Zhen H.,Shandong College of Electronic Technology |
Yewei L.,Shandong Normal University |
Jinjiang L.,China Institute of Technology
Journal of Convergence Information Technology | Year: 2011
Visual attention theory in cognitive psychology is referenced, and visual attention model is used to accurately detect the salient region in the images, so that image salient region extraction algorithm based on improved visual attention model is proposed. The traditional visual attention model is firstly improved, and the improved visual attention model is used to extract the visual feature of the input image. The center - surrounding operator is used to obtain the salient degree of the concern point, and the salient degree value is used to constitute many feature maps, so as to the salient region map is obtained by the normalized operation. At last, BP neural network algorithm is used to make a distinction between the salient region and the background region, so as to achieve the image salient region extraction. The experimental results verify that this paper algorithm not only realize image salient region extraction and guarantee the integrality of the salient region content, show good visual effect, improve the quality of the image extraction.
Min L.,Shandong College of Electronic Technology |
Shoulin L.,Pu Lian Software Ltd Liability
International Journal of Smart Home | Year: 2015
While there are a majority of enterprises using the J2EE technological structure design and solutions, it is difficult to fulfill the requirements in the complex and various, multi-point and wide-spread business implementation level by relying solely on object and interface oriented technology. SmartADF (Smart Application Develop Framework-Java) is a rapid application development framework. It is a framework suitable for developing the enterprise applications software based on J2EE framework. It adopts a strict hierarchical design. Hence, different levels of developers can find the right level to work. The use of interface coupling between the layers could be realized. It has been proved by the practice result that SmartADF provides a more abundant set of components and technical specifications to develop fast, stable and efficient enterprise applications compared with the traditional J2EE development framework, which improves the efficiency of developers. © 2015 SERSC.
Leng Y.,Shandong Normal University |
Qi G.-H.,Shandong Jiaotong University |
Xu X.-Y.,Shandong College of Electronic Technology
Radioengineering | Year: 2013
To construct a classification system or a detection system, large amounts of labeled samples are needed. However, manual labeling is dull and time consuming, so researchers have proposed the active learning technology. The initial training set selection is the first step of an active learning process, but currently there have been few studies on it. Most active learning algorithms adopt random sampling or algorithms like sampling by clustering (SBC) to select the initial training samples. But these two kinds of method would lose their effectiveness in detecting events of small probability because sometimes they could not select or select too few samples of the small probability events. To solve this problem, this paper proposes a BIC based initial training set selection algorithm. The BIC based algorithm performs clustering on the whole training set first, then uses BIC to judge the status of clusters. Finally, it adopts different selection strategies for clusters of different status. Experimental results on two real data sets show that, compared to random sampling and SBC, the proposed BIC based initial training set selection algorithm can efficiently solve the detection problem of small probability events. In the mean time, it has obvious advantages in detecting events of non-small probability.
Leng Y.,Shandong Normal University |
Xu X.,Shandong College of Electronic Technology |
Qi G.,Shandong Jiaotong University
Knowledge-Based Systems | Year: 2013
One key issue for most classification algorithms is that they need large amounts of labeled samples to train the classifier. Since manual labeling is time consuming, researchers have proposed technologies of active learning and semi-supervised learning to reduce manual labeling workload. There is a certain degree of complementarity between active learning and semi-supervised learning, and therefore some researches combine them to further reduce manual labeling workload. However, researches on combining active learning and semi-supervised learning for SVM classifier are rare. Of numerous SVM active learning algorithms, the most popular is the one that queries the sample closest to the current classification hyperplane in each iteration, which is denoted as SVMAL in this paper. Realizing that SVMAL is only interested in samples that are more likely to be on the class boundary, while ignoring the usage of the rest large amounts of unlabeled samples, this paper designs a semi-supervised learning algorithm to make full use of the rest non-queried samples, and further forms a new active semi-supervised SVM algorithm. The proposed active semi-supervised SVM algorithm uses active learning to select class boundary samples, and semi-supervised learning to select class central samples, for class central samples are believed to better describe the class distribution, and to help SVMAL finding the boundary samples more precisely. In order not to introduce too many labeling errors when exploring class central samples, the label changing rate is used to ensure the reliability of the predicted labels. Experimental results show that the proposed active semi-supervised SVM algorithm performs much better than the pure SVM active learning algorithm, and thus can further reduce manual labeling workload. © 2013 Elsevier B.V. All rights reserved.
Qi W.,Shandong Jianzhu University |
Liu X.,Shandong College of Electronic Technology |
Zhao J.,Shandong University
CSAE 2012 - Proceedings, 2012 IEEE International Conference on Computer Science and Automation Engineering | Year: 2012
This paper addresses flower image classification. The extent of blossom, deformation and inter-class appearance blur of flowers add great difficulties to flower classification task in addition to view, color, illumination changes that commonly occurred in other objects classification tasks. In this paper, SIFT-like feature descriptors and feature context method are used in coding local and spatial information, then LibLinear SVM classifier is employed for classification. Experimental results show that CSIFT is more robust and stable than SIFT and Dense SIFT in representing flower image. The accuracy of classification with CSIFT and feature context is comparable to state-of-the-art method. Since we do not need segment flower out of image in advance, practically, our method is better in performance and efficiency. © 2012 IEEE.
Xu X.-Y.,Shandong College of Electronic Technology |
Liu Z.-G.,Shandong University of Finance and Economics
Proceedings - 17th IEEE International Conference on Computational Science and Engineering, CSE 2014, Jointly with 13th IEEE International Conference on Ubiquitous Computing and Communications, IUCC 2014, 13th International Symposium on Pervasive Systems, Algorithms, and Networks, I-SPAN 2014 and 8th International Conference on Frontier of Computer Science and Technology, FCST 2014 | Year: 2015
A fast macroblock mode decision scheme is proposed in this work for arbitrary downsizing video transcoder based on H.264/AVC. For macroblock to be re-encoded in downsized frame, the corresponding area in the pre-encoded frame is not aligned with the boundary of blocks when downsizing factor is arbitrary. The concepts of 'mapped block' and 'overlapped block' are proposed in this paper to represent the corresponding area. To reduce the computational complexity, method based on blocks or based on pixels is adopted according to the value of downsizing factor. We compared the compression performance and computational complexity of the proposed method with the full mode search method. The proposal can significantly reduce the computational complexity while the compression performance loss is slight. © 2014 IEEE.
Kang W.,China University of Petroleum - East China |
Zhang L.,China University of Petroleum - East China |
Liu C.,Shandong College of Electronic Technology
Diangong Jishu Xuebao/Transactions of China Electrotechnical Society | Year: 2012
The power accumulator battery is sensitive to the charging current. But it has tinny inner resistance and opposing electromotive force, and ordinary converters will output super standard ripple wave when charging the power accumulator battery. The current-source converter can output steady current, but its capacity is limited by the energy storing inductance in the DC side. To resolve the above-mentioned problems, this paper compares the decay characteristics between Butterworth and Chebyshev filter in two aspects which include the order setting when the two filters are of the same stop-frequency decay and the filtering result and response speed when the two filters are of the same order, barrier frequency and decay. Based on the conclusion above, the DC filter parameter design method in the current-source converters is proposed. This method adopts filter parameters according to the appropriate arithmetic to reduce the energy storing inductance in the current-source converter. The voltage drop of the inductance is decreased and the system energy transfer efficiency is increased. This two power accumulator battery charging systems are designed. Simulation and test results demonstrate the correctness and feasibility of the filter design method.
Liu J.-M.,Jinan University |
Qu Q.,Shandong College of Electronic Technology
Proceedings - 3rd International Symposium on Information Processing, ISIP 2010 | Year: 2010
A substitution-diffusion based image cipher using chaotic standard and logistic maps was proposed recently. However, the cipher can be broken with a simple method by using only one known plain and cipher image pair. The attacking process need not to be complex because of the two vital defects of the encryption algorithm. Suggestions for improving the security of the algorithm are also provided. © 2010 IEEE.
Wang S.-Z.,Shandong College of Electronic Technology
International Journal of Multimedia and Ubiquitous Engineering | Year: 2014
In this paper, based on the K - SVD and residual error than the low SNR image sparse representation denoising algorithm. On the basis of the foregoing contents, this paper expounds on the build process and mechanism analysis of the algorithm, the paper on the basis of the subjective evaluation reference peak signal-to-noise ratio (PSNR) as the objective evaluation standard. Can be seen from the results of simulation experiments for different kinds of image denoising, image sparse decomposition based on a complete atom library has a better effect of denoising algorithm often, this is because after complete the atoms in the dictionary has redundancy, to show more abundant characteristic information, can more effectively extract the image features. In the proposed algorithm, the K - SVD algorithm for image sparse decomposition to optimize dictionary and residual error than the threshold for accurate division of image information and provide evidence for image noise effectively, a combination of both in image denoising, especially in low SNR image denoising experiment obtained good effect. The experimental results also from another side shows the sparse decomposition based on a complete atom library on image denoising application potential. © 2014 SERSC.
Feng Z.,Shandong College of Electronic Technology
Applied Mechanics and Materials | Year: 2014
Excessive working pressure and psychological stimulate make people a great stress. The purpose of this study was to investigate the teachers' job pressure, job performance and their relationship. The main methods is questionnaires, the sample consists of 250 employees from seven higher universities. The results indicate that job satisfaction is not independent in all job facets and that satisfaction with one facet might lead to satisfaction with another. © (2014) Trans Tech Publications, Switzerland.