The problem of ghost objects seriously affects the intelligent video surveillance application in background subtraction method for moving objects detection in videos. According to the characteristic that the ghost’s foreground and the true object’s foreground are different in the similarities of edge, and by defining edge similarity measure functions, this paper presents a ghost discriminant algorithm based on edge similarity. Experimental results show that compared with traditional methods, this method is effective and robust.
In order to study the interaction between user behavior and Peer to Peer(P2P) file sharing systems, this paper proposes a mathematical model. The contributions include: considering the guide strategy factor for the first time in P2P model based on user behaviors; analyzing the impacts of guide strategy on user behaviors and system performance; comparing the effects of three guide strategies. Simulation results show that guide strategy can effectively improve the probability of first downloading good file, and reduce the number of retrial and overall traffic.
Based on the research of SMBIOS entry point structure, SMBIOS structure tables and memory-mapped I/O, this paper puts forward a method of universal Basic Input/Output System(BIOS) image file sampling and formatting, and establishes a BIOS image file security detection model. Experiments prove this method can analyze multiple IBV’s(Independent BIOS Vendor) BIOS image files successfully, thus it solves a critical problem in BIOS security risk analysis and detection system.
According to the existing problems of dominate cells information missing, an automatic data mining method of contingence table is purposed. According to the location of dominate cells, the categories in the original table are classified and represented by tree structure models on the basis of relevant theories of Multinomial Processing Tree(MPT) models, the following processes including hypothesis generations, parameter estimation and goodness-of-fit test is performed automatically. Application result shows the approach can effectively extract the latent interactions and peculiarity association rules.
In order to improve water quality evaluation of multi-spectral image accurately, this paper puts forward a model for water quality evaluation of Radial Basis Function(RBF) neural network based on optimized parameters of Genetic Algorithm(GA). The method uses SPOT-5 data and the water quality field data. It chooses four representative water quality parameters. RBF neural network is trained and tested, and the parameters of RBF neural network are optimized by particle swarm optimization algorithms. Water quality parameters of COD, NH3-N, DO, CODmn are retrieved by the trained RBF neural network. Experimental result shows that the method has more accuracy than the routine method.
It exists problems of big size, high power and low efficiency when cloud computing center is composed of traditional sever. This paper designs and implements a new cloud computing server. It discusses the key technologies, such as cloud processors, 3D-torus net, hardware virtualization, etc. Compared with traditional servers, the new one has the characteristics of small size and low power.
Efficient resource provision and schedule is the key to build a private cloud environment. This paper presents a resource provision and schedule approach based on the proposed cloud application meta-model for cloud applications in a private cloud environment. It creates a minimum spanning tree according to the characteristics among virtual machines, and splits the set of virtual machines and maps them to the physical hosts by using the graph theory based on split clustering algorithm. The experiments and comparative analysis of the method for virtual machines show that the approach is efficient in enhancing the overall performance of cloud applications while reducing the consumption of physical network equipments and bandwidth resources.