标书文档识别处理系统VHTender的设计与实现

本文介绍了英文标书文档处理系统VHTender的设计和实现过程，该系统能够从纸质标书的扫描图象中抽取文档信息，将其转换成电子标书，本文从功能实现的角度，介绍了该系统在一些关键性技术中所采用的方法和策略。     

自然图像中混叠的概率模型

我们为建立自然图像中混叠的概率模型提供了一个数学框架。假设：(1)图像是由三维物体的二维投影轮廓组成．(2)在简化条件下．我们推导出自然图像的一种解析概率密度函数。已经证明．这种解析表达式可以很好地比配观测到的密度函数(直方图)。我们首先推导图像像素的边缘概率密度函数，然后推广到多维边缘概率密度函数，最后提出了几种改进措施。     

Learning mixtures of point distribution models with the EM algorithm

This paper demonstrates how the EM algorithm can be used for learning and matching mixtures of point distribution models. We make two contributions. First, we show how shape-classes can be learned in an unsupervised manner. We present a fast procedure for training point distribution models using the EM algorithm. Rather than estimating the class means and covariance matrices needed to construct the PDM, the method iteratively refines the eigenvectors of the covariance matrix using a gradient ascent technique. Second, we show how recognition by alignment can be realised by fitting a mixture of linear shape deformations. We evaluate the method on the problem of learning the class-structure and recognising Arabic characters.     

Stochastic models for recognition of occluded targets

Recognition of occluded objects in synthetic aperture radar (SAR) images is a significant problem for automatic target recognition. Stochastic models provide some attractive features for pattern matching and recognition under partial occlusion and noise. In this paper, we present a hidden Markov modeling based approach for recognizing objects in SAR images. We identify the peculiar characteristics of SAR sensors and using these characteristics we develop feature based multiple models for a given SAR image of an object. The models exploiting the relative geometry of feature locations or the amplitude of SAR radar return are based on sequentialization of scattering centers extracted from SAR images. In order to improve performance we integrate these models synergistically using their probabilistic estimates for recognition of a particular target at a specific azimuth. Experimental results are presented using both synthetic and real SAR images.     

Analysis of a Classifier with Random Thresholds

A neural classifier with random thresholds is considered. Probabilistic analysis of functional characteristics depending on the classifier parameters is performed, and recommendations for their selection are made. The classifier structure optimization is proposed for input data distribution.     

Statistical Pattern Modeling in Vision-Based Quality Control Systems

Machine vision technology improves productivity and quality management and provides a competitive advantage to industries that employ this technology. In this article, visual inspection and quality control theory are combined to develop a robust inspection system with manufacturing applications. The inspection process might be defined as the one used to determine if a given product fulfills a priori specifications, which are the quality standard. In the case of visual inspection, these specifications include the absence of defects, such as lack (or excess) of material, homogeneous visual aspect, required color, predetermined texture, etc. The characterization of the visual aspect of metallic surfaces is studied using quality control chars, which are a graphical technique used to compare on-line capabilities of a product with respect to these specifications. Original algorithms are proposed for implementation in automated visual inspection applications with on-line execution requirements. The proposed artificial vision method is a hybrid between the two usual methods of pattern comparison and theoretical decision. It incorporates quality control theory to statistically model the pattern for defect-free products. Specifically, individual control charts with 6-sigma limits are set so the inspection error is minimized. Experimental studies with metallic surfaces help demonstrate the efficacy and robustness of the proposed methodology.     

Speech-as-data technologies for personal information devices

 For small, portable devices, speech input has the advantages of low-cost and small hardware, can be used on the move or whilst the eyes & hands are busy, and is natural and quick. Rather than rely on imperfect speech recognition we propose that information entered as speech is kept as speech and suitable tools are provided to allow quick and easy access to the speech-as-data records. This paper summarises our work on the technologies needed for these tools – for organising, browsing, searching and compressing the stored speech. These technologies go a long way towards giving stored speech the characteristics of text without the associated input problems. Received: 5 March 2002 / Accepted: 1 September 2002 Nick Haddock Consultant Acknowledgements The authors would like to thank the whole HP Labs Gryphon team for their valuable contributions to this work - Mike Collins for the hierarchical chunking algorithm, Erik Geelhoed and David Frohlich for the users perspective, Richard Hull for starting off the compression work, Steve Loughran for productisation, and Dave Reynolds for his consistent advice and support. We would also like to thank our partners at Cambridge University, Steve Young and Tony Robinson, whose expertise and technology formed the foundation for this work, as well as the efforts of Kate Knill on wordspotting, Carl Seymour on compression, James Christie on recognition, and Robin Valenza whose brief excursion into the world of speech technology helped develop a simple and effective summarisation technique. Finally we would like to thank the reviewers for their many helpful comments.     

系统生存性分析与需求定义

通过对计算机分布式系统安全问题的介绍 ,提出了生存性的概念 ,并对其进行了分析 ,给出了生存性的定义。生存性的着重点在于当系统受到攻击 ,甚至已被成功入侵的情况下保护关键性服务。     

藏文自动分词系统的设计与实现

藏文自动分词系统的研制目前在国内仍是空白。本文从四个方面详细报告了书面藏文自动分词系统的具体实现过程,内容包括系统结构、分词知识库的组织与实现以及分词策略、算法设计及其详细的自动分词过程实例。文章最后给出了实验结果,结果表明系统具有较高的切分精度和较好的通用性。     

基于神经网络的入侵识别方法

分析了正常和异常的网络行为模式所带来的网络传核数据包的不同特征，提出了基于神经网络的入侵识别方法。采用带偏差单元的回归BP算法使神经网络收敛性能良好，并对端口扫描类型的攻击进行了仿真识别。