兰州铁路局召开信息部门领导会议

2003年1月25日，兰州铁路局信息技术处在兰州组织召开全局信息部门领导会议，深入贯彻“十六”大信息化建设精神，总结2002年工作，讨论和修订2003年工作计划。

Vision based terrain reconstruction for planet rover using a special binocular bundle adjustment

This paper presents a pure vision based technique for 3D reconstruction of planet terrain. The reconstruction accuracy depends ultimately on an optimization technique known as 'bundle adjustment'. In vision techniques, the translation is only known up to a scale factor, and a single scale factor is assumed for the whole sequence of images if only one camera is used. If an extra camera is available, stereo vision based reconstruction can be obtained by binocular views. If the baseline of the stereo setup is known, the scale factor problem is solved. We found that direct application of classical bundle adjustment on the constraints inherent between the binocular views has not been tested. Our method incorporated this constraint into the conventional bundle adjustment method. This special binocular bundle adjustment has been performed on image sequences similar to planet terrain circumstances. Experimental results show that our special method enhances not only the localization accuracy, but also the terrain mapping quality.

FUZZY PRINCIPAL COMPONENT ANALYSIS AND ITS KERNEL-BASED MODEL

Principal Component Analysis(PCA)is one of the most important feature extraction methods,and Kernel Principal Component Analysis(KPCA)is a nonlinear extension of PCA based on kernel methods.In real world,each input data may not be fully assigned to one class and it may partially belong to other classes.Based on the theory of fuzzy sets,this paper presents Fuzzy Principal Component Analysis(FPCA)and its nonlinear extension model,i.e.,Kernel-based Fuzzy Principal Component Analysis(KFPCA).The experimental results indicate that the proposed algorithms have good performances.

Neural Network Based on Rough Sets and Its Application to Remote Sensing Image Classification

This paper presents a new kind of back propagation neural network (BPNN) based on rough sets,called rough back propagation neural network (RBPNN).The architecture and training method of RBPNN are presented and the survey and analysis of RBPNN for the classification of remote sensing multi_spectral image is discussed.The successful application of RBPNN to a land cover classification illustrates the simple computation and high accuracy of the new neural network and the flexibility and practicality of this new approach.

Mining Maximal Frequent Patterns in a Unidirectional FP-tree

Because mining complete set of frequent patterns from dense database could be impractical, an interesting alternative has been proposed recently. Instead of mining the complete set of frequent patterns, the new model only finds out the maximal frequent patterns, which can generate all frequent patterns. FP-growth algorithm is one of the most efficient frequent-pattern mining methods published so far. However, because FP-tree and conditional FP-trees must be two-way traversable, a great deal memory is needed in process of mining. This paper proposes an efficient algorithm Unid_FP-Max for mining maximal frequent patterns based on unidirectional FP-tree. Because of generation method of unidirectional FP-tree and conditional unidirectional FP-trees, the algorithm reduces the space consumption to the fullest extent. With the development of two techniques: single path pruning and header table pruning which can cut down many conditional unidirectional FP-trees generated recursively in mining process, Unid_FP-Max further lowers the expense of time and space.

BOOTSTRAPPING FOR EXTRACTING RELATIONS FROM LARGE CORPORA

A new approach of relation extraction is described in this paper. It adopts a bootstrap- ping model with a novel iteration strategy, which generates more precise examples of specific relation. Compared with previous methods, the proposed method has three main advantages: first, it needs less manual intervention; second, more abundant and reasonable information are introduced to represent a relation pattern; third, it reduces the risk of circular dependency occurrence in bootstrapping. Scalable evaluation methodology and metrics are developed for our task with comparable techniques over TianWang 100G corpus. The experimental results show that it can get 90% precision and have excellent expansibility.

Understanding visual-auditory correlation from heterogeneous features for cross-media retrieval

Cross-media retrieval is an interesting research topic,which seeks to remove the barriers among different modalities.To enable cross-media retrieval,it is needed to find the correlation measures between heterogeneous low-level features and to judge the semantic similarity.This paper presents a novel approach to learn cross-media correlation between visual features and auditory features for image-audio retrieval.A semi-supervised correlation preserving mapping(SSCPM)method is described to construct the isomorphic SSCPM subspace where canonical correlations between the original visual and auditory features are further preserved.Subspace optimization algorithm is proposed to improve the local image cluster and audio cluster quality in an interactive way.A unique relevance feedback strategy is developed to update the knowledge of cross-media correlation by learning from user behaviors,so retrieval performance is enhanced in a progressive manner.Experimental results show that the performance of our approach is effective.

Trimming self-intersections in swept volume solid modeling

Swept volume solid modeling has been applied to many areas such as NC machining simulation and verification, robot workspace analysis, collision detection, and CAD. But self-intersections continue to be a challenging problem in the boundary representation of swept volume solids. A novel algorithm is presented in this paper to trim self-intersection regions in swept volume solids modeling. This trimming algorithm consists of two major steps: (1) roughly detecting self-intersection regions by checking intersections or overlapping of the envelop profiles; (2) splitting the whole envelop surfaces of the swept volume solid into separate non-self-intersecting patches to trim global self-intersections, and to trim local self-intersections, dividing local self-intersecting regions into patches and replacing self-intersecting patches with non-self-intersecting ones. Examples show that our algorithm is efficient and robust.

A Distributed Data Mining System Based on Multi-agent Technology

Distributed Data Mining is expected to discover preciously unknown, implicit and valuable information from massive data set inherently distributed over a network. In recent years several approaches to distributed data mining have been developed, but only a few of them make use of intelligent agents. This paper provides the reason for applying Multi-Agent Technology in Distributed Data Mining and presents a Distributed Data Mining System based on Multi-Agent Technology that deals with heterogeneity in such environment. Based on the advantages of both the CS model and agent-based model, the system is being able to address the specific concern of increasing scalability and enhancing performance.

Modeling and designing for ship steering integrated simulator

The modeling of a ship steering integrated simulator(SSIS)applied to the design,debugging and maintenance of an autopilot is discussed.A nonlinear responsive model is proposed and applied to the design of SSIS.The SSIS generates real signals of the ship heading,the rudder angle,the ship position and the output to the autopilot.A variety of factors,such as ship speed variety,shallow water effect,nonlinearity of yaw and actuator,and environmental disturbances like wind,wave and current are considered carefully.Detailed formulas for calculating relevant parameters are provided.Taken a naval ship as an example,the physical-digital simulations on SSIS and the digital simulation on a Marine System Simulator(MSS)were conducted separately in various sailing conditions.Simulation results show that the simple nonlinear responsive model can be applied to ship motion control and simulation with sufficient accuracy and effectiveness.

STUDY ON EXTRACTING METHODS OF BURIED GEOLOGICAL INFORMATION IN HUAIBEI COAL FIELD

It is discussed features and tbe producing mechanism of buried geological information in geological, geophysical and remote sensing data in Huaibei coal field, and studied the methods extracting buried tectonic and igneous rock information from various geologicaI data using digital image processing techoiques.

A Hybrid Task Scheduling Algorithm in Grid

Task scheduling in Grid has been proved to be NP-complete problem. In this paper, to solve this problem, a Hybrid Task Scheduling Algorithm in Grid (HTS) has been presented, which joint the advantages of Ant Colony and Genetic Algorithm. Compared with the related work, the result shows that the HTS algorithm significantly surpasses the previous approaches in schedule length ratio and speedup.

Multiuser Pre-Coding Aided Quadrature Spatial Modulation for Large-Scale MIMO Channels

Pre-coding aided quadrature spatial modulation(PQSM) is a promising multiple input multiple output(MIMO) transmission technology. The multiuser(MU) detection in PQSM system is investigated in this paper. Based on the known channel state information, pre-coding matrix is designed to pre-process the in-phase and quadrature signals of quadrature spatial modulation(QSM) to reduce the inter-channel interference. In order to lower the complexity at the receiver brought by the orthogonality of the PQSM system, an orthogonal matching pursuit(OMP) detection algorithm and a reconstructed model are proposed. The analysis and simulation results show that the proposed algorithm can obtain a similar bit error rate(BER) performance as the maximum likelihood(ML) detection algorithm with more than 80% reduction of complexity.

A Scalable Communication Mechanism in Service Emergence Based on Bio-network

A scalable communication mechanism is proposed for service emergence based on bio-network.Service emergence is a novel model inspired by the characteristics of emergence and self-evolution in biological neuroendocrine and immune system,emergent communication means a group communication of mobile bio-entities.A series of protocol and algorithm are presented within frequently bio-entities migration and failure,it includes distribution and parallelization of message propagation method,a token-ring protocol that considerably improves the performance of emergence,and failure detection mechanisms.Experiment results show the desired capability via the proposed solution.

Connecting Language with Perception and Action for Human-robot Interaction

To build robots that engage in intuitive communication with people by natural language, we are developing a new knowledge representation called conceptual network model. The conceptual network connects natural language concepts with visual perception including color perception, shape perception, size perception, and spatial perception. In the implementation of spatial perception, we present a computational model based on spatial template theory to interpret qualitative spatial expressions. Based on the conceptual network model, our mobile robot can understand user's instructions and recognize the object referred to by the user and perform appropriate action. Experimental results show our approach promising.

On the Relation Redundancy in Fuzzy Databases

This paper concentrates on the problem of data redundancy under the extended-possibility-based model. Based on the information gain in data classification, a measure - relation redundancy - is proposed to evaluate the degree of a given relation being redundant in whole. The properties of relation redundancy are also investigated. This new measure is useful in dealing with data redundancy.

THE THEORETICAL BER PERFORMANCE ANALYSIS OF A COOPERATIVE DIVERSITY SCHEME IN FREQUENCY SELECTIVE FADING CHANNEL

A novel cooperative diversity scheme based on Distributed Space-Time Block Coding and Multi-Carrier Code Division Multiple Access (DSTBC-MC-CDMA) is proposed which works well in frequency selective fading channels with multiple single-antenna users. And an analytical error model is established to describe the symbol decoding errors between interusers, based on which a close form expression for theoretical Bit Error Rate (BER) performance of the scheme is derived to analyze the influence of the interuser decoding errors on the BER performance of the scheme. Then simulation is complimented to verify the analytic result above, which also shows that the BER performance of DSTBC-MC-CDMA outgoes that of non-cooperative MC-CDMA with considerable gains. Further- more, the simulations coincide with the theoretical results well.

Application of fuzzy logic in weighted information of fusion fingerprints and palm prints

It is a developing job to distinguish identifications with information fusion of fingerprints and palm prints. It is also a very effective way to resolve the problem of low identification rate and low stability of single biology characteristic identification. Based on the theory of fuzzy logic theory, we bring out the method of obfuscating weigh coefficient and reliability to fuse the information of fingerprints and palm prints to realize high identification rate. The experiment proves the feasibility and effectiveness of this method and the identification rate can be more than 90%, which contributes useful experience to the research of identification using biology characteristics.

IMPROVING VOICE ACTIVITY DETECTION VIA WEIGHTING LIKELIHOOD AND DIMENSION REDUCTION

The performance of the traditional Voice Activity Detection （VAD） algorithms declines sharply in lower Signal-to-Noise Ratio （SNR） environments. In this paper, a feature weighting likelihood method is proposed for noise-robust VAD. The contribution of dynamic features to likelihood score can be increased via the method, which improves consequently the noise robustness of VAD. Divergence based dimension reduction method is proposed for saving computation, which reduces these feature dimensions with smaller divergence value at the cost of degrading the performance a little. Experimental results on Aurora Ⅱ database show that the detection performance in noise environments can remarkably be improved by the proposed method when the model trained in clean data is used to detect speech endpoints. Using weighting likelihood on the dimension-reduced features obtains comparable, even better, performance compared to original full-dimensional feature.