Advanced Materials Research Vols. 1049-1050

Abstract: For some special blade section, an accurate partitioning algorithm for feature point cloud is presented to avoid the drawbacks of traditional convex hull algorithm. On this basis, the method based on graph transformation is used to extract the chord length. The experimental results show that these two algorithms have strong generality, high accuracy and efficiency, can meet the testing requirements of blades.

Abstract: In view of the problems that high complexity, large calculation and the difficulty to apply to real-time systems in the current moving target tracking algorithm, this paper introduce the BRISK feature extraction algorithm, and proposed the object tracking algorithm based on BRISK. Set up the background model and use the background difference method to detect the moving target template. Then match in the next frame and track the target. In order to reduce the search feature matching area, further improve the real-time of the algorithm, we also introduce the kalman filter algorithm to estimate the target motion trajectory. The experimental result show that comparing with the SURF, SIFT feature tracking algorithm, the algorithm of this paper has greatly improved in real-time.

Abstract: English, as one of the most important information carriers, has become the most widely used language worldwide, and its importance is increasingly prominent in each field of our society. Examinations are an important means of evaluating the level of English language learning and teaching outcomes. For the problems existed in traditional English exam, this paper research English online examination system intelligent test paper based on random algorithm. First, the design test paper constraints, including the question amount constraint, questions type constraints, knowledge point constraints, difficulty constraints, answer time constraints and so on; Then, using the linear congruence generator study of a random number generation, list recursive formula and describe parameters; Finally, using flow diagram to study random test paper process, for each step describing in detail. Advantages of random algorithm are easy to use, the disadvantage is not retrospective, applies to the number of exam papers and extraction requirements are relatively few cases.

Abstract: Traditional anomaly detection algorithm has improved to some degree the mechanism of negative selection. However, there still remain many problems such as the randomness of detector generation, incompleteness of self-set and the generalization ability of detectors, which would cause a lot of loopholes in non-self-space. This paper proposes a heuristic algorithm based on the second distribution of real value detectors for the remains of loopholes of the non-self-space in the first distribution. The algorithm proposed can distribute real value detectors through omission data based on the methods of partition and movement. A method is then proposed to solve the problem on how to get the optimal solution to the parameters related in the algorithm. Theoretical analysis and experimental results prove the universality and effectiveness of the method. It is found that the algorithm can effectively avoid the generation of loopholes and thus reduce the omission rate of detector sets.

Abstract: Subjective question marking system at present is affected by the attention of people, the subjective topic grading principles are common contrast degree of exam questions similar to those of the reference answer, and based on the improved semantic similarity algorithm, calculation of sentence similarity, the similarity degree of exam questions and reference answer is obtained, thus give scores.And design based on semantic similarity experiment, the experiment results show that the proposed multi-level fusion similarity calculation method to improve the original method, on the basis of integration advantages of various methods, make the calculation results meet the requirements of the scoring system.

Abstract: The plasma sheath can be regarded as a model of a phenomenon of spacecraft reentry or hypersonic vehicles and causes so called radio blackout, which interferes with radio signals. This paper gives a survey on several aspects of plasma sheath, including the characteristic of plasma sheath, experiments to deal with the blackout and numerical solutions. The numerical methods includes WKB, finite-difference time-domain method, particle-in-cell method, CFD based method and Monte Carlo method. Some discussions are also presented.

Abstract: In such a developed day of information communication, communication is an important essential way of interpersonal communication. As a carrier of information, expression is rich in human behavior information. Facial expression recognition is a combination of many fields, but also a new topic in the field of pattern recognition. This paper mainly studied the facial feature extraction based on MATLAB, by MATLAB software, extracting the expression features through a large number of facial expressions, which can be divided into different facial expressions more accurate classification .

Abstract: The theory of down sampling technique base on band-pass sampling and using inSoftware radio is introduced in this paper. We provided an achievement method ofNumerically controlled oscillator (NCO) in digital down converter of down samplingtechnique in the article.

Abstract: An efficient ear recognition method by weighted wavelet transformation and Bi-Directional principal component analysis was proposed. First, each ear image was decomposed into four sub-images by wavelet transformation ,the four sub-images were low frequency image , vertical detail image ,horizontal detail image and high frequency image .Then the low frequency image was decomposed into four sub-images, the four-images were weighted by different coefficients, then ,the four sub-images were reconstructed into a image .On this basis ,the feature was extraction by the BDPCA method ,and then we use the k-Nearest Neighbor Classification to recognition .Experimental results show that the method have high recognition rate and shorted training time.

Abstract: Individuals’ privacy protection when publishing data for research has recently put great attention on data mining and information resources sharing fields. Privacy preservation is an important and challenging problem in micro-data publishing. This paper aimed to find an available directly way protect patient privacy. Processing numeric values which got from body sensor network (BSN). Firstly, we analyze the characteristics of medical data which collected from BSN, and then the records will be grouped according to the Quasi-identifier. The last step is to inspect the diversity of sensitive attributes.