Stereoscopic images created from video images and their associated depth data partly contain artifacts due to misalignment between depth and video data. To solve this problem, we present an algorithm of registration of depth and video data based on edge detection. First we use edge detection method to get the edge of the video image and the depth image respectively. Then we set the edge of the video image as the baseline to register the edge of the depth image. Experimental results show that by using the proposed algorithm the spatial alignment of depth and video data can be improved, leading to an alleviation of the observed artifacts in stereoscopic images.

In order to design a current mode filter which has the property of current controllability, the linear transformation filters by employing current mirrors as the elementary active elements are proposed. Based on the linear transformation, a simple and systematic design table is given to realize currentmode filters using current mirror. Using the proposed design table, all pole and elliptic filters can be designed easily. As an example, a new thirdorder elliptic lowpass filter is realized using only current mirrors and grounded capacitors. PSPICE simulations have been carried out using 0.18 μm CMOS technology. The simulation results that comply with the theoretical analysis are obtained. The circuit has several advantages, such as low voltage（0.9 V）, simple circuit, low filter sensitivities, adjustable cutoff frequency, which is suitable for full integration.

Because the current NTP(Network Time Protocol) clock synchronization algorithm can not meet the new networks demands of high synchronization precision, an improved algorithm of NTP based on DS（Dempster/Shafer） theory is proposed. Firstly the current shortage of NTP clock synchronization is analyzed; then based on DS evidence theory, an improved NTP clock synchronization algorithm is established; at last a simulation test is done, and results show that the improved algorithm can enhance the synchronization precision.

In order to promote the digital, intelligent and humanized development of modern teaching process,combining virtual technology with embedded wireless technology and using full\|LCD(Liquid Crystal Display) touch screen as the control terminal to complete all system control order.Through the 24 G wireless transmission, the instructions and data on the touch screen can be sent or received to realize the interactive communication. PC software is written by LabVIEW and combines the teacher mode with interactive mode perfectly. The experiment has shown that six groups of student terminals and one host terminal can connect with each other smoothly and simultaneously, the touch panel shows quick processing speeds, the wireless modules and PC display speeds are satisfactory as well.

To reduce distortion due to AWGN（Additive White Gaussian Noise） in transmission channel of DCSK(Differential Chaos Shift Keying) chaotic communication system, an improved system scheme was proposed Which is called PIDCSK（Phase Inversion Differential Chaos Shift Keying） chaotic communication system based on DCSK modulation system. According to the delay retransmission mechanism of information signal and reference signal had a
halfsymbol period delay at the transmitter in the DCSK system; then at the PIDCSK receiver the phase inversion correlated demodulation during halfsymbol period for received signal was exploited. By the fact that the adjacent noises within a symbol period in the channel was correlated, the proposed scheme could reduce the adjacent noise interference obviously and further decrease the bit error rate of DCSK modulation system with the same SNR（Signal to Noise Ratio）. Simulation results show that PIDCSK system performance is better than DCSK system, especially in larger correlation coefficient, and the lower SNR condition.

The V-PREPT（Versatile Preemption) algorithm only concerns with minimizing the total cost of preemption leaving out the number of LSPs（Label Switching Paths) to be preempted, leading to excessive number of rerouting decisions. In this paper, an algorithm providing utilization rate of network resources and QoS（Quality of Service), named N-PREPT is proposed. The N-PREPT algorithm executes the preemption of LSPs according to the main preemption optimization criteria, under the premise of minimizing the total cost of preemption. It considers minimizing number of LSPs to be preempted and avoiding the occurrence of cascading preemption at first and uses hierarchical approximation method to raise restriction to the number of preemption. Simulation results show that the proposed algorithm significantly outperforms the V-PREPT algorithm when preemption occurs in the network.

Cycle distribution of LDPC(Low-Density Parity-Check) codes affects the codes ^,decoding performance and encoding complexity, however it is commonly NP hard to analyse.We propose the rotationdistance for analysis of QCLDPC（Quasi-Cyclic Low-Density Parity-Check） code based on circulant matrices. The circulant submatrices within the paritycheck matrix are treated as a “matrix node” to simplify theTanner graphs of the codes. Thus cycles of QC-LDPC codes can be found efficiently, and we demonstrate the usefulness of the new method by a simple proof of the known result that 12 is an upper limit of the girth of the QC-LDPC codes we considered. Moreover, the cycle analysis based on the new method also reveals relations between decoding performance and the cycle distribution of the code.

Multi-radar data fusion was a typical system of multisensor information fusion and track correlation was one of key technology.Integralcorrelation technology was proposed. Its foundation of application was fuzzy set theory, and the error correlation from sensor^,s  measure error and uncertainty was solived. The processing rules was given for plotplot correlation or plot-track correlation. The simulation results showed that this method was an effective datacorrelative method. 

In order to improve the management system and operational efficiency of IP QoS（Quality of Service） strategy. SLA(Service Level Agreem) parameters are designed based on the SLS(Service Level Specification); describes the strategies; design strategy is stored in the directory server and policy management tool (PMT: Policy Management Tool). Policy Management tools policy setting process is completed the user's network quality of service policy setting function, policy management tools to achieve policy query program on the strategies of network or user to query the warehouse storage strategy functions.Technical development of network quality of service has certain practical significance.

In order to improve the security level of wireless sensor network, by analyzing the typical key management scheme and combing with specific applications, a CKMS (Cluster\|based Key Management Scheme) for wireless sensor networks is presented. It includes five types of keys and key update mechanism,and supporting network expansion. Comparing with the existing key management schemes, this scheme has better connectivity, less communication overhead and computational overhead, using less storage.

In order to disytiguish different nellrons,the morphological classification of neurons in the understanding of the characteristics of neuron structure and function has great significance. Using LMeasure software to extract geometry of neurons for feature, characteristics extracted through principal component analysis to reduce the dimension. Using probabilistic neural network, BP（Back Propagation） neural network, fuzzy classifier consisting of “Note” on the pyramidal neurons, purkinje neuron, motor neurons, sensory neurons, Bipolar interneuron, tripolar interneuron and multipolar interneuron to classify 7 kinds of neuron. Experiments show that the classification of three combined is better than any of them has good performance, a higher recognition rate.

Evolution of circuit is a focus of EHW（Evolvable Hardware）. To improve the evolution efficiency and success rate of circuits, the multiobjectives during evolution of circuit are defined and quantized. To solve the multiobjective optimization, an evolutionary algorithm of circuits based on MGEP（MultiObjective Gene Expression Programming） is presented. The chromosome encoding of GEP(Gene Expression Programming) in evolution of circuit is designed; the multiobjectives during evolution of circuit are defined and quantized. And the uniformly scattered search direction is enhanced by nondominated sorting and fitness sharing strategy. The experiments on evolution of digital circuits show that MGEP improves the evolutionary efficiency. Compared with GP, the evolutionary time of MGEP drops 72.9%. MGEP is also capable of searching out simple and practical circuit. Compared with GP and GEP, the ratio of searching optimal circuit increases 50.4% and 38.9%.

In the optically pumped magnetometer, the signal to noise ratio is very low when use the method of detecting Zeeman energy level transition directly. However, the detected signal to noise ratio can be greatly improved by using the technique of magnetic resonances optical detection which can transform the microwave quantum detection to the photon detection, because the energy of photon is much greater than that of microwave quantum.The method of magnetic resonances optical detection in the OPM(Optically Pumped Magnetometer) has been studied, the principles and experimental devices are discussed which includes the method of detection of cells transparency, detection of orthogonal beams, spin exchange collision and the method of pump beam modulation. The characteristics of these methods are analyzed and compared, the application and complexity are considered, and all these will provide the basis for the design of the optically pumped magnetometer.

In order to maintain data consistency under a distributed environment,with a particular disaster recovery system,the data consistency for distributed data base were analyzed. Based on current data consistency maintenance technology research and comparison, proposed a technique based on a message queue data consistent method of data synchronization through the study of the actual disaster recovery system to ensure that under certain conditions,the data consistency.

Because the complexity of ChanVese model is expensive and its segmentation time is long, an improved ChanVese active contour model is presented. In the model, a level set regularization term is introduced into ChanVese model, which can force level set function close to a signed distance function so that reinitialization can be avoided，and the complexity of ChanVese model is less expensive，and we also change the function which is set as the initial level set function, so that new contours emerge more easily. Finally the experimental results demonstrate that the improved ChanVese model is robust to noise, converges more quickly, and can achieve global segmentation.

To change impossibility of live teaching affairs meeting at different places,the JXTA(Juxtapose) is applied in distance teaching affairs meeting for the first time.Study to routing mechanism of JXTA platform shows that JXTA is able to meet the needs of distance teaching affairs meeting but the routing is unstable and unreliable. So, a scheme introduced consistency hash algorithm Chord ring to cover JXTA routing is proposed to solve the problem. The scheme also solves technical and routing problems in distance teaching affairs meeting, providing stability and reliability of the network in reality and makes distance education system  more perfect.

In order to detect radioactive materials, the pulse detection of radioactive materials and energy calibration, are designed based on the photomultiplier. The system includes photomultiplier detector, the radioactive pulse conditioning circuit and the circuit of MCU MSP430F1612, which has 12bit A/D inside, to sample the peak of radioactive pulse and so on. The data through the Single chip microcomputer dealing will be transmitted to PC through the RS232 serial port. The PC uses the Visual Basic 6.0 software to filter and fit the data in order to search the peak. Analyzed the building materials  have which radioactive element contained and it is energy range by the obtained energy.This system is much more accurate to achieve the energy scale,automatic and manual peak search,which could accurately determine the energy between 0.01 to 2.97 MeV.The system has realized the detection of radioactive materials on the scene.

In order to overcome the performance deficiencies of the hydraulic system in old drilling machine, and satisfy the need of high power drilling and the complex drilling control,we design and build a testing drill system, which has the ability of monitoring, recording and controlling the drilling parameters in realtime. We adopt the ameliorat PID（Proportional Integral and Differential） algorithm and fuzzyPID control algorithm, to improve the stability and accuracy to achieve good control effect of the system. Within the set range, the average error is about 10%, achieving good results.