Focusing on the problem of high overhead and frequent overflow of the counter mode encryption, this paper proposed an efficient scheme to protect data confidentiality and integrity. Based on the local characteristic of data accessing, the scheme set different counter length for memory area according to different accessing frequencies and the counter length can be dynamically adjusted. The analysis and the simulation results indicated that compared with the counter mode encryption, the scheme can decrease memory space overhead and the number of overflow. The proposed scheme can be applied to other schemes of protecting confidentiality and integrity based on counters and can satisfy performance requirements for most applications.

In order to improve the bandwidth utilization of IP over WDM (Wavelength Division Multiplexing) optical Internet, a static traffic grooming scheme was proposed by introducing the knowledge of fuzzy mathematics and microeconomics. It introduced the user traffic request delay satisfaction degree and the network relative cost. With the help of gaming analysis and layered graph, based on predatory search algorithm (PSA), it tried to find the optimal grooming solution with the comprehensive user QoS (Quality of Service) satisfaction degree maximized and the network relative cost minimized. It was simulated over actual network topologies and its performance was compared with certain existing traffic grooming scheme. Simulation results have shown that it is both feasible and effective with better performance.

Hungarian method is a classical method for solving assignment problems. It also can be widely used in other problems, such as matching problem. This paper researches its application on using structural index reduction method to solve high-index DAEs, based on the combinatorial relaxation theory. Combinatorial relaxation theory converts the complex mathematical problem to the matching problem of bipartite graph. Based on this theory this paper presents the main idea of Hungarian method and puts up three implementations for Hungarian method. At last, it compares the time performance of the three implementations by running a set of experiments.

Automatic image annotation is a critical and challenging problem in pattern recognition and image understanding areas. There are some problems in existing automatic image annotation areas. For example, the size of unlabeled data is much larger than the labeled data. Besides, most image annotation models can only use one kind of image segmentation strategy and certain image description method. According to the above problems, an automatic image annotation model based on Co-training is proposed. In this model, four independent feature properties are constructed and then four corresponding sub-classifiers are built. In this way, different image segmentation strategies and feature representation methods can be integrated into a unified framework. An adaptive algorithm based on vote and consistency is proposed to extend the training dataset. The proposed method use Co-training algorithm and mass unlabeled data to improve the performance of automatic image annotation. Experiments conducted on Corel 5 K dataset verify the effectiveness of proposed method.

Lip model plays an important role in virtual facial expression and speech simulation. However, the structures and movements of existing lip models are very simple, and the common motions such as protruding, pouting and sipping are failed to totally realize. In order to solve these problems, this paper presents a new double-layer muscle-controlled lip model. This method adopts the approach from anatomy to classify lip muscles into deep and superficial ones. The deep muscle is employed to control the corners of the mouth position and the size of mouth fissure. The superficial muscle is treated as a new arc muscle, and it uses catenary function to fix the parameters under different shrinkages and fit the process of protruding and sipping etc. The experiments show that the method can control the model flexibly and simulate some lip movements that accord with physiological motion characteristics.

Operation command transformation is one of key technologies of the synchronized collaborative design upon heterogeneous CAD systems and becomes a significant challenge because of the great difference in the modeling operations and underground data structures between different CAD systems. A feasible way is to build a neutral operation to aid the conversion of operations among different CAD systems. A method of constructing neutral modeling command set is proposed based on the general parametric feature modeling technique. The set is built by taking the intersection of operation commands from all commercial CAD systems according to parametric feature modeling operations. In order to accomplish the translation of system modeling command (SMC) between different CAD systems, a representation language of neutral modeling command (NMC) is presented to describe these neutral operation commands. By means of the language the translations of SMC-to-NMC and NMC-to-SMC are presented of operation commands from different CAD systems such as Pro/E, UG, SolidWorks. A prototype of synchronized collaborative design is developed based on two common CAD systems Pro/E and UG with their APIs and VC++. The experiments show that our middleware structure increases greatly the efficiency of the system.

Artificial immune network (AIN) as a branch of artificial immune system has been widely used in many application fields, and shows good ability of global optimization, especially in parameters optimization of the pharmacokinetic models. The search process of AIN for global optimum is based on the principles of clonal selection and immune network. However, as one of the heuristic-based optimal algorithms, the evolution of memory cells in the AIN is more time consuming compared with gradient-based optimal algorithms. In this paper, an AIN with distributed clonal selection strategy is proposed to improve the efficiency of the AIN. Then the distributed AIN is implemented with MATLAB Distributed Computing Engine (MDCE). One of the advantages of MDCE is that it is convenient to run optimal algorithms programmed with MATLAB platform. In the experiments, parameters of the [ 18 F] Fluoro-2-deoxy2D-glucose (FDG) tracer kinetic model are optimized with the distributed AIN algorithms, theory analysis and experiments results indicate the algorithm is capable of improving search speed significantly in successful rate and algorithm stability.

A new image fusion algorithm of the feature level is proposed incorporating with quantum-behaved particle swarm optimization algorithm (QPSO) clustering in this paper. Wavelet decomposition is performed on the source images, clustering algorithm that incorporates the fuzzy C- means (FCM) into QPSO algorithm (QPSO-FCM) is developed to segment the image in the feature space formed by multi-channel Gabor filters, and then the weighting factors are constructed. Finally, the fused image is obtained by taking inverse wavelet transform. The QPSO has less parameters and higher convergent capability of the global optimization. So QPSO-FCM has a strong global searching capacity and avoids the local minimum problems of FCM. The performance of the image fusion method is evaluated using five criteria including root mean square error, peek-to-peek signal-to-noise ratio, entropy, cross entropy and mutual information. Owing to the improved the clustering effect, the evaluation results indicate that the proposed algorithm outperforms the image fusion algorithm based on FCM.

It is well known that fuzzy clustering and level set are two important tools for image segmentation. The former focuses on analyzing the statistical characteristics of image features, while the latter aims to acquire the good geometrical continuity of segmentation boundaries. Obviously, two kinds of methods may be complement to each other. Inspired by this idea, a new level set model integrated with fuzzy c-means (FCM) clustering FCMLS has been presented in our previous studies. Compared with FCM and original level set methods, some remarkable characteristics and better performance have been demonstrated. In this paper, we mainly concern with the foreground extraction problem, wherein some extended works are also reported including detailed analysis on the convergence of FCMLS, multiregional FCMLS. Our research results indicate that FCMLS has good performance and is very worthful for foreground image extraction.