Advanced Materials Research Vols. 998-999

Abstract: A novel method based on Empirical Mode Decomposition(EMD) is approached to process the geometry signal. The main idea is to decompose the signal into some different detail components called Intrinsic Mode Function (IMF). The key steps are as follows: First, the signal is spherical parameterization; Second it is transformed into the plane signal and sampled regularly; Third, the translated signal is processed as an image using Bid-Empirical Mode Decomposition, getting several image IMFs; Finally, invert mapping these IMFs to geometry signal and getting the geometry signal’s IMFs.We demonstrate the power of the algorithms through a number of application examples including de-noising and enhancement.

Abstract: In recent years, mass incidents occurred frequently. In order to identify and warn the incidents proactively and timely. To this problem, we propose an algorithm based on adaptive LBP to estimate the crowd density. Firstly, use three-dimensional Hessian matrix to detect characteristic point. Secondly, use improved adaptive LBP to extract the dynamic texture and analyze it, then get the local feature. Thirdly, learn for global characteristic vectors, and then estimate the density level with support vector machine (SVM). Through simulation comparison, the density estimation method is more accurate and more real-time.

Abstract: In the course of the face feature match, many classifiers have been designed. The neural network is usually selected as a classifier because of its validity and universality, whereas its training time, training epochs, and its convergence, all are not satisfied to us. It is often influenced by the author’s experience. In the case, a collaborative genetic algorithm and neural network is presented as a new face recognition classifier. The one thing is to train the NN weights by the GA until the stopping criterion is met, and the next thing is to use the BP algorithm to continue to train the network. The training time and training epochs have been improved in the experiment of the face recognition on ORL face database. The simulation shows the validity of methods.

Abstract: To reduce the impact of irrelevant attributes on clustering results, and improve the importance of relevant attributes to clustering, this paper proposes fuzzy C-means clustering algorithm based on coefficient of variation (CV-FCM). In the algorithm, coefficient of variation is used to weigh attributes so as to assign different weights to each attribute in the data set, and the magnitude of weight is used to express the importance of different attributes to clusters. In addition, for the characteristic of fuzzy C-means clustering algorithm that it is susceptible to initial cluster center value, the method for the selection of initial cluster center based on maximum distance is introduced on the basis of weighted coefficient of variation. The result of the experiment based on real data sets shows that this algorithm can select cluster center effectively, with the clustering result superior to general fuzzy C-means clustering algorithms.

Abstract: This paper deals with maximum-likelihood (ML) detection of symbol sequence in the absence of synchronization information. A novel iterative scheme is proposed to obtain the ML estimates of the symbols without an estimation of synchronization parameters. Instead of the optimal sampling point recovery and explicit carrier phase compensation, the detection of symbols employs the direct calculation of the matched filter output, eliminating the need for a separate synchronizer. The detection problem is treated as ML estimation from incomplete data, which is solved by means of an iterative scheme based on the expectation-maximization algorithm. The proposed scheme is compared with conventional non-data-aided and iterative ML synchronizers. Accordingly, the simulation results indicate that the proposed detector enables improvements on both the bit error rate and convergence property.

Abstract: In View of the Multi-View Face Detection Problem under Complex Background, an Improved Face Detection Method Based on Multi-Features Boosting Collaborative Learning Algorithm Integrating Local Binary Pattern (LBP) is Presented. Firstly, Facial Skin Color Information is Used to Exclude most of the Background Regions. then, Haar-like Feature and LBP Feature are Extracted from Facial Candidate Regions and Inputted into a Modified Adaboost Algorithm to Obtain a Strong Classifier. Lastly, in Order to Improve the Detection Speed, Pyramid Classifier System Structure is Adopted to Determine the Face. the Experimental Results on CMU Standard Test Set and Life Photos, the Proposed Method has Achieved the Rapid Detection of Multi-View Face Image.

Abstract: Non-Line-of-Sight (NLOS) propagation problems badly degrade the accuracy of wireless mobile positioning algorithms, which incurs a large positive bias in the Time-of-Arrival (TOA) measurements. Under several assumptions, the Hankel matrix of TOA data can be decomposed into a low-rank distance matrix and a sparse error matrix. This paper utilizes the robust principal component analysis (RPCA) method to solve the decomposition problem. After estimating the distance, the positioning problem can use existing Line-of-Sight (LOS) based algorithms to calculate the coordinate of the mobile station (MS). Simulation results show our method outperforms other existing NLOS positioning methods and the RPCA based matrix decomposition process can eliminate NLOS effect efficiently.

Abstract: A novel palmprint recognition method based on sparse two-dimensional local discriminant projections (S2DLDP) is proposed. After a description of the basic theory and resolution method for S2DLDP, the paper presents the detail palmprint feature extraction method based on S2DLDP, and tests the algorithm performance by various non-zero elements size and neighborhood size. S2DLDP considerers the class information, local separability, two-dimensional image inherent properties of training samples and sparse projection, which provides an intuitive, semantic and interpretable feature subspace for palmprint representation. The optimal recognition accuracy of EER=2.2% is obtained on PolyU palmprint database, which also illuminates the effectiveness of the proposed algorithm.

Abstract: Existing data miming algorithms have mostly implemented data mining under centralized environment, but the large-scale database exists in the distributed form. According to the existing problem of the distributed data mining algorithm FDM and its improved algorithms, which exist the problem that the frequent itemsets are lost and network communication cost too much. This paper proposes a association rule mining algorithm based on distributed data (ARADD). The mapping marks the array mechanism is included in the ARADD algorithm, which can not only keep the integrity of the frequent itemsets, but also reduces the cost of network communication. The efficiency of algorithm is proved in the experiment.

Abstract: Mobile banking as a fusion of mobile technology and banking business is becoming the bank to develop business. This paper uses historical induction and logical reasoning, then combine variety research methods in general and specific measures theory research to study. Transferring value from the customer point of view, we design integrated marketing communication system of mobile banking solutions..