Pub Date : 1997-08-18DOI: 10.1109/ICDAR.1997.619868
Benoit Poirier, M. Dagenais
The proliferation of electronic document formats impedes the dissemination and management of documents. Indeed, a common format with structural information is required to obtain document indexing and navigation. While in some formats it is easy to decode and preserve the document structure information, often the only easily obtainable representation is Postscript, where only the geometrical information remains. Even if an organization is willing to convert all its document producing activities to a structure preserving format such as HTML, the existing documents need to be converted. The paper addresses the difficult problem of extracting the structure of a document from a geometrical representation. An interactive tool to extract the document content and structure from a geometric representation (Postscript) has been developed. It successfully analyzes several documents produced with different tools, and produces structural information using the HyperText Markup Language (HTML). The end user, when presented with the extracted document structure, can interactively modify it, if needed. The tool is easily extended to recognize new constructs and is aimed at organizations needing to convert numerous documents for searching and browsing on intranets or on the Internet.
{"title":"An interactive system to extract structured text from a geometrical representation","authors":"Benoit Poirier, M. Dagenais","doi":"10.1109/ICDAR.1997.619868","DOIUrl":"https://doi.org/10.1109/ICDAR.1997.619868","url":null,"abstract":"The proliferation of electronic document formats impedes the dissemination and management of documents. Indeed, a common format with structural information is required to obtain document indexing and navigation. While in some formats it is easy to decode and preserve the document structure information, often the only easily obtainable representation is Postscript, where only the geometrical information remains. Even if an organization is willing to convert all its document producing activities to a structure preserving format such as HTML, the existing documents need to be converted. The paper addresses the difficult problem of extracting the structure of a document from a geometrical representation. An interactive tool to extract the document content and structure from a geometric representation (Postscript) has been developed. It successfully analyzes several documents produced with different tools, and produces structural information using the HyperText Markup Language (HTML). The end user, when presented with the extracted document structure, can interactively modify it, if needed. The tool is easily extended to recognize new constructs and is aimed at organizations needing to convert numerous documents for searching and browsing on intranets or on the Internet.","PeriodicalId":435320,"journal":{"name":"Proceedings of the Fourth International Conference on Document Analysis and Recognition","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1997-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126402025","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1997-08-18DOI: 10.1109/ICDAR.1997.620583
F. Alimoglu, Ethem Alpaydin
We investigate techniques to combine multiple representations of a handwritten digit to increase classification accuracy without significantly increasing system complexity or recognition time. We compare multiexpert and multistage combination techniques and discuss in detail in a comparative manner methods for combining multiple learners: voting, mixture of experts, stacking, boosting and cascading. In pen based handwritten character recognition, the input is the dynamic movement of the pentip over the pressure sensitive tablet. There is also the image formed as a result of this movement. On a real world database, we notice that the two multi layer perceptron (MLP) neural network based classifiers using these representations separately make errors on different patterns, implying that a suitable combination of the two would lead to higher accuracy. Thus we implement and compare voting, mixture of experts, stacking and cascading. Combined classifiers have an error percentage less than individual ones. The final combined system of two MLPs has less complexity and memory requirement than a single k nearest neighbor using one of the representations.
{"title":"Combining multiple representations and classifiers for pen-based handwritten digit recognition","authors":"F. Alimoglu, Ethem Alpaydin","doi":"10.1109/ICDAR.1997.620583","DOIUrl":"https://doi.org/10.1109/ICDAR.1997.620583","url":null,"abstract":"We investigate techniques to combine multiple representations of a handwritten digit to increase classification accuracy without significantly increasing system complexity or recognition time. We compare multiexpert and multistage combination techniques and discuss in detail in a comparative manner methods for combining multiple learners: voting, mixture of experts, stacking, boosting and cascading. In pen based handwritten character recognition, the input is the dynamic movement of the pentip over the pressure sensitive tablet. There is also the image formed as a result of this movement. On a real world database, we notice that the two multi layer perceptron (MLP) neural network based classifiers using these representations separately make errors on different patterns, implying that a suitable combination of the two would lead to higher accuracy. Thus we implement and compare voting, mixture of experts, stacking and cascading. Combined classifiers have an error percentage less than individual ones. The final combined system of two MLPs has less complexity and memory requirement than a single k nearest neighbor using one of the representations.","PeriodicalId":435320,"journal":{"name":"Proceedings of the Fourth International Conference on Document Analysis and Recognition","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1997-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128079718","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1997-08-18DOI: 10.1109/ICDAR.1997.619805
Kyusik Chung, Jongmin Yoon
The paper presents a performance comparison of several feature selection methods based on neural network node pruning. Assuming the features are extracted and presented as the inputs of a 3 layered perceptron classifier, we apply the five feature selection methods before/during/after neural network training in order to prune only input nodes of the neural network. Four of them are node pruning methods such as node saliency method, node sensitivity method, and two interactive pruning methods using different contribution measures. The last one is a statistical method based on principle component analysis (PCA). The first two of them prune input nodes during training whereas the last three do before/after network training. For gradient and upper down, left right hole concavity features, we perform several experiments of handwritten English alphabet and digit recognition with/without pruning using the five feature selection algorithms, respectively. The experimental results show that node saliency method outperforms the others.
{"title":"Performance comparison of several feature selection methods based on node pruning in handwritten character recognition","authors":"Kyusik Chung, Jongmin Yoon","doi":"10.1109/ICDAR.1997.619805","DOIUrl":"https://doi.org/10.1109/ICDAR.1997.619805","url":null,"abstract":"The paper presents a performance comparison of several feature selection methods based on neural network node pruning. Assuming the features are extracted and presented as the inputs of a 3 layered perceptron classifier, we apply the five feature selection methods before/during/after neural network training in order to prune only input nodes of the neural network. Four of them are node pruning methods such as node saliency method, node sensitivity method, and two interactive pruning methods using different contribution measures. The last one is a statistical method based on principle component analysis (PCA). The first two of them prune input nodes during training whereas the last three do before/after network training. For gradient and upper down, left right hole concavity features, we perform several experiments of handwritten English alphabet and digit recognition with/without pruning using the five feature selection algorithms, respectively. The experimental results show that node saliency method outperforms the others.","PeriodicalId":435320,"journal":{"name":"Proceedings of the Fourth International Conference on Document Analysis and Recognition","volume":"215 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1997-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134431965","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1997-08-18DOI: 10.1109/ICDAR.1997.620627
F. Buddrus, Marco Bellavia
Deals with the idea of an object-oriented DBMS (O/sub 2/) working as an Internet server. The approach introduced offers fine-grained storage and retrieval of World Wide Web (WWW) documents, providing sophisticated document maintenance, supporting referential integrity, document integration, integration with application objects, access control and versioning. In contrast to comparable systems, it acts transparently to the WWW users, i.e. without using proprietary query languages or protocols. While achieving the full benefits of a modern document repository, it keeps the ease of the traditional file system it replaces.
{"title":"Surfing an ODBMS (maintaining WWW documents with O/sub 2/)","authors":"F. Buddrus, Marco Bellavia","doi":"10.1109/ICDAR.1997.620627","DOIUrl":"https://doi.org/10.1109/ICDAR.1997.620627","url":null,"abstract":"Deals with the idea of an object-oriented DBMS (O/sub 2/) working as an Internet server. The approach introduced offers fine-grained storage and retrieval of World Wide Web (WWW) documents, providing sophisticated document maintenance, supporting referential integrity, document integration, integration with application objects, access control and versioning. In contrast to comparable systems, it acts transparently to the WWW users, i.e. without using proprietary query languages or protocols. While achieving the full benefits of a modern document repository, it keeps the ease of the traditional file system it replaces.","PeriodicalId":435320,"journal":{"name":"Proceedings of the Fourth International Conference on Document Analysis and Recognition","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1997-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131059574","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1997-08-18DOI: 10.1109/ICDAR.1997.620559
D. Guillevic, C. Suen
We describe a hidden Markov model (HMM) based word recognition engine being developed to be integrated with the CENPARMI bank cheque processing system. The various modules are described in detail, and preliminary results are compared with our previous global feature recognition scheme. The engine is tested on words from a database of over 4,500 cheques of 1,400 writers.
{"title":"HMM word recognition engine","authors":"D. Guillevic, C. Suen","doi":"10.1109/ICDAR.1997.620559","DOIUrl":"https://doi.org/10.1109/ICDAR.1997.620559","url":null,"abstract":"We describe a hidden Markov model (HMM) based word recognition engine being developed to be integrated with the CENPARMI bank cheque processing system. The various modules are described in detail, and preliminary results are compared with our previous global feature recognition scheme. The engine is tested on words from a database of over 4,500 cheques of 1,400 writers.","PeriodicalId":435320,"journal":{"name":"Proceedings of the Fourth International Conference on Document Analysis and Recognition","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1997-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133695958","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1997-08-18DOI: 10.1109/ICDAR.1997.620647
H. Bunke, R. Ammann, Guido Kaufmann, T. M. Ha, M. Schenkel, R. Seiler, F. Eggimann
On-line recognition differs from off-line recognition in that additional information about the drawing order of the strokes is available. This temporal information makes it easier to recognize handwritten texts with an on-line recognition system. In this paper we present a method for the recovery of the stroke order from static handwritten images. The algorithm was tested by classifying the words of an off-line database with a state-of-the-art on-line recognition system. On this database with 150 different words, written by four cooperative writers, a recognition rate of 97.4% was obtained.
{"title":"Recovery of temporal information of cursively handwritten words for on-line recognition","authors":"H. Bunke, R. Ammann, Guido Kaufmann, T. M. Ha, M. Schenkel, R. Seiler, F. Eggimann","doi":"10.1109/ICDAR.1997.620647","DOIUrl":"https://doi.org/10.1109/ICDAR.1997.620647","url":null,"abstract":"On-line recognition differs from off-line recognition in that additional information about the drawing order of the strokes is available. This temporal information makes it easier to recognize handwritten texts with an on-line recognition system. In this paper we present a method for the recovery of the stroke order from static handwritten images. The algorithm was tested by classifying the words of an off-line database with a state-of-the-art on-line recognition system. On this database with 150 different words, written by four cooperative writers, a recognition rate of 97.4% was obtained.","PeriodicalId":435320,"journal":{"name":"Proceedings of the Fourth International Conference on Document Analysis and Recognition","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1997-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128834482","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1997-08-18DOI: 10.1109/ICDAR.1997.619837
C. Cracknell, A. Downton, L. Du
Virtually all form processing systems rely upon a form description file which holds some form design specific information. Such files are written in a form description language. This paper presents a novel form description language and a new approach to form processing which makes full use of the language. The form description language which is presented is novel in two ways: firstly, it allows the comprehensive utilisation of contextual information; secondly, it enables the storage of form models which economically describe the variations in the form design which are allowed for a particular class of forms.
{"title":"An object-oriented form description language and approach to handwritten form processing","authors":"C. Cracknell, A. Downton, L. Du","doi":"10.1109/ICDAR.1997.619837","DOIUrl":"https://doi.org/10.1109/ICDAR.1997.619837","url":null,"abstract":"Virtually all form processing systems rely upon a form description file which holds some form design specific information. Such files are written in a form description language. This paper presents a novel form description language and a new approach to form processing which makes full use of the language. The form description language which is presented is novel in two ways: firstly, it allows the comprehensive utilisation of contextual information; secondly, it enables the storage of form models which economically describe the variations in the form design which are allowed for a particular class of forms.","PeriodicalId":435320,"journal":{"name":"Proceedings of the Fourth International Conference on Document Analysis and Recognition","volume":"75 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1997-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127387349","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1997-08-18DOI: 10.1109/ICDAR.1997.620578
Jing Zheng, Xiaoqing Ding, Youshou Wu
The paper presents a novel method for online handwritten Chinese character recognition. In our method, each category of character is described by a fuzzy attributed relational graph (FARG). A relaxation algorithm is developed to match the input pattern with every FARG. For decision making, a similarity measure is established via statistical technique to calculate the matching degree between the input pattern and referenced FARG, according to which the recognition result is determined. The principle of our method makes it very robust against stroke connection and stroke order variation as well as stroke shape deformation. A database of 22530 samples collected from 6 subjects is used to test our recognition system which can recognize 3755 categories of Chinese characters. The result shows that our method is very effective: a top 1 recognition rate of 98.8% and a top 10 of 99.7% are reached.
{"title":"Recognizing on-line handwritten Chinese character via FARG matching","authors":"Jing Zheng, Xiaoqing Ding, Youshou Wu","doi":"10.1109/ICDAR.1997.620578","DOIUrl":"https://doi.org/10.1109/ICDAR.1997.620578","url":null,"abstract":"The paper presents a novel method for online handwritten Chinese character recognition. In our method, each category of character is described by a fuzzy attributed relational graph (FARG). A relaxation algorithm is developed to match the input pattern with every FARG. For decision making, a similarity measure is established via statistical technique to calculate the matching degree between the input pattern and referenced FARG, according to which the recognition result is determined. The principle of our method makes it very robust against stroke connection and stroke order variation as well as stroke shape deformation. A database of 22530 samples collected from 6 subjects is used to test our recognition system which can recognize 3755 categories of Chinese characters. The result shows that our method is very effective: a top 1 recognition rate of 98.8% and a top 10 of 99.7% are reached.","PeriodicalId":435320,"journal":{"name":"Proceedings of the Fourth International Conference on Document Analysis and Recognition","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1997-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114827209","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1997-08-18DOI: 10.1109/ICDAR.1997.620654
A. Myka, Ulrich Güntzer
The vector-space model offers an easy and robust model for Information Retrieval. Thereby, the similarities between queries and documents as well as the similarities between documents themselves are of importance. Document similarities may be used in order to generate links between documents that lead users from one document to related ones. Studies have shown that the vector-space model is robust in the context of OCR-processing if manually constructed queries are used. However it is not clear whether this model, if used for hypertext construction, is robust with regard to data corruption as caused by OCR engines. In this paper, we describe the performance of automatic hypertext construction, based on the vector-space model, with regard to three different measures: the number of overtakings within the used rankings, the accumulated distance of a document's position within the rankings and a comparison based on recall-precision graphs.
{"title":"Measuring the effects of OCR errors on similarity linking","authors":"A. Myka, Ulrich Güntzer","doi":"10.1109/ICDAR.1997.620654","DOIUrl":"https://doi.org/10.1109/ICDAR.1997.620654","url":null,"abstract":"The vector-space model offers an easy and robust model for Information Retrieval. Thereby, the similarities between queries and documents as well as the similarities between documents themselves are of importance. Document similarities may be used in order to generate links between documents that lead users from one document to related ones. Studies have shown that the vector-space model is robust in the context of OCR-processing if manually constructed queries are used. However it is not clear whether this model, if used for hypertext construction, is robust with regard to data corruption as caused by OCR engines. In this paper, we describe the performance of automatic hypertext construction, based on the vector-space model, with regard to three different measures: the number of overtakings within the used rankings, the accumulated distance of a document's position within the rankings and a comparison based on recall-precision graphs.","PeriodicalId":435320,"journal":{"name":"Proceedings of the Fourth International Conference on Document Analysis and Recognition","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1997-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116898667","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1997-08-18DOI: 10.1109/ICDAR.1997.620566
S. Liao, Q. Lu
Moment descriptors have been developed as features in pattern recognition since the moment method was first introduced. In this paper, new moment features for Chinese character recognition are proposed. These provide significant improvements in terms of Chinese character recognition, especially for those characters that are very close in shapes.
{"title":"A study of moment functions and its use in Chinese character recognition","authors":"S. Liao, Q. Lu","doi":"10.1109/ICDAR.1997.620566","DOIUrl":"https://doi.org/10.1109/ICDAR.1997.620566","url":null,"abstract":"Moment descriptors have been developed as features in pattern recognition since the moment method was first introduced. In this paper, new moment features for Chinese character recognition are proposed. These provide significant improvements in terms of Chinese character recognition, especially for those characters that are very close in shapes.","PeriodicalId":435320,"journal":{"name":"Proceedings of the Fourth International Conference on Document Analysis and Recognition","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1997-08-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116986164","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：