Time filter

Source Type

Xiaolei Z.,Beijing Information Technology College
Journal of Convergence Information Technology | Year: 2012

The image registration method based on mutual information is high precision and robust, so it is focused in the image registration research in recent years and is extensively applied in many fields. This paper proposes a new image registration method based on the minimal probability distance (namely symmetric KL distance) and gives a PV interpolation policy for this new algorithm to effectively the inherent weakness of the PV interpolation based on the image joint histogram. Finally the test validates correctness and high efficiency of the registration method based on the minimal probability distance and improved PV interpolation algorithm. Source

Zhan Y.,Beijing Information Technology College
Advanced Materials Research | Year: 2014

In the TV program production, the Media Assets Management system should have integration of production, material management, browsing, information management, friendly interface, standardized operation and open system and other functions. The MAM system consists of the Media Assets Management server, WEB application server, streaming media server, cataloging workstation and data streaming tape library. The MAM system in the Whole station network should be an extended full service platform to achieve unified management of various media assets automatically. © (2014) Trans Tech Publications, Switzerland. Source

Xu Z.,Beijing Information Technology College
Journal of Networks | Year: 2014

In the information of security situation, traffic is necessary data to describe the network performance. It is also an important indicator to measure attacks such as worms or DoS. During the process of situational awareness, the traffic measuring methods are often needed without the privilege of routers. Our work emphasizes on the measuring algorithms based on packet pair in end-to-end measuring technologies. The simulations have shown that the congestion status of network and the length of packet have greater effect on the measurement of path capacity. So on the basis of improved packet pair measuring algorithms, a novel traffic measuring method DTM for demand is put forward. DTM aims to acquire the optimum compromise between measuring overhead and accuracy through the utilization of detecting packets. The relation between sending rate and available bandwidth are decided by oneway delay and changes of packets. Simultaneously, DTM assigns corresponding weight to the transition bandwidth, according to the disturbed degree of packet which is set as transition point, and its adjacent packets. Then it can better reflect the influence caused by background traffic. DTM also provides the rules to judge whether two packets are transition points and it discusses the method to determine the measuring range dynamicly according to specific application bandwidth requirements. The experiments have studied the trend of relativity and the difference of packet pair, when the following factors are changing: size of detecting packet, network load, features of background traffic and bottleneck bandwidth. It is shown that DTM has characteristics in low overhead, high accuracy, good smoothness and sensitivity for network situation, compared to traditional algorithms. © 2014 ACADEMY PUBLISHER. Source

Jing Z.,Beijing Information Technology College
Journal of Networks | Year: 2014

In the process of the optimization iteration in high-dimensional space, standard particle swarm optimization algorithm falls into local optimum easily and its convergence speed is also very slow. In order to solve these problems, a self-adaptive particle swarm optimization based on the directed-weighted complex networks is proposed by introducing the complex networks. In our algorithm, we use the small-scope network with directed links to initialize the particle swarm topological structure and introduce the evolution mechanism of directed dynamic networks, which will make the particle topological structure is evolved in scale-free networks when the in-degree is complied with the power-law distribution. Therefore, our proposed algorithm will improve the learning diversity among these particles, and avoid these particles fall into local optimum easily. The simulation experiment demonstrates that the proposed algorithm will improve the premature convergence problem and enhance the algorithm convergence speed. © 2014 ACADEMY PUBLISHER. Source

Zhang H.-J.,Beijing Information Technology College
Proceedings - 2015 International Conference on Intelligent Transportation, Big Data and Smart City, ICITBS 2015 | Year: 2015

This article mainly aims at the sorting algorithm of the era of big data analysis, for the ranking method of a series of corresponding research, and for the realization of the algorithm in this series. © 2016 IEEE. Source

Discover hidden collaborations