ZHU Yun-min, ZHANG Li-wei, WANG Sheng-yuan, DONG Yuan, ZHANG Su-qin
Vol. 37, Issue S1, Pages: 1-6(2009)
摘要:As the multi-core processor is widely used and advanced high-trusted software is required,the verification of parallel programs for multi-core processor becomes more and more important.This paper presents a proof framework about the verification of parallel programs,including the definition of our abstract machine,the formal specification for object code,logic inference rules and the proof of soundness theory.The classic spin lock technology is introduced to implement the mutually exclusive access to shared memory.Our proof framework supports Hoare-logic style reasoning.In addition,we use high-order logic to describe both operational semantics and security policy.Pro grammers can verify the partial correctness of multi-core parallel programs in our framework.
摘要:This paper presents an application-level multicast using a tree-bone overlay and an auxiliary mesh.We firstly define the strictly simple reso urce tree-based overlay,and then propose the relaxed simple reso urce tree-based overlay according to the observation of self-contained performance of peers in pervasive environment.Next we analyze the churn of tree-based overlay and the resulting issue of bandwidth bottleneck.With an auxiliary mesh and state transitions based on the dynamic size of bandwidth,we design an adaptive bandwidth strategy which enables the multicast better bandwidth utilization and usability.Experimental results show that the application-level multicast achieves a good performance.
摘要:Eye-tracking as a new method of human computer interaction has the merits of naturalness,efficiency and easy intent under standing.The authors have designed a Desktop Eye-Tracking System which used the technique of image processing to obtain the user's gaze focus in virtual environment scene.Based on the theory of pupil center corneal reflection,the authors improved and realized a highly efficient linear prediction algorithm for real-time eye gaze image processing.Experiments proved that this algorithm can well support real-time eye-tracking as an interaction modality for desktop virtual environment.
关键词:eye-tracking;human computer interaction;linear prediction algorithm;pupil center corneal reflection
CAO Yuan-yuan, HUANG Fei-yue, TAO Lin-mi, XU Guang-you
Vol. 37, Issue S1, Pages: 16-21(2009)
摘要:Recognition of actions in daily living is challenging because: (1) actions are continuous; (2) human location is changeful; (3) human body is partially occluded sometimes.This paper proposes a multi-view framework for on-line recognition of actions in daily living.Action representation based on "Envelope shape" enables view-invariant action recognition.A sliding window concatenates the feature vectors for action representation into a stream as the input to a bank of HMMs.A maximum likelihood based classifier detects action.The HMMs are chosen by an ontology based environment knowledge model.Besides,occlusion detection and action representation in occlusion are also discussed.Implementations demonstrate the efficacy of our approach.
摘要:Software documentation is usually expressed in natural languages and free text,in which it captures large useful information.Establishing traceability links between documentation and source code can be helpful in Software Engineering Management.Currently,the recovery of traceability links is mostly based on information retrieval techniques,e.g.,probabilistic model,vector space model and Latent Semantic Indexing (LSI).But previous work only treats documentation and source code as plain text files without considering the features with respect to Software Engineering.Four enhancing strategies are proposed to improve the traditional LSI method based on the features of software documentation and source code,namely,source code clustering,identifiers classifying,similarity thesaurus and hierarchical structure enhancement.Experimental results show that the four enhancement strategies can increase the precision by about 15%.So,the special characteristics of documentation and source code should be considered carefully during the recovering traceability links between them.
摘要:With the development of network application,network traffic is constantly increasing,and the web application testing is becoming more important.According to statistical results,the network traffic is self-similar in the time scale.Therefore,the research of self-similar traffic model is very important.Network traffic simulation has limitation by single computer.So this paper improves the Fraction Gaussian Noise (FGN) model so that it becomes a high-speed parallel algorithm to generate self-similar traffic.Using the improved algorithm,we propose a distributed system and present the design and implementation of this system.The paper also includes the work of achieving a network traffic detection tool.The results show that the testing samples generated by the distributed system have self-similar property.
关键词:self-similar model;network traffic simulation;distributed system
WEI Da, JIN Ying, ZHANG Jing, ZHENG Xiao-juan, LI Zhuo
Vol. 37, Issue S1, Pages: 36-41(2009)
摘要:Safe execution of untrusted mobile code is one of the most important problems in mobile code security.Model Carrying Code(MCC) provides a model-centered framework for safe execution of untrusted mobile code by taking both code producers and code consumers into consideration.Security policy specification and enforcement are important parts of MCC.An approach to formal specification and runtime enforcement of security policy has been proposed and implemented for Java mobile code running on open source JVM Kaffe and Linux operating system.Some useful works have been done on specification of security policies,the enforceable Extend Finite State Automata(EFSA) model and the method of monitoring on process level to catch system call information.It is indicated that our approach present a effective way to support MCC and implement security policy enforcement.
关键词:model carrying code;specification and enforcement of security policies;open source JVM;process level monitoring
HUANG Pei-jie, ZHU Li-hua, LIU Xue-hui, WU En-hua, WANG Chuan-ming
Vol. 37, Issue S1, Pages: 42-50,56(2009)
摘要:We present a novel technique for efficient visual communication,where a compact representation is abstracted from arbitrary image sequences automatically.We aim at improving the entire visual communication process,including transmission of video data and perception of visual signal by human eyes.Therefore,guided by some related theory in Psychology,our work increases both the compressibility and comprehensibility of images.This goal is achieved by two steps.One step is an edge extraction process for preserving object boundaries which are believed to be the most sensitive features to human vision,and the other step is a non-linear diffusion process for reducing extraneous details.Both are designed in a spatiotemporal manner to guarantee the temporal coherence of resulting animations,while real-time processing speed is maintained by facilitating parallel computation on a GPU.We additionally build a complete visual communication system with the proposed algorithm as its core to demonstrate the practicality of our technique.Experimental statistics collected on the system indicates that transmission bandwidth is obviously saved while perceptibility of image sequences is improved.
摘要:Researches on requirement engineering focus more on the requirement analysis and requirement specification,while less on requirement elicitation.In this paper we propose a scenario-based collaborative requirements elicitation approach for Enterprise Information Systems(EIS).In this approach,we use scenario to organize different stakeholders,to facilitate the collaboration among them and thus change the elicitation pattern from traditional requirement experts centered way to stakeholder-equal way.This approach proposes the elicitation steps as: the stakeholders should first construct the as-is scenario,then to-be scenario and use-Case,and elicit the requirement based on the to-be scenario..We also present how to extract scenario-role-data view from natural language description and activity diagram from semi-formal description,and use these two views to validate the completeness and conflict free of the scenario descriptions and the users' requirements.
关键词:scenario;collaborative;requirement elicitation;enterprise information system
摘要:We propose a novel distributed component framework named ACF,which introduce a dynamic contextual composition mechanism based on shared coordination object,contrastively based on shared coordination data traditionally.ACF directly supports explicit expression and dynamic evolution of coordination logic,by applying runtime software architecture.At present,ACF has been preliminarily implemented and employed in an integrated development environment (IDE) name Artemis Studio.
HAO Xiao-lei, DONG Meng-gao, MAO Xin-jun, QI Zhi-chang
Vol. 37, Issue S1, Pages: 65-69(2009)
摘要:The self-adaptation logic and enterprise logic of self-adaptive systems are often tangled with together in existing approaches to engineering complex self-adaptive system,which makes it difficult and complicated to develop and maintain self-adaptive systems.We believe autonomy is the basis of self-adaptation and it is necessary to separate the self-adaptation logic and enterprise logic of self-adaptive systems in order to simplify the development of such complex systems.A Self-adaptive Agent strategy Description Language SADL is therefore designed to specify the self-adaptation of systems.Self-adaptive agent can sense the changes of the situated environment and dynamically execute self-adaptive operations such as join,quit,deactivate and activate in order to adapt to the environment changes.The paper introduces the syntax and semantics of SADL and the design and implementation of its compiler.
关键词:agent technology;dynamic binding mechanisms;self-adaptive agent strategy description language
PAN Gang, LI Tong, REN Hao-yi, LI Shi-jian, YAO Min
Vol. 37, Issue S1, Pages: 70-74(2009)
摘要:Pervasive computing puts forward a number of challenges on reasoning engine,such as: to be context-driven,to deal with frequently change of contexts,and to work in a real-time manner.Traditional reasoning engines can hardly meet these new requirements.In this paper,we present a context-driven reasoning engine ScudCORE (Scud COntext-driven Reasoning Engine),which takes the first order predicate logic as basis.When contexts vary frequently,ScudCORE reduces the reasoning problems of excessive frequency and inefficiency by the context filtering mechanism.We employ ontology to make the reasoning engine convenient to share,reuse and understand context information both inside and outside it.Also we increase its flexibility with a dynamic rule management module.
摘要:In this paper,structured laser pointer (SLP) is proposed to sufficiently express the movement of user's hand.Firstly,the SLP is designed as input device,and then the corresponding signal processing application is programmed to filtrate the sequence of laser spots,and extract the trajectory of the laser spots and user gesture from the sequence.And the laser trajectory and user gesture are translated into original interactive language to increase the semantic interaction.Using a SLP can acquire a target quickly because the dwell of laser spots are not required.Instead,we search the plain section in the sequence of laser spots to quickly and accurately select the target that people want to point to.Finally,according to Fitts' law,a user study is designed that requires participants to achieve several tasks.It proved that SLP performs much better than traditional laser pen in speed,and using SLP can also express more semantic interaction.
关键词:structured laser pointer;rotation gesture;plain section;user study
WANG Xiang, GUO Yan-wen, DU Zhen-long, WU Gang-shan, ZHANG Fu-yan, PENG Qun-sheng
Vol. 37, Issue S1, Pages: 79-86(2009)
摘要:We present a new approach for automatically adjusting the brightness of under-exposed digital images and video sequence.Our approach employs "divide-and-conquer" scheme with gradient domain operation.The under-exposed image is first segmented into different regions according to the brightness.We then compute the brightness enhancement function for each region.Finally a constrained energy function on gradient domain is solved for preserving continuity among different regions.The algorithm is further extended to video.Several key frames are selected from the input video.We adjust them using the image brightness adjustment algorithm.Afterwards,through optical flow,we find the relationship of the regions between the key frames and intermediate frames,and use such relation to adjust the brightness of intermediate frames.Our approach works effectively for dealing with spatially non-uniformly exposed image and temporally non-uniformly under-exposed video.Meanwhile,detail information,such as strong structures as well as textures,is faithfully preserved,as demonstrated by experimental results.
摘要:In a group signature scheme,any legal member of this group can sign on behalf of this group,and no one can find out which member is the actual signer except the group manager.The focus of this paper is to design a new ID-based group signature without trusted PKG (group manager) or random oracles,and the security of the scheme,such as forward security.According to the analysis result,legal group members can sign on behalf of the group; an arbiter (OM) can open a legal signature to find out the actual signer,and he can also point out that whether the group manager is legal or not.So this is a scheme without trusted PKG without random oracles.Besides,this paper takes advantage of the information of time section to implement the forward security of this scheme in case of key exposure or revocation of group members.
关键词:identity-based group signature;without random oracle;without trusted PKG;forward security;bilinear pairing
摘要:Blind mixing matrix recovery is one of the most important steps in blind separation of sparse sources,which impacts significantly on the recovery accuracy of source signals.A novel improved K-means clustering algorithm is proposed based on differential evolution,to avoid the partial convergence problem of the K-means algorithm.The proposed algorithm is applied to allocate the sparse mixture data to several clusters,thus guaranteeing the robustness of the clustering.Then the cluster centers are amended through Hough transform to recover the mixing matrix.Experimental results show that the proposed mixing matrix recovery algorithm has advantages of high robustness and accuracy compared with conventional algorithms.
摘要:We achieve a dynamic update of the authentication-tree through creating the sub tree and dynamic update of the key.In the original signature,the number of the signature influence the scale of the authentication-tree,and the increase of the number of signature makes lower efficient.In our scheme,the number of signature is not affected by the scale of the authentication-tree.In addition,the paper do a safety analysis to the impro ved authentication Merkle tree signature,the results show that the signature of the improver Merkle authentication is as safe as the original Merkle authentication signature tree,and as used key management according the time,the The signature also has forward-secure.
摘要:Threshold signature is designed to distribute the power of signature,therefore it is more secure than normal single signature.Most of the existing threshold signatrue schemes are provable secure in the random oracle model.An identity-based signature (IBS) is proposed from the bilinear parings based on Paterson's signature scheme In the proposed threshold signatue scheme,a trusted private key generator (PKG) is required to generate and manage the private secret key.The security of the signature scheme is proven in this paper.We show that it is strong and resisted against adapt chosen message attack.
关键词:identity-based;threshold signature;without random oracles
摘要:Focusing on model-driven development for collaborative systems,a new description approach for the formal semantics of architecture models is proposed by combining category theory with algebraic specification and process algebra.On the ba-sis of component specification,morphisms are used to describe the relationships between components,and the morphism types imply the different semantics of component relations.Thus architecture models are described within typed category diagrams,and functors are used to describe the mapping relations between different levels of models.The formal approach can be used to judge whether a transformation satisfies some property preservation constraints or not.A collaborative editing system is given as a case to illustrate the application of this approach.
LIU Chun-xiao, PENG Qun-sheng, YANG Ying-zhen, WANG Jin, CHEN Wei
Vol. 37, Issue S1, Pages: 112-117(2009)
摘要:We present a new image completion method based on a large displacement view (LDV) for faithfully repairing large missing regions on the target image with complex structure information.A coarse-to-fine distortion correction algorithm is proposed to minimize the perspective distortion in the LDV image,which performs following three steps in order,i.e.homography based LDV warping,energy optimization of overlap correspondences and energy optimization for hole filling.Then,the rectified LDV image is used to restore the missing pixels.We finally eliminate the ghost effect between the repaired region and its surroundings by Poisson image blending.Experiments show that our method outperforms recent state-of-the-art image completion methods.
WANG Zhang-ye, CAO Mei-xuan, LI Li, PENG Qun-sheng
Vol. 37, Issue S1, Pages: 118-124(2009)
摘要:We propose a new image based individual prototyping method for facial aging image synthesis.Based on the human facial aging image database of the yellow race,we present an individual prototyping match algorithm based on the local standard deviation of curvatures of facial contour.By calculating the local standard deviation of those curvatures,we match the input facial image with several corresponding ones from the face database and synthesize the prototypes with texture enhancement.By the shape and color transformation model we generate the aged facial images.Our method can realize several important aged features such as wrinkles,pouches,darken skin,and somewhat white hair and eyebrows.Experiments show that our method can be easily manipulated and synthesize the anile facial images at different ages for various kinds of people realistically.
摘要:Because distributed system had features of physical distribution,simultaneo us access,being sensitive to sequence and platform isomerism,distributed test system must fit tho se features.TTCN-3 distributed test execution platform was able to fulfill requirements of distributes test.However,with the feature of static binding test adapter,TTCN-3 test system can not communicate with different kinds of SUT sata single test execution node.Therefore the concept of test adapter framework was present in this paper.On the basis of fundamental adaptation with supporting TTCN-3 standard interface,the framework integrated test adapters for different kinds of SUTs,dynamically selected and used adaptive test adapter.As a result,the framework fulfills requirements of TTCN-3 distributed test execution platform,and strengthens the ability of TTCN-3 for distributed test domain as well.
关键词:TTCN-3;distributed test;TTCN-3 distributed test execution platform;test adapter;test adapter framework
摘要:The detector generating and matching algorithm is important for efficiency and accuracy of the artificial immune system.In order to reduce the time and space consumption of detector generating,and improve efficiency and validity of detector,we present a cutting based detector generating and matching algorithm.Using super cube to present detector and providing the basis of reducing matching algorithm overhead.Using the spatial inclusion relationship to express the matching relationship between detector and antigen,then reducing time and space consumption of matching and in favour of analyzing accuracy.Cutting space to generate detector by location of self in domain,elemiating redundance between detectors,reducing hole of inspecting,and ensuring high efficiency and accuracy of inspecting antigen.Realizing prototype to test and compare with current algorithms.The results prove cutting based detector generating and matching algorithm can improve the performance of the artificial immune system.
摘要:To retrieve the non-free choice structure of the process model quickly and precisely,this paper propose a simulated annealing process mining approach to address this issue.Main contribution of the work includes: (1) Apply the simulated annealing approach under the setting of process mining.(2) Represent process model as "causal matrix".(3) Use heuristic rules to select annealing operations.(4) Evaluate the mining result with a quantitative measurement,incorporate the ideas above into existing simulated annealing algorithm to form an integrated solution.We give experimental results which produced by the ProM,a platform for business process mining.
摘要:Soft ware product line (SPL) can increase the efficiency and quality of software development.Software architecture (SA),as the blueprint of SPL,defines the inter-relationships between components and guides the component composition implementation.However,the existing interface connection architecture is limited to describe the direct interactions between components.It cannot support the more complex interaction situations which emerge with the SPL variability.In this paper,we propose a method of software product line architecture modeling and component composition implementation with extension of aspectual mechanism.The core is an architecture description language (ADL) which extends xADL210 and combines with aspect-oriented techniques.The ADL supports the design and customization for SPL architecture based on variability,and instructs the component composition process for applications.Furthermore,we have developed a prototype tool FdSPLC which provides the visual modeling of architecture as well as the automatic application derivation.
摘要:To reduce the subjective influences and errors from traditional usability evaluation,an approach based on eye-tracking was proposed for the mobile computing user interface.It analyzed the recognition issues in human-computer interaction,and selected fixation durations,numbers,matrix of fixation transfer frequency and other data related to eye movements as measuring criteria.Evaluation strategies were generated for efficiency of visual search and information processing,which explored the domain task and top-down visual mechanism to guide the optimization for interface design.The approach was applied for the usability evaluation with smart phones; as a result,it differentiated usability levels effectively and made consistency with subjective feedback.It was well done to illustrate the feasibility and effectiveness for the proposed approach.
JIANG Yong-xin, WANG Xiao-tong, XU Xiao-gang, HUANG Hua
Vol. 37, Issue S1, Pages: 151-155(2009)
摘要:A method for image enhancement based on light compensation was proposed to enhance the image with imbalanced local illumination.Firstly,to get several different luminance images,original image multiplied by different factors.Then,all images were decomposed by NLEMD.The details of images were added together to enhance texture of image.The illumination of the image was balanced by Retinex algorithm.Lastly,we reconstructed the enhancing result.The experimental results indicate that detail of the image is enhanced effectively and the problem of light compensation is resolved effectively.
摘要:Rhythm is one of the three main factors of music.It's of great importance to analyze and extract rhythm features from music signal which mainly includes beat and tempo.In this paper,firstly we introduce a method which uses autocorrelation phase-entropy,to analyze meter and tempo of music.With this method we get a 97 percent of accuracy degree in music tempo induction on a database of 100 songs(including 50 popular and 50 Chinese folk); Then based on the result of tempo induction,we propose a roughly Bayesian model to keep the beats sequence consistent with the global tempo of music in the music beats locating; Finally we introduce a new beat tracking algorithm using dynamic programming to finish the beat tracking.By comparing its result with others',this algorithm proves to be very effective.