Tracking of human head with particle filter

To cope with the problem of tracking a human head in a complicated scene,we propose a method that adopts human skin color and hair color integrated with a kind of particle filter named condensation algorithm.Firstly,a novel method is presented to set up human head color model using skin color and hair color separately based on region growing.Compared with traditional human face model,this method is more precise and works well when human turns around and the face disappears in the image.Then a novel method is presented to use color model in condensation algorithm more effectively.In this method,a combination of edge detection result,color segmentation result and color edge detection result in an Omega window is used to measure the scale and position of human head in condensation.Experiments show that this approach can track human head in complicated scene even when human turns around or the distance of tracking a human head changes quickly.

Application of fuzzy logic in content-based image retrieval

We propose a fuzzy logic-based image retrieval system, in which the image similarity can be inferred in a nonlinear manner as human thinking. In the fuzzy inference process, weight assignments of multi-image features were resolved impliedly. Each fuzzy rule was embedded into the subjectivity of human perception of image contents. A color histogram called the average area histogram is proposed to represent the color features. Experimental results show the efficiency and feasibility of the proposed algorithms.

Design of acquisition algorithm based on delay-finding detector for indoor GNSS signals

A delay-finding detector is proposed to enhance acquisition sensitivity for indoor GNSS signals. Traditionally, the over-threshold detector on amplitude or peak-to-mean ratio is employed for signal acquisition. Instead, the proposed detector exploits peak positions of successive multiple detections covering a full range of code phase offset to refine the detection performance. Due to each peak position corresponding to an optimal estimate of current code delay, the new detector is named delay-finding detector. On the other hand, a pre-detection processing involved in delay-finding detector aims at confronting the indoor multipath effects. A time domain filter is applied to combine the power of inner-chip multipath components. In addition, a self-adaptive method for dynamically adjusting integration time is designed to further enhance acquisition efficiency in the whole algorithm. The simulation results demonstrate that the proposed algorithm is capable of acquiring the signal when the Carrier-to-Noise Ratio is as low as 22 dB-Hz, fixing the residual Doppler offset at the worst situation. It is capable of increasing detection probability significantly for two-path channel.

供应链系统异构知识集成方法研究

供应链系统流动着大量的异构信息和知识,为了有效地利用现有知识对供应链进行优化管理和运作,供应链不仅要实现信息系统的集成而且需要实现知识集成。本文运用领域本体沦(Ontology)、多Agent等技术研究供应链系统异构知识集成的方法,应用资源描述框架模式(RDFS)作为信息和知识的核心转换技术,实现异构知识的转换和集成,为供应链系统异构知识集成提供一个切实可行的技术解决方案。该方案可以保证全程信息数据和知识的正确性、及时性和完整性,进一步促使信息流为决策人员提供及时、可靠和科学的有效决策支持。

集成工作流与Multi-agents技术的供应链协同管理

提出了集成工作流与软件Agent技术的供应链协同管理框架及方法，通过工作流管理系统及多Agent的设计，应用XML描述的工作流过程规范，由软件Agent实时地执行和重置具体的工作流任务，从而达到供应链协调管理的目的。

Predictive control for mechanical system with backlash based on hybrid model

The mechanical system with backlash is distinguished between a＂backlash mode＂and a＂contact mode＂.The inherent switching between the two operating modes makes the system a prime example of hybrid system.For eliminating the bad effect of backlash, a piecewise affine（PWA） model of the mechanical servo system with backlash is built.The optimal control of constrained PWA system is obtained by taking advantage of model predictive control（MPC） method, and the explicit solution of MPC in a look-up table form is figured out by combining the dynamic programming and multi-parametric quadratic programming, thereby establishing an explicit hybrid model predictive controller.Furthermore, a piecewise quadratic（PWQ） function for guaranteeing the stability of closed-loop control is found by formulating the search of PWQ function as a semi-definite programming problem.In the tracking experiments, it is demonstrated that the explicit hybrid model predictive controller has a good traction control effect on the mechanical system with backlash.The error meets the demands of real system.Further, compared to the direct on-line computation, the computation burden is reduced by the explicit solution, thereby being suitable for real-time control of system with short sampling time.

Improved chirp scaling algorithm for parallel-track bistatic SAR

The curvature factor of the parallel-track bistatic SAR is range dependent, even without variation of the effective velocity. Accounting for this new characteristic, a parallel-track chirp scaling algorithm （CSA） is derived, by introducing the method of removal of range walk （RRW） in the time domain. Using the RRW before the CSA, this method can reduce the varying range of the curvature factor, without increasing the computation load obviously. The azimuth dependence of the azimuth-FM rate, resulting from the RRW, is compensated by the nonlinear chirp scaling factor. Therefore, the algorithm is extended into stripmap imaging. The realization of the method is presented and is verified by the simulation results.

Two Hybrid Methods Based on Rough Set Theory for Network Intrusion Detection

In this paper,we propose two intrusion detection methods which combine rough set theory and Fuzzy C-Means for network intrusion detection.The first step consists of feature selection which is based on rough set theory.The next phase is clustering by using Fuzzy C-Means.Rough set theory is an efficient tool for further reducing redundancy.Fuzzy C-Means allows the objects to belong to several clusters simultaneously,with different degrees of membership.To evaluate the performance of the introduced approaches,we apply them to the international Knowledge Discovery and Data mining intrusion detection dataset.In the experimentations,we compare the performance of two rough set theory based hybrid methods for network intrusion detection.Experimental results illustrate that our algorithms are accurate models for handling complex attack patterns in large network.And these two methods can increase the efficiency and reduce the dataset by looking for overlapping categories.

Computer Forensics Under Cloud Computing Environment

Cloud computing is becoming the developing trend in the information field.It causes many transforms in the related fields.In order to adapt such changes,computer forensics is bound to improve and integrate into the new environment.This paper stands on this point,suggests a computer forensic service framework which is based on security architecture of cloud computing and requirements needed by cloud computing environment.The framework introduces honey farm technique,and pays more attention on active forensics,which can improve case handling efficiency and reduce the cost.

A new method for PQ disturbance detection and identification

A new method based on phase-shift and N-1 Support Vector Machines(SVMs)is presented for power quality(PQ)disturbance detection and identification.Through phase-shift and simple algebra operation,the method detects out the PQ disturbances easily and effectively.Then a data dealing process is carried out to extract features from the detecting outputs.Then SVM theory is introduced into the identification of PQ disturbances.N kinds of PQ disturbances are classified with an N-1 SVMs classifier.The testing results show that the proposed method can detect and classify the PQ disturbances successfully.Moreover,the classifier has a good performance on training speed and correct ratio.