Let G ⊆ GL;n; (R be a finite matrix group and X⊆Rn be a G-variety. We propose a new approach for computing a stratification of X with respect to the orbit type of Rn respectively of the quotient X/G and we present new algorithms for this task. For X = Rn these algorithms yield an optimal description of each stratum and of the orbit space in terms of polynomial equations and inequalities (optimal with respect to the number of inequalities). Moreover we show that the dimension d of a stratum ∑ ;d; of R;n;/G is an upper and lower bound for the number of inequalities needed for a description of ∑ ;d; and its closure, which improves the upper bound d(d+1)/2, which holds for general basic closed semialgebraic sets of dimension d. Additionally, our algorithms allow to compute strata of particular interest of X/G, which demands less computational resources. By performing computations as long as possible in Rn (and not in Rn/G) and by refining results of Procesi and Schwarz, it seems that our algorithms are more efficient than the present approach. We conclude by giving an application of our algorithms to the problem of constructing a potential for Nickel-Titanium alloys and compare the runtime with other algorithms.

While current postal address readers have been adapted and fine tuned once and off-line in a central development lab, the concept of next generation readers presented here will contain a built-in learning capability which can be referred to as continuous learning from letters. This will be performed on different levels of reading and comprehension, enabling the system to adapt itself to slow changes of mail mix and writing conventions. Preconditions of an adaptive reading system are discussed and an outline of such a system is given, which not only keeps track, with the slow changes of input, but also optimizes itself to sire specific conditions, thus guaranteeing an optimal solution at each site and at any time within life cycle. The system presented here is funded by BMBF (German Federal Ministry of Education and Research) within the projects READ and ADAPTIVE READ.

This paper presents a generic system which automatically extracts the requested items from invoices with arbitrary form layout in arbitrary domains. The system consists of two components, an OCR tool which need not be adapted to the current domain and an information extraction component FRESCO which contains the knowledge about the domain. The automation rate for interpreting invoices in the domain of health insurance is above 50% with an error rate below 1% with respect to the items to be extracted.

Extracting structural information from paper documents supports the daily document processing by, for example, automatically finding index terms, document topics, etc. Knowledge about such components are modeled in a semantic net, which describes geometric properties, spatial relationships, lexical entities as well as lexical relationships. The document model is used to extract the sender, date, recipient, opening and closing formula from a business letter. 181 business letters have been processed, divided into a training set of 20 and the remaining ones for testing. The error rates for the test set range from 0.022 to 0.049 by an average rejection rate of 0.4. Results show that the computational effort can be limited to O(n/sup 2/) given n primitive objects for matching.

A document analysis system which is capable of extracting the semantics of specific text portions of structured documents is presented. The architecture of the analysis system is based on a knowledge representation scheme, a semantic network, called Resco-Frame Representation of Structured Documents. It allows the definition of knowledge about document components as well as knowledge about analysis algorithms in a uniform, simple, but powerful representation formalism. Hence, this architecture enables the analysis system to exploit the specific power of both the algorithmic knowledge describing the properties of algorithms and the declarative knowledge about properties of text objects in documents. The inference engine and the control algorithm show how these two knowledge sources are combined and utilized. The flexibility of the representation formalism Fresco, the recognition results and the computational complexity of the inference algorithm are presented in two different applications.>

In optical character recognition (OCR) and document analysis, many reading errors are not caused by inadequate classifier power, but by segmentation errors. In particular, merged characters are a major remaining problem. An efficient and powerful method of determining cut hypotheses for the segmentation of merged characters is presented. The method is based on a classifier deciding for each column of the character image, whether it represents a cut hypothesis or not. Since in the training phase the classifier is adapted by a sample set consisting of images of merged character patterns, the decision rules are created automatically rather than being man-made heuristics. The results obtained from a large test set show that a high recognition rate can be achieved with a reasonable computational effort.>

In optical character recognition (OCR) and document analysis many errors are not caused by inadequate classifier power, but by segmentation errors. Besides broken characters, merged characters constitute the major remaining problem. This paper presents an efficient method for segmenting merged characters. The algorithm combines a statistical adapted cut classifier and a search algorithm, which employs further experts for selecting the proper cut positions from a set of hypotheses. It is designed especially for proportional fonts and even succeeds, if the characters are in italics font style.>

The authors present a conceptual framework for solving the task of document analysis, which, in essence, consists in the conversion of the document's pixel representation into an equivalent knowledge network representation holding the document's content and layout. Starting on the pixel level, the formation of elementary geometric objects on which layout analysis as well as the definition of character objects is based is described. Character recognition accomplishes the mapping from geometric object to character meaning in ASCII representation. On the next level of abstraction words are formed and verified by contextual processing. Modeled knowledge about complete documents and about how their constituents are related to the application forms the highest level of abstraction. The various problems arising at each stage are discussed. The dependencies between the different levels are exemplified and technical solutions put forward.>

Synthetic aperture radar (SAR) images reveal radiometric image distortions that are caused by terrain undulations. The authors present the results of a study extracting and investigating the various components of these terrain influences. An imaging model, is set up for the geometric rectification of the SAR image and for a reconstruction of the imaging geometry. A prerequisite for the setup of this model is the use of a digital elevation model. Eight different geometric parameters are derived and investigated for their influence on grey-value variations in the geocoded SAR image. Image grey-value variations of three major land-use classes-forest, agricultural land, and urban/suburban areas-are examined. Empirical models of the SAR-backscatter variations are used to describe the relations between image grey values and various geometric parameters.>