Applied Mechanics and Materials Vols. 347-350

Abstract: A moving model of close target in a certain velocity is established aiming at the characteristic of low maneuverability. The Extended kalman filter (EKF) is used to reduce the error in location tracking. From the simulation result, it can be concluded that the moving model can describe the moving characteristic of close ballistic target perfectly. The Extended kalman filter can reduce the error in tracking location clearly.

Abstract: The existing methods of attack traceback still have many defects, such as input debugging method requires the professional intervention; ICMP traceback message takes up the network bandwidth resources and so on, so a new collaborative network is picked up, based on ant source of ideological attack tracking method. This method is mainly used strategy from ICMP traceback message, that is, data packet attacked is backed-up on the network monitor, and then the scope of the query path information is reduced using the ant colony algorithm, enabling rapid construction of the attack path. Experiments show that the method improves the query speed of tracking information and the accuracy of positioning the source attack.

Abstract: In this paper, we construct a wavelet linear estimator for the component of a finite mixture under independent identically distributed biased observations. We evaluate its performance by determining an upper bound of risk of function classes.

Abstract: A video copy detection method based on fusion of spatio-temporal features is proposed in this paper. Firstly, trajectories are built and lens boundaries are detected by SURF features analyzing, then normalized histogram is used to describe spatio-temporal behavior of trajectories, the bag of visual words is constructed by trajectories behavior clustering, word frequency vectors and SURF features with behavior labels are extracted to express spatio-temporal content of lens, finally, duplicates are detected efficiently based on grade-match. The experimental results show the performance of this method is improved greatly compared with other similar methods.

Abstract: Gas emission is basically non-stationary time series, Based on this view, wavelet multi-resolution analysis was applied to the predication of gas emission. Firstly, the gas emission data was decomposed by wavelet multi-resolution analysis. Secondly, the single branch reconstruction of each layer was predicted by establishing AR forecasting model. Synthesized all of the results from every layers, the forecasting result was obtained. The simulation showed that predication for mine gas emission with the method of this paper has much higher accuracy than AR model forecasting model.

Abstract: For the strong flurry interrupting, the body will suffer large swaying motion when it is in erecting state ,the output of its strapdown inertial navigation system (SINS) will be disturbed for the high gravitational. center of IMU, the conventional methods are difficult to achieve alignment rapidly and accurately, to solve this problem, an anti-interference self-alignment algorithm for SINS which under strong flurry is presented, which utilizes the continuous attitude update in inertial reference frame to record the attitude changes caused by sway interrupt to remove the angular interrupting, and uses the characteristics that the body exists a shake center whose speed is zero to remove the linear movement interrupting by acquiring the equivalent specific force of the shake center, and then uses the estimation of the initial attitude to determinate the attitude of the body. The simulation result show that the presented algorithm can accomplish alignment quickly even in the presence of strong flurry interference without coarse alignment phase.

Abstract: Fast and optimal motion estimation method is proposed for electronic image stabilization. First, an approach for macro-block judgment is presented. Before motion vectors calculation, gradient information is analyzed, only useful reference blocks that are indispensable for accurate motion estimation are selected, by which the number of macro-blocks for subsequent calculation is reduced. Second, in the block matching, an improved SSDA is used to reduce computing cost. Finally, the affine transformation model and similarity transformation model of image motion are created and using least squares method for solving the optimal estimation of model parameters. Experimental results show the accuracy and fast computing speed of the proposed method.

Abstract: Grid technology has become the computer and network technologys research hotspot and frontier, Grid technology development and the consummation will promote the development of educational information. The intelligent has shortcoming which are resource isolated, update difficulty etc, Based on the grid technology research and application, using OGSA (Open Grid Services Architecture) grid system structure, Taking GLOBUS as the grid middleware to build grid platform. This paper discuss the educational grid minimum implementation unit - - teaching system of campus grid and put forward the network teaching system model based on the grid technology. Then establish a distributed, heterogeneous sharing information resources of the library, it provides an efficient, fully shared virtual teaching space for network tutoring system. Finally, intelligent network teaching system and the grid binding laid certain foundation.

Abstract: In order to realize the intelligent recognition for the maturity grade of fresh corn ear, intelligent inspection system was studied based on computer vision, which could automatically complete the collection and handling of ear graphic and the recognition of maturity of the corn ear. Based on the study, a kind of intelligent recognition method was put forward under the graphic of certain frequency domain. An energy chain was established, and the characters of energy spectrum was extracted through the two-dimensional inverse discrete Fourier transformation on the graphics collected. With the above characters, a probabilistic neural network was developed, the accuracy rate of the recognition method could be 96.7%.

Abstract: Traditional image matching algorithms has poor accuracy in image comparing, such as histogram intersection method. A new image matching algorithm based on the similarity comparison of irregular shape is presented in this paper, which divides the image into a number of irregular regions according to different colors, and extracts the boundary points of the irregular region to compose an irregular shape. The direction and distance is used to comparing the two irregular shapes if the rotation of the image is not considered, otherwise circular list is used to ignore the image rotation. It can be used widely. If two irregular shapes are similar, the two images are considered similar. Experiment proves that this method can effectively improve the image matching accuracy.