Applied Mechanics and Materials Vol. 415

Abstract: This paper presents CU-Track, a multi-camera framework for real-time multi-object tracking. The developed framework includes a processing unit, the target object, and the multi-object tracking algorithm. A PC-cluster has been developed as the processing unit of the framework to process data in real-time. To setup the PC-cluster, two PCs are connected by using PCI interface cards that memory can be shared between the two PCs to ensure high speed data transfer and low latency. A novel mechanism for PC-to-PC communication is proposed. It is realized by a dedicated software processing module called the Cluster Module. Six processing modules have been implemented to realize system operations such as camera calibration, camera synchronization and 3D reconstruction of each target. Multiple spherical objects with the same size are used as the targets to be tracked. Two configurations of them, active and passive, can be used for tracking by the system. The algorithm based on Kalman filter and nearest neighbor searching is developed for multi-object tracking. Two applications have been implemented on the system, which confirm the validity and effectiveness of the developed framework.

Abstract: For detecting the defects of the vertical polarized electrolytic capacitors in the process of production, an online machine vision detection method based on DSP is proposed in this paper. The hardware framework is designed emphatically with DM642 as the core processor, which has the advantages of small size, fast processing speed, low cost and low power consumption. The software workflow of the capacitors defects detection is introduced. To meet the requirement of high production and real-time detection of capacitors, a rapid appearance detection algorithm is introduced by using a few number of feature points instead of the whole. The experimental results show that the proposed method can achieve many types of defects detection of the capacitors.

Abstract: This paper proposes a new image Multi-Instance (MI) bag generating method, which models an image with a Gaussian Mixed Model (GMM). The generated GMM is treated as an MI bag, of which the color and locally stable invariant components (SIFT) are the instances. Agglomerative Information Bottleneck clustering is employed to transform the MIL problem into single-instance learning problem so that single-instance classifiers can be used for classification. Finally, ensemble learningis involved to further enhance classifiers generalization ability. Experimental results demonstrate that the performance of the proposed framework for image recognition is superior to some common MI algorithms on average in a 5-category scene recognition task Key words:Multi-Instance Learning; Gaussian Mixed Model; AIB Clustering; image modeling; Single-Instance Bag; Ensemble Classifier; Scene Recognition

Abstract: In this paper, a hybrid differential evolution harmony search (HDEHS) algorithm was presented for solving power economic dispatch problems. In this algorithm, mutation and crossover operation instead of harmony memory consideration and pitch adjustment operation, this improved the algorithm convergence rate. Moreover, dynamically adjust the key parameter (e.g. mutagenic factor F, crossover rate CR) to balance the local and global search. Based on a 13 units power system experiment simulations, the HDEHS has demonstrated stronger convergence and stability than original harmony search (HS) algorithm and its three improved algorithms (IHS, GHS and NGHS) that reported in recent literature.

Abstract: Differential evolution (DE) is a population-based stochastic function minimizer (or maximizer), whose simple yet powerful and straightforward features make it very attractive for numerical optimization. However, DE is easy to trapped into local optima. In this paper, an improved differential evolution algorithm (IDE) proposed to speed the convergence rate of DE and enhance the global search of DE. The IDE employed a new mutation operation and modified crossover operation. The former can rapidly enhance the convergence of the MDE, and the latter can prevent the MDE from being trapped into the local optimum effectively. Besides, we dynamic adjust the scaling factor (F) and the crossover rate (CR), which is aimed at further improving algorithm performance. Based on several benchmark experiment simulations, the IDE has demonstrated stronger convergence and stability than original differential (DE) algorithm and other algorithms (PSO and JADE) that reported in recent literature.

Abstract: Harmony search (HS) algorithm is a good meta-heuristic intelligent optimization method and it does depend on imitating the music improvisation process to generate a perfect state of harmony. However, intelligent optimization methods is easily trapped into local optimal, HS is no exception. In order to improve the performance of HS, a new variant of harmony search algorithm is proposed in this paper. The variant introduce a new crossover operation into HS, and design a strategy to adjust parameter pitch adjusting rate (PAR) and bandwidth (BW). Several standard benchmarks carried out to be tested. The numerical results demonstrated that the superiority of the proposed method to the HS and recently developed variants (IHS, and GHS).

Abstract: With the gradual development of industrial technology, visual measuring technique has been widely used in industrial field. In the automatic production, it is used in the analysis of objects characteristics, the detection of working condition and the control of quality. This paper mainly studies how the machine vision theory is applied in the practical detections, and develops the overall visual detection platform. It uses the serial port communication between VC++6.0 and Single-Chip Microcomputer to control the experimental platform. In order to realize to test the production defects, it uses the knowledge of mathematical morphology to put forward a defect detection method of production image based on regional morphology. It can realize to simulate actual production line, and has a certain development prospects for real-time detections of product quality.

Abstract: Binocular stereo matching is a hot and difficult problem in machine vision. In this paper, based on the matching method of Halcon which is visual software perform image matching. First, performing binocular stereo vision system calibration, based on the calibration results acquired the epipolar standard geometric structure. Then, image matching researched under this structure. At last, using ncc matching algorithm, through comparing the different parameters matching window obtain ideal match results. Experiments prove that this method not only can effectively shorten matching time, but also can achieve higher matching accuracy.

Abstract: With the digital age with the rapid development of information technology, marketing, networking, multimedia devices into people's lives, colleges and universities have also appeared in the popularity of multimedia equipment and technology and universal application. Professional teaching dance for universities, from the multimedia technology in university teaching dance development status of the actual situation, respectively, from the college dance theoretical courses, practical courses dance, dance integrated curriculum, explains the meaning and role of multimedia technology in university dance the importance of teaching.

Abstract: As the informationization of the medicare sector in China grows up and the sources of the rehabilitation data are distributed and multi-leveled from different centers, it is urgently needed to integrate the resources of various application systems, implement unified data exchange of the distributed and heterogeneous multiple data sources and encapsulate the data resources into various kinds of services. For this purpose, this paper puts forward the idea of SOA-based cloud services encapsulation and integrated architecture. At first, the multi-source and heterogeneity of data and the integration of information in the rehabilitation management are analyzed. Then, the cloud services encapsulation and integration technology is study to find out how to encapsulate data resources into various kinds of services, build a SOA-based cloud services platform, find a solution for the integration of multi-sourced and heterogeneous data, and meet the function extension, encapsulation and release of cloud services arising out of changing demands. At last, A case of the cloud platform system is presented to show the feasibility and effectiveness of the architecture platform.