Advanced Materials Research Vols. 532-533

Abstract: To determine the optimal thresholds in image segmentation, a new multilevel thresholding method based on improved particle swarm optimization (IPSO) is proposed in this paper. Firstly, use the conception of independent peaks to divide the histogram to several regions, secondly, the optimization object function using maximum between-class variance (MV) method can be gotten in each area, by the non-uniform mutation and Geese-LDW PSO optimization of the object function, the optimal thresholds can be gotten, and the image can be segmented with the thresholds. Compared with the basic MV algorithm and genetic algorithm (GA) modified MV, the experimental results show that the new method not only realizes the image segmentation well, but also improves the speed.

Abstract: An efficient and robust star acquisition algorithm based on facet fitting is presented to improve the performance of star sensors. The location of star central pixels can be determined by searching extremum intensity pixels among the point spread function (PSF) of stars, which is well fitted by the cubic facet model. According to extremum theory, the second derivative operators are pre-calculated and the searching process can be completed using convolution operations thrice. Simultaneously, cluster formation is also a time consuming routine, which is accomplished using specific maximum and minimum threshold to speed up it. A variety of experiments are carried out to validate the performance of proposed algorithm, moreover, the performance evaluation index M is presented. The results clearly show that the proposed algorithm makes a great progress than the vector method in time expense and accuracy under intense noise conditions.

Abstract: The development of the internet and exponential growth of network information produce a large number of duplicated pages on the network, reducing the retrieval of recall and precision and affecting the retrieval efficiency. The accuracy of the web, therefore, influences the quality of search engine. On the basis of the structural text description, this paper proposes an improved eliminating repetitive algorithm method, which is based on MD5 of Near-replicas. It proves that the method has a good effect on improving the recall and the precision through experiment.

Abstract: In view of the virtue and shortage of genetic algorithm and BP network, this paper proposes a new BP network training method based on improved genetic algorithm (IGA-BP). This algorithm uses hierarchical code, adaptive crossover and mutation, pruning similar chromosomes, dynamic supply new chromosomes and other operations, so the network structure and weight are optimized at the same time and the "premature" phenomenon is avoided. The simulation results show that the IGA-BP network architecture is simple, the convergence rate is quick, and has good approximation and generalization ability.

Abstract: This thesis proposes a new algorithm of the Chaos-based audio data hiding. The Chaos theory is introduced in design a new algorithm of the audio data hiding: with one section of audio as the watermarking, the Chaotic sequences select one part of the original audio signal as the carrier, and then embed the Chaos-encrypted audio watermarking into the carrier’s wavelet coefficients. Experimental results show that embedded watermark is imperceptibility and robust to many attacks, such as noise adding, re-sampling, low pass filtering, reverberation, MP3 compression and re-quantization and so on.

Abstract: This paper proposes an interactive video foreground segmentation method based on modeling and graph cut algorithm. User interactions are required at initial frame or key frame of video sequence at first. Secondly we make use of user interactions information to develop background/foreground model and get foreground segmentation result of the current frame in term of graph cut algorithm. And automatic updated methods are proposed to obtain foreground segmentation results automatically on the later sequence of video without user interaction. The developed system of interactive video foreground segmentation has performances with extracting object and editing segmentation results. Experimental results on kinds of video demonstrated that our interactive segmentation system is efficient.

Abstract: We present an efficient K-shortest paths routing algorithm for computer networks. This Algorithm is based on enhancements to currently used link-state routing algorithms such as OSPF and IS-IS, which are only focusing on finding the shortest path route by adopting Dijkstra algorithm. Its desire effect to achieve is through the use of K-shortest paths algorighm, which has been implemented successfully in some fileds like traffic engineering. The correctness of this Algorithm is discussed at the same time as long as the comparison with OSPF.

Abstract: This paper introduces fundamental theory and mathematic model of Support Vector Machine(SVM), and also covers applying SVM algorithm in data assorting. In conventional SVM model, sample set always has noisy and isolated points, for solving this problem this paper proposes a SVM boundary sample cut algorithm: first, pre-select boundary samples, then apply Remove-Only algorithm to remove some inappropriate points, then the result will be final sample set for SVM. At last, we compared conventional and improved algorithms by applying them on categorizing two medical data sets; the accuracy of improved algorithm achieves 100%. The result shows this improved algorithm is with significant practical advantage and value.

Abstract: In this study, a genetic algorithm simulating human reproduction mode (HRGA) is proposed. The genetic operators of HRGA include selection operator, help operator, crossover operator and mutation operator. The sex feature, age feature and consanguinity feature of genetic individuals are considered. Two individuals with opposite sex can reproduce the next generation if they are distant consanguinity individuals and their age is allowable. Based on this genetic algorithm, an improved evolutionary neural network algorithm named HRGA-BP algorithm is formed. In HRGA-BP algorithm, HRGA is used firstly to evolve and design the structure, the initial weights and thresholds, the training ratio and momentum factor of neural network roundly. Then, training samples are used to search for the optimal solution by the evolutionary neural network. HRGA-BP algorithm is used in motor fault diagnosis. The illustrational results show that HRGA-BP algorithm is better than traditional neural network algorithms in both speed and precision of convergence, and its validity in fault diagnosis is proved.

Abstract: Ship-radiated noises are able to reflect feature of same type vessel. This paper designed a general algorithm framework for ship-radiated noise target recognition, and given the design and implementation of key modules. The algorithm framework can improve the identification accuracy, support multi-dimensional feature space , multi-classify unit and heterogeneous classifier, overcome shortcoming of identify target property of single feature and single classifier from only one aspect, so it has widely applicability. The sample assessment mechanism is introduced by the framework, which can eliminate low quality sample in order to improve the identification accuracy and efficiency of training. A simulation example is given and the results shows that the method is reasonable and effective.