Stereo matching algorithms based on  BP(Belief Propagation) can yield better matching accuracy over the local methods, but large quantity of calculating is needed. For this problem, this paper presents a stereo video matching algorithm based on motion estimation. The traditional BP algorithm is used to get the disparity map of I frame, and the propagating messages with optimal disparity of each pixel are stored. Then, for P frame, the rearranged propagating messages through referring to motion estimation information from I frame are used as the initial values for iteration of BP algorithm. This reduced the number of iterative times. Experiment results show that the proposed stereo video matching algorithm of belief propagation based on motion estimation can dramatically enhance the efficiency of stereo video matching.

Fast routing technology of space access networks is studied in case of dynamic network topology, big transmission delay and a topologybased routing method of rapid construction techniques is According to the analysis about predictable movement characteristics of space access point, space access network is organized by means of WMN（Wireless Mesh Networks) to transmit data. Through WMN structural analysis, we designed a mobile backbone topology construction algorithm for space access network, which spread flooding information only among backbone nodes and reduced the amount of control and routing packets. So this algorithm is able to quickly construct routing information and improve the efficiency of routing establishment.

Smart antenna and power control are combined to reduce interference and error rate and to improve the performance of the system.The signal is beamformed through smart antenna, utilizing game theory to realize power control of the transmiter.Considering the influence of the power level strategy from other users, power iteration is used to improve the transmission power. Compared to only using power control, it can improve the SINR(Signal to Interference Ratio), throughput, sending power etc.and the simulation results show its advantagement.

Tracking around the zenith requires a very rapid motion in azimuth, approaching infinity in the limit of a star passing exactly through the zenith.  The result is a  blind zone around the zenith for altitudeazimuth mounts. The zenith blind zone is mainly considered and some formulae are given for estimation of the dimensions of it. The effect of field rotation, which is another disadvantage for altitudeazimuth optoelectronic tracking system, is analyzed briefly. Two solutions to overcome the zenith blind zone are presented, which also leads to some directions for future investigation. 

To achieve diversity gain and reduce algorithm complexity, cooperative diversity is employed to cluster MB-OFDM (Multiband orthogonal Frequency Division Multiplexing) UWB systems, in which cluster is the minimum granularity of resource allocation. Improved TFC(Time-Frequency Coded) model is proposed. We propose a cluster allocation and planar recursion power allocation algorithm based on convex optimal theory for the purpose of maximum system capacity under the overall power limitation. To reduce the complexity of algorithm, a quick crosslayer allocation algorithm judged by the relative requirement of cluster is further introduced. Simulation results show that this algorithm can satisfy usersQoS requirement,achieve a good tradeoff between system throughput and the fairness. 

To addreress various difficulties faced in solving network congestions by hardware methods such as capacity expansion, addition of microcells, adjustment of base station shapes, adjustment of antennas, and so on, the influence of GSM (Global System for Mobile Communications) wireless network parameters on congestions sanalyyed, and the network optimization methods such as network parameter adjustment ane emmployed to ease wireless channel congestion. The computer based simulation result shows that proper adjustment of network parameters will lower network congestion, balance traffic, and ease comparatively long, not so serious traffic congestion.

To improve reliability of the electronic devices in space and efficiency of the radiationhardened SOI（Silicon on Insulator） ICs design, a library developing flow is established. The radiation-hardened standard cells library for Synopsys EDA（Electronic Design Automation) tools is designed based on 3.3 V-0.35 μmPD（Partly） SOI CMOS(Complementary Metal-Semiconductor) technology platform. The standard cells are composed of body contacted devices featuring H-gate， asymmetric source and drain to improve the radiation hardening performance. The cells library is validated by EDA tools and measurement. Experiments show that the EDAC(Error Detection And Correction) circuit functions properly with total dose of 300  krad (Si).

In order to meet space camera system application, a highspeed CCD(Charge Coupled Device) image acquisition system based on Camera Link is designed. CPLD（Complex Programmable Logic Device） is used for timing and logic control; CCD processor TDA8783 is used for highspeed A/D converting; pingpong structure memories are used for high speed transmission and meeting Camera Link protocol.TDI（Time Delay Integral）mode is used to meet the interline CCDs bad quality and lower SNR（Ratio Signal to Noise） result for low fill factor and quantum efficiency,through setting the TDI stage and integral time of TDI.The results show that the CCD driver system can output two groups of CCD signals simultaneously in data output speed of 15 frame/s,SNR exceed 50 dB, channel asymmetric less than 2％.The driver system can meet the neqiciremeut of the camera system because of its good function and low cost.

With the development of network technology, computer crimes are increasing.As an interdiscipline of law and computer science,computer forensic becomes an essential part in society.We outline the emergence and development of forensic science, introduce the basic forensic principal and analyze several typical forensic models or tools，to understand the forensic precisely.

A new parameter variation chaotic synchronization method based on parameter variation chaotic synchronization conditions is persented.Only one variable of driving system is used in the method, feedback control and adaptive control methods are also used to guarantee the chaotic synchronization.The method need not parameter identification and use only part of variable of driving system which loosen the restriction conditions of chaotic synchronization. The Lorenzs group chaotic system is considered,theorem on the sufficient conditions for attaining chaotic synchronization has been proposed. The simulations have shown the effectiveness and feasibility of the method.

The level of experts is a decisive factor for the success and failure of the review of science research projects.An approach to evaluating the level of experts for the assessment of science research projects is proposed. The evaluation indexes related to the situations on the science research projects reviewed by experts and obtained by experts are described, respectively, and computation formulas of the evaluation indexes are presented. Then, alternative experts are screened by the constructed bidimension coordinate system. An approach based the TOPSIS(Technique for Order Performance by Similarity to Ideal Solution) method is proposed to obtain the rank of the rest experts. And an illustrative example is used to show the practicality and validity of the proposed approach. The proposed approach can provide decision support to expert selection for the assessment of science research projects.

In order to reduce the rate of premature convergence and to escape from local optimum, GEP(Gene Expression Programming) based on diversified development strategy is proposed,  which assigns the population with different development strategies to enhance the optimizing ability of GEP through GSBS（Gene Space Balance Strategy), ACMO（Adaptive Crossover and Mutation Operators) and obsolete operator (OBSO). Experiments on function mining show that all of strategies play roles of mining. Compared with the result of GEP. The number of average evolution generations is decreased by 11%, evolution time is decreased by 8%, and the success rate is increased by 20%.

To overcome the deficiency of Laplacian score algorithm which makes feature selection mostly depending on the local geometrical structure of samples, an improved semi-supervised feature selection algorithm was proposed, based on the constraint Laplacian score. It utilized the cannot-link pairwise constraints among samples as the global structure. Then the selected features were those can preserve the local structure from the nearest neighbor graph,and preserve the global structure from the cannot-link constraints. Experiments on Yale and PIE(Fave pose,Illamination,Expression dadbase) datasets show that the performance of proposed algorithm outperformed Laplacian score algorithm significantly, and was equivalent to the supervised Fisher score algorithm and the latest semi-supervised constraint score algorithms. And it is even better than constraint score algorithm in terms of stability.

The real-time image to enlarge engine IP( Intellectual Property ) core is designed, to minimize the hardware resources of their occupation and the shortest clock cycle consumption, and to achieve the ratio of the image to enlarge. The IP core using VHDL（Very-High-Speed Integrated Circuit Hardware Description Language） language description can be integrated into a variety of display output integrated circuit chip. A simplified format D1 to XGA(Extended Graphics Array) effective method is proposed. An algorithm based on FPGA（Field Programmable Gate Array）, and the form and pipeline processing structure are also given. Experiments show that the design of image magnification system can display per second rate 65 high-quality D1 to XGA format, the same time keep a good image edge information.

In order to solve problems that the data of dynamic computing systems LAMMPS are too huge and difficult to control,the paper developed an intelligent and efficient clusters computing system architecture firstly, which based oncloud computingconcept.The method combine FTP(File Transfer Protocol) storage technology, UDP fast network transmission with LAMMPS(Largescale Atomic/Molecular Massively Parallel Simulator) are too huge and difficult to control. The method combines FTP storage technology, UDP(User Datagram Protocol) fast network transmission with LAMMPS, reduces manual intervention, improves the processing efficiency of the molecular model and proposes a new clustering solution for molecularlevel physics, chemistry computer simulation.

We studied watermark scheme used in content features complete authentication to velidate if the image was juggled in processing. Based on the application requirements analysis, we have built general protocol to be followed during watermark embedding. A fragile watermarking scheme was implemented followed by watermarking protocol and fragile watermarking algorithm. As hash elements, participating product singular value of image, the authentication codes have been generated in the scheme. We compressed low bit flat of image by MHC（Midified Huffman Codeing） coding. According to the experimental results, it can completely meet the application requirements.

In order to better monitor and verify the network traffic and to deal with problems in the existing network sniffers, we designed a new simple and practical network sniffer based on the detailed analysis of the fundamental principles of software network sniffers. And it is implemented by programming in the Windows XP by making use of Visual C++ 60. Finally, a network sniffer designed was analyzed simply compared with the previous ones. Results of the analysis showed that the new network sniffer designed is fully equipped with data capture, analysis of the data packets and other basic functions. And it is simple and practical, so it has better application value.