that implementing GSkiplist on the GPU can achieve speeds up to 8 times the speed when it is implemented on the CPU.

Multi-label learning is one of the most important branches in the data mining field, and many algorithms have been proposed to solve many recent problems such as automatic annotation for multimedia contents, web mining, tag recommendation, etc. However, most of the classic methods have a hypothesis that the labels are independent, which may not hold in every situation, especially in tariff package recommendation where packages are always connected. In this paper, we present a novel algorithm which has a selfadaptive mechanism to implement the correlations between labels. We also analyzed the performance of the framework on tariff package data set. The simulation results show that the novel algorithm achieves better results compared to classical methods in both theory and practice.

The existing market schedule strategies in a cloud environment mostly seek to maximize income without paying sufficient attention to the credibility of service suppliers. The market schedule model based on credit price and cost computing is proposed for the purpose of income maximization of service suppliers and good credibility among users in combination. The model introduces credit evaluation of service suppliers into the task preemption scheduling mechanism based on cost computing, employing economical concepts including credit evaluation coefficient, credit price, task penalty and income discriminant function. It also uses the discriminant function value of service suppliers’ income for decision making in task preemption. The simulation experiment shows that the model achieves income maximization of service suppliers and schedules the tasks with low credibility first.

With the development of smart grid interactive, the database interaction between internal and external information is increasing. In this paper, we proposed a detailed design of a method of SQL access log compression for power business system, so as to respond to the demands of database security audit of the power system. This method can achieve the compression of SQL access log and also has the function of simple classification for SQL logs. The test results show that the method can achieve a large margin compression of SQL access logs and meet the requirements of log preprocessing for the database behavior audit of power business system.

Forecasting of stock market trends has always been an important issue in the field of finance. Neural networks are useful for understanding non-linear relationship between historical data and stock price changes. In this study, recurrent neural network (RNN) is used for time series data. The dataset contains large amount of intraday data from the constituent stock of China Shanghai Shenzhen 300 Index. Using the intraday data, we created and classified different daily features for the RNN model. Instead of using accuracy as a parameter to measure the effectiveness of the model, precision and average profit are used. Our experimental results demonstrate that this RNN model performs well with large amounts of testing data, which can provide useful information for stock market investors.

Since medical insurance is relevant for everyone, it is very important to know how to study the formulation and implement the system. In this paper, I propose a text association analysis method, TACMA, which is based on medical insurance policies by combining the text information of medical insurance policy with the characteristics of semantic relations of the audit knowledge. Compared with the traditional TP-growth algorithm, the TACMA algorithm has improved significantly in mining effectiveness and time efficiency in text knowledge discovery, and can be successfully applied to practical projects.

With the development of computer and database technology, people have a multitude of ways to access information. However, the rapid increase of big data information has created new problems for people. Due to the vast amounts of monitoring data and ineffective extraction of intrinsic data, more and more information are ignored. There is demand for user-oriented growth in data analysis, which is a technology that presents data in a more graphical way for easier visualization. This data visualization technology will impact users greatly and aid them in making decisions in the future. By combining the D3 visualization library with practical applications in the field of health care, data visualization can be used to effectively explore health care research.

This paper presents a new CPU scheduler to facilitate concurrent applications in Xen. The new scheduler uses the completely fair scheduler to reduce the gap between concurrent and non-concurrent virtual machines (VMs), and to balance physical CPU time distribution and resource utilization. It uses the red-black tree to shorten the virtual CPU (VCPU) lookup time, achieve efficient task scheduling and to reduce the runtime. To assist with concurrent VCPU synchronization and scheduling, we set a concurrent waiting queue which can handily pick up marked VCPUs for execution. Extended simulation runs are conducted to evaluate the performance of our proposed scheduler and the simulation results showed that our scheduler outperforms existing schedulers in reducing the runtime for both concurrent and non-concurrent applications.

This paper studies the laser rotating precision fuzzy PID computer control system, using laser technology combined with computer intelligent control technology, the control technology of stepper motor control method based on fuzzy PID control algorithm of fuzzy computer, PID computer, PID computer fuzzy controller is designed for stepper motor, stepper motor built computer control system also, the stepper motor control system was analyzed by MATLAB computer simulation in MATLAB/Simulink environment to build the stepper motor fuzzy PID computer control model, the implementation of the MATLAB computer simulation was established based on the laser rotary precision machining machine computer intelligent control system, with the elimination of the stepping motor start and stop. The swing, running well, precision rotating laser processing.

Cardiovascular disease has become one of the most serious diseases which not only endangers human health, but also causes death. The microcirculatory hemodynamic parameters contain abundant cardiovascular information. Through hemodynamic detection, cardiovascular health status can be understood. Results: The data measured by the fingertip pulse wave detection was transmitted to the mobile phone via Bluetooth. The data was obtained at the phone of the Android platform and finally calculated to the relevant microcirculatory hemodynamic parameters. Using this program, users can easily learn about their physical condition through their mobile phones. This equipment allows users to pay attention to their own health at home and improve their health consciousness. Not only that, it also makes the allocation of medical resources more efficient.

Due to slow adjustment process of production parameters of food production line control system, production efficiency is slowed down, impacting food production schedules. Thus, in order to improve the control efficiency of food production line and to ensure rapid development of the food industry, this paper focuses on the food production line of computer control system. The production line of the thousand layer cake is analyzed in this paper, where the hardware and software aspects of food production line computer control system are delved into for detailed design and analysis. Research results show that improvements in design of the food production line computer control system can effectively improve the efficiency of food production and management.

In this paper, an extended permission is introduced. In the permission model, the permission is split into the permission component by analyzing the relationship between the permission and the operation. Quantified permission component is spilt from permission, and the permission component is calculated. The size of the permissions of the two roles is then compared. The size of permissions is compared before authorization, and the authorization operation is performed only if the granted permission is greater than the current permission. This reduces the workload of both the authorized operation and the administrator.

This paper focuses on a queuing model of a multi-skill call center in M-design. In this model, there are two types of customers and three groups of servers who have different skills. We obtained the state-transition rates by using results from an M/M/c/c loss queuing system. We then established equations for the steady-state probabilities of the system. Following that, we obtained the computational formula for the service level and established the staffing calculation model for the optimal number of agents in each group. Subsequently, we adopted the implicit enumeration method to find the solution, and through the numerical examples, we investigated the effects of model’s parameters on the results.

With the development of the dynamic analysis algorithm and behavior detection method, it has become increasingly important to design an effective system to manage and control behavior detecting resources. This paper introduces a design and architecture of a resource management system to manage the virtual environment for malware behavior detection, including its resource management algorithm, network design, system architecture, et al. The design introduced here focuses on the high standards placed on performance, reliability and scalability of malware behavior detection system. From the results obtained from both experimental simulations and real world implementation, this system has proved to be competent and fulfil those requirements.

With the increase in popularity of Portable Document Format (PDF) documents and increasing vulnerability of PDF users, effective detection of malicious PDF documents becomes a more and more significant issue. In this paper, we proposed a method for the detection of JavaScript-bearing malicious documents and established the prototype detection system. We de-obfuscate the JavaScript code extracted from PDF documents through static analysis, and emulate the code execution during dynamic analysis. The combination of static and dynamic analysis in our approach makes the detection immune to obfuscation. Our experimental evaluation shows that our method can detect a broad range of malicious PDF documents and markedly enhance the detection accuracy with an acceptable overhead.

Focusing on the defects of the traditional artificial meter reading system, the paper designed a set of intelligent remote meter reading system. The system has core components such as collector and concentrator which can cut the cost of meter copying and enable water management to be concentrated, reliable and convenient by using correct networking solution and communication modes. With mobile apps that connect to public network, users can also view their water consumption and cost data. This paper will introduce the hardware architecture and key modules of the intelligent remote meter reading system. The measurements made by the proposed system are not only accurate, but installation and maintenance are also relatively easy, which meet the demands of system design.

Banana, as an important-part of the fruit project in China, is a pillar industry of tropical agriculture, which is also a major source of farmers’ income. The change in the field environment has a great effect on the production of bananas. The key environment factors and characteristics which affect banana high efficient cultivation, are yet to be resolved. We realized that some environmental changes may affect banana yield, especially natural disasters such as typhoon or drought. A kind of mathematical model which can reflect the relationship between banana cultivation and field environment should be built. Based on the problem, we will systemically research the monitoring model between banana growth and field environment and focus on the key factors. We will analysis monitoring model characteristics in order to reveal the change law of the environment parameters of banana growth. The basic data and framework can be provided for banana expert system. Results in this thesis will set up the foundation for serving relevant government offices, farmers and consumers.

As the medium for direct interaction with various participants, internet-based Peer-To-Peer (P2P) lending platforms play a vital role. To assess their operational quality in a relatively objective way, we established a theoretical basis and came up with an indicator system to evaluate the operational quality of P2P platforms. This was done by consulting local and foreign documents and by collaborating with researchers on current developments. We also analyzed real-time data of 75 internet-based P2P lending platforms via factor analysis and drew the conclusion that the indicator of website history and scale is the biggest contributor. Thus, we proposed relevant suggestions to regulatory institutions, internet-base lending platforms and relevant investors.

In order to solve the problems of high table memory access, an efficient table memory access optimized algorithms arepresented for CAVLC decoding in H.264/AVC in this paper. The contribution of this paper rests that we use hash table query technology in table look-up to lower high table memory access by reducing high table memory access in H.264/AVC. Specially, after finding the internal relationships between code length and numbers of 0 in code prefix, we use hash table query technology in table look up to reduce lots of table memory access. The simulation results show that our proposed schemes based on hash query can lower about 25% table memory access for CAVLC decoding compared with TLSS method.

The Internet has experienced rapid growth in capacity, access device number and applications. However, issues such as network security, energy efficiency and network flexibility have arisen. To address these issues, a networking investigation platform, NetFPGA, is set up. This paper analyzes different research aspects of the NetFPGA and presents our findings about the NetFPGA platform.

Complex Event Processing (CEP) often involves processing huge amounts of information in real time on the fly. In this paper, we present a new approach to deal with the uncertainty in complex events in terms of complex event patterns and provide an algorithm to demonstrate the improvement for Complex Event Processing (CEP). Two complex event detecting principles, Strong Match Principle and Weak Match Principle,adopted in different situations, were studied. Our results show that if basic events occur in random, all basic events in a rule may not completely occur in real world situations. Type I error and Type II error are also discussed when weight was introduced to support fuzzy reasoning to deal with the uncertainty in CEP. We have found that a good uncertain CEP detecting system should not only indicate which event occurred, but also when the event is going to happen with fuzzy reasoning and forecasting functions.

In this paper, an integrable differential-difference equation is investigated based on its Lax representation. N-fold Darboux transformation (DT) are constructed with the aid of computer symbolic computation. As applications of N-fold DT, N-soliton solutions in terms of determinants are given. Also, soliton elastic interaction behavior of those solutions are shown graphically, which might be helpful for understanding physical phenomena.

For aviation enterprise collaborative design process involving a large number of complicated design knowledge, knowledge leads to cumbersome screening, the use of low efficiency. The knowledge management modeling approach is proposed in aviation enterprise collaborative design. In the analysis of aviation enterprise knowledge management infrastructure on collaborative design, collaborative design management is analyzed on the basis of given aviation enterprise collaborative knowledge management model design, and an aviation enterprise collaborative design mode is given. The method solves the low utilization efficiency of collaborative design knowledge, and looks for such knowledge complex issues to meet the needs of designers.

The following section is included:

• Author Index