Applied Mechanics and Materials Vols. 397-400

Abstract: Trademark represents the quality of the goods and the manufacturer's reputation, which is an important symbol of goods; automatic trademark recognition has a wide range of application prospects in many fields such as e-commerce, advertising and logistics transportation. In target recognition algorithm, the most critical is to extract the target feature. Because of the moment feature has better invariant characteristics; it is widely used in image feature extraction. In this paper we use Hu modified moments and the minimum distance classifier for trademark recognition, the experimental results show that the modified moments have higher recognition rate than the original Hu moments, this method can improve the recognition rate of the target image, it is conducive to trademark recognition.

Abstract: Visualizing high dynamic range image or video on traditional display devices need fast tone mapping technique. Traditional bin-based histogram equalization is fast enough but cannot handle high dynamic range data, e.g. single or double precision. Other common tone mapping methods require a high computational complexity. In this paper, we propose a new continuous data equalization method based on simple sorting, which can quickly and efficiently enhance the visualization of high dynamic range data. This method can be applied to real-time visualization of high dynamic range video with cost effective hardware acceleration.

Abstract: Modern printing industry detective techniques, which have transferred from traditional single discrete manual operation into line detection based on machine vision. The line detection research under the highly automated process requirements, which is based on color graphics, gravure text, high-speed stamping and laser hologram hot stamping technology, is the multi-feature defect detection and recognition. By CDD study linear measurement techniques and computer vision technology, the research built the visual inspection platform, and design detection scheme combined with the specific requirements of the printing process in database technology and image processing technology. Research obtained the online testing solutions and system information flow characteristics, which achieved requirement of multi-feature detection, high precision and high speed continuous production technologies.

Abstract: The research takes the campus network users logging on the Internet as the analysis object, using the data preprocessing technology to clean up the original data, combined with statistical analysis and data mining technology to analyze the users access log records, which will result in the form of dynamic charts for Web display, by using Microsoft SQL Server 2008 and Microsoft Visual Studio 2010. Take use of intelligent .NET platform, combined with K-means algorithm to cluster the students information. DMX (Data Mining Extensions) will be used to show the mining results on the Web. The realization of the system can not only carry on correct guide to Internet users and regulate the behavior of students, but also have important guiding meaning to managers and policy makers for analysis and making decision.

Abstract: Focus on the estimation for DOA and polarization parameters of electromagnetic sensor array, this paper proposes the method that uses cyclic relation function to substitute for covariance matrix. Because cyclic statistic is less sensitive to stable noise and any cyclic stable noise with different cyclic frequency, the method proposed is immune to any stable color noise. This method use minimum norm method to solve the spectrum function. So this method can restrict the effect of the cyclic relation matrix estimating error. Computer simulation experiments prove the performance of this method.

Abstract: In order to realize highly intelligent and automatic species identification and recognition, we obtained the images of 11 varieties and each variety includes 50 seeds. For each image, we acquired 33 characteristics including shape, color and texture characteristics. And then we constructed the Artificial Neural Network and Support Vector Machine model to train and identify different varieties. We built the recognition system based on Visual C++ 6.0 and OpenCV library.Results shows that the SVM method has higher recognition effect than neural network overall and the recognition effect is more stability, the overall self-_recognition performance can reach 100% and test accuracy can reach 85%. The recoginition System base on Visual C++ runs faster than that of Matlab, which is more suitable for real-time varieties identification.

Abstract: This paper presents a medical bottle detecting system based on image processing, and completing the tightness detection task for the installation of vial and cap. Firstly, the image of the bottle is captured by using the background subtraction method; secondly, getting the ideal threshold T by the iterative formulae and carrying on linearization processing for image; then, the ideal target image of medical bottle can be obtained by using morphologic operations. Finally, using the 8-directional chain code to calculate the circumference, area and height of bottle. The image processing course and experimental results are discussed at the end of the thesis. The experimental result shows that this algorithm is effective and accurate.

Abstract: According to the distribution and diversity of logging information, well-logging information data manipulation platform is established by Agent software technology, in a highly flexible, robust and unified information management environment, provide support for comprehensive well-logging information distributed storage and reuse. Examples of well-logging information reuse, verified the feasibility of application mode.

Abstract: This paper has innovatively put forward a requirements management integration platform which takes the requirements review as the core. The integration platform proposes a model of gaining discrete demand, realizes integration and the sharing of resources with other requirements tools, and also gives the management and selection methods of the requirements analysis review staff. The paper finally achieves a collaborative requirements review wizard system which can improve the standardization and efficiency of the requirements review.

Abstract: This paper firstly analyzes the reliability Influence factors of the embedded software, then puts forward an embedded software reliability evaluation model based on BP neural network, finally the preliminary application and verification of the model is given.