We report an improved methodology for training classifiers for document image content extraction, that is, the location and segmentation of regions containing handwriting, machine-printed text, photographs, blank space, etc. Our previous methods classified each individual pixel separately (rather than regions): this avoids the arbitrariness and restrictiveness that result from constraining region shapes (to, e.g., rectangles). However, this policy also allows content classes to vary frequently within small regions, often yielding areas where several content classes are mixed together. This does not reflect the way that real content is organized: typically almost all small local regions are of uniform class. This observation suggested a post-classification methodology which enforces local uniformity without imposing a restricted class of region shapes. We choose features extracted from small local regions (e.g. 4-5 pixels radius) with which we train classifiers that operate on the output of previous classifiers, guided by ground truth. This provides a sequence of post-classifiers, each trained separately on the results of the previous classifier. Experiments on a highly diverse test set of 83 document images show that this method reduces per-pixel classification errors by 23%, and it dramatically increases the occurrence of large contiguous regions of uniform class, thus providing highly usable near-solid 'masks' with which to segment the images into distinct classes. It continues to allow a wide range of complex, non-rectilinear region shapes.
{"title":"Iterated Document Content Classification","authors":"Chang An, H. Baird, Pingping Xiu","doi":"10.1109/ICDAR.2007.148","DOIUrl":"https://doi.org/10.1109/ICDAR.2007.148","url":null,"abstract":"We report an improved methodology for training classifiers for document image content extraction, that is, the location and segmentation of regions containing handwriting, machine-printed text, photographs, blank space, etc. Our previous methods classified each individual pixel separately (rather than regions): this avoids the arbitrariness and restrictiveness that result from constraining region shapes (to, e.g., rectangles). However, this policy also allows content classes to vary frequently within small regions, often yielding areas where several content classes are mixed together. This does not reflect the way that real content is organized: typically almost all small local regions are of uniform class. This observation suggested a post-classification methodology which enforces local uniformity without imposing a restricted class of region shapes. We choose features extracted from small local regions (e.g. 4-5 pixels radius) with which we train classifiers that operate on the output of previous classifiers, guided by ground truth. This provides a sequence of post-classifiers, each trained separately on the results of the previous classifier. Experiments on a highly diverse test set of 83 document images show that this method reduces per-pixel classification errors by 23%, and it dramatically increases the occurrence of large contiguous regions of uniform class, thus providing highly usable near-solid 'masks' with which to segment the images into distinct classes. It continues to allow a wide range of complex, non-rectilinear region shapes.","PeriodicalId":279268,"journal":{"name":"Ninth International Conference on Document Analysis and Recognition (ICDAR 2007)","volume":"57 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130404835","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}
Tong-Hua Su, Tian-Wen Zhang, Hu-Jie Huang, Yu Zhou
This paper proposes a skew detection method for real Chinese handwritten documents. After analyzing the characteristics of Chinese characters, it utilizes the horizontal stroke histogram. Its accuracy, ability to increase the recall rate of text line separation, and CPU time consuming are investigated using 853 real Chinese handwritten documents. The results show that: 1) the method can identify 98.83% of the skew angles within one degree, with an improvement of 8.44% than Wigner-Ville distribution (WVD) method; 2) when incorporated into text line separation, the recall rate has an improvement of 2.54% than WVD method; 3) the method only consumes one-twentieth of WVD method on the same test environment.
{"title":"Skew Detection for Chinese Handwriting by Horizontal Stroke Histogram","authors":"Tong-Hua Su, Tian-Wen Zhang, Hu-Jie Huang, Yu Zhou","doi":"10.1109/ICDAR.2007.233","DOIUrl":"https://doi.org/10.1109/ICDAR.2007.233","url":null,"abstract":"This paper proposes a skew detection method for real Chinese handwritten documents. After analyzing the characteristics of Chinese characters, it utilizes the horizontal stroke histogram. Its accuracy, ability to increase the recall rate of text line separation, and CPU time consuming are investigated using 853 real Chinese handwritten documents. The results show that: 1) the method can identify 98.83% of the skew angles within one degree, with an improvement of 8.44% than Wigner-Ville distribution (WVD) method; 2) when incorporated into text line separation, the recall rate has an improvement of 2.54% than WVD method; 3) the method only consumes one-twentieth of WVD method on the same test environment.","PeriodicalId":279268,"journal":{"name":"Ninth International Conference on Document Analysis and Recognition (ICDAR 2007)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117177738","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}
Christoph Ringlstetter, Ulrich Reffle, Annette Gotscharek, K. Schulz
Most systems for correcting errors in texts make use of specific word distance measures such as the Levenshtein distance. In many experiments it has been shown that correction accuracy is improved when using edit weights that depend on the particular symbols of the edit operation. However, most proposed approaches so far rely on high amounts of training data where errors and their corrections are collected. In practice, the preparation of suitable ground truth data is often too costly, which means that uniform edit costs are used. In this paper we evaluate approaches for deriving symbol dependent edit weights that do not need any ground truth training data, comparing them with methods based on ground truth training. We suggest a new approach where special error dictionaries are used to estimate weights. The method is simple and very efficient, needing one pass of the document to be corrected. Our experiments with different OCR systems and textual data show that the method consistently improves correction accuracy in a significant way, often leading to results comparable to those achieved with ground truth training.
{"title":"Deriving Symbol Dependent Edit Weights for Text Correction_The Use of Error Dictionaries","authors":"Christoph Ringlstetter, Ulrich Reffle, Annette Gotscharek, K. Schulz","doi":"10.1109/ICDAR.2007.99","DOIUrl":"https://doi.org/10.1109/ICDAR.2007.99","url":null,"abstract":"Most systems for correcting errors in texts make use of specific word distance measures such as the Levenshtein distance. In many experiments it has been shown that correction accuracy is improved when using edit weights that depend on the particular symbols of the edit operation. However, most proposed approaches so far rely on high amounts of training data where errors and their corrections are collected. In practice, the preparation of suitable ground truth data is often too costly, which means that uniform edit costs are used. In this paper we evaluate approaches for deriving symbol dependent edit weights that do not need any ground truth training data, comparing them with methods based on ground truth training. We suggest a new approach where special error dictionaries are used to estimate weights. The method is simple and very efficient, needing one pass of the document to be corrected. Our experiments with different OCR systems and textual data show that the method consistently improves correction accuracy in a significant way, often leading to results comparable to those achieved with ground truth training.","PeriodicalId":279268,"journal":{"name":"Ninth International Conference on Document Analysis and Recognition (ICDAR 2007)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130971510","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}
The effectiveness of kernel fisher discrimination analysis (KFDA) has been demonstrated by many pattern recognition applications. However, due to the large size of Gram matrix to be trained, how to use KFDA to solve large vocabulary pattern recognition task such as Chinese Characters recognition is still a challenging problem. In this paper, a two-stage KFDA approach is presented for handwritten Chinese character recognition. In the first stage, a new modified linear discriminant analysis method is developed to get the recognition candidates. In the second stage, KFDA is used to determine the final recognition result. Experiments on 1034 categories of Chinese character from 120 sets of handwriting samples shows that a 3.37% improvement of recognition rate is obtained, which suggests the effectiveness of the proposed method.
{"title":"Handwritten Chinese Character Recognition Using Modified LDA and Kernel FDA","authors":"Duanduan Yang, Lianwen Jin","doi":"10.1109/ICDAR.2007.128","DOIUrl":"https://doi.org/10.1109/ICDAR.2007.128","url":null,"abstract":"The effectiveness of kernel fisher discrimination analysis (KFDA) has been demonstrated by many pattern recognition applications. However, due to the large size of Gram matrix to be trained, how to use KFDA to solve large vocabulary pattern recognition task such as Chinese Characters recognition is still a challenging problem. In this paper, a two-stage KFDA approach is presented for handwritten Chinese character recognition. In the first stage, a new modified linear discriminant analysis method is developed to get the recognition candidates. In the second stage, KFDA is used to determine the final recognition result. Experiments on 1034 categories of Chinese character from 120 sets of handwriting samples shows that a 3.37% improvement of recognition rate is obtained, which suggests the effectiveness of the proposed method.","PeriodicalId":279268,"journal":{"name":"Ninth International Conference on Document Analysis and Recognition (ICDAR 2007)","volume":"55 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132559338","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}
In this paper we present a method to improve the performance of individual page segmentation engines based on the combination of the output of several engines. The rules of combination are designed after analyzing the results of each individual method. This analysis is performed using a performance evaluation framework that aims at characterizing each method according to its strengths and weaknesses rather than computing a single performance measure telling which is the "best" segmentation method.
{"title":"Combination of OCR Engines for Page Segmentation Based on Performance Evaluation","authors":"Miquel A. Ferrer, Ernest Valveny","doi":"10.1109/ICDAR.2007.83","DOIUrl":"https://doi.org/10.1109/ICDAR.2007.83","url":null,"abstract":"In this paper we present a method to improve the performance of individual page segmentation engines based on the combination of the output of several engines. The rules of combination are designed after analyzing the results of each individual method. This analysis is performed using a performance evaluation framework that aims at characterizing each method according to its strengths and weaknesses rather than computing a single performance measure telling which is the \"best\" segmentation method.","PeriodicalId":279268,"journal":{"name":"Ninth International Conference on Document Analysis and Recognition (ICDAR 2007)","volume":"49 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132733743","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}
India is a multilingual multiscript country with more than 18 languages and 10 different major scripts. Not enough research work towards recognition of handwritten characters of these Indian scripts has been done. Tamil, an official as well as popular script of the southern part of India, Singapore, Malaysia, and Sri Lanka has a large character set which includes many compound characters. Only a few works towards handwriting recognition of this large character set has been reported in the literature. Recently, HP Labs India developed a database of handwritten Tamil characters. In the present paper, we describe an off-line recognition approach based on this database. The proposed method consists of two stages. In the first stage, we apply an unsupervised clustering method to create a smaller number of groups of handwritten Tamil character classes. In the second stage, we consider a supervised classification technique in each of these smaller groups for final recognition. The features considered in the two stages are different. The proposed two-stage recognition scheme provided acceptable classification accuracies on both the training and test sets of the present database.
{"title":"A Two Stage Recognition Scheme for Handwritten Tamil Characters","authors":"U. Bhattacharya, S. Ghosh, S. K. Parui","doi":"10.1109/ICDAR.2007.37","DOIUrl":"https://doi.org/10.1109/ICDAR.2007.37","url":null,"abstract":"India is a multilingual multiscript country with more than 18 languages and 10 different major scripts. Not enough research work towards recognition of handwritten characters of these Indian scripts has been done. Tamil, an official as well as popular script of the southern part of India, Singapore, Malaysia, and Sri Lanka has a large character set which includes many compound characters. Only a few works towards handwriting recognition of this large character set has been reported in the literature. Recently, HP Labs India developed a database of handwritten Tamil characters. In the present paper, we describe an off-line recognition approach based on this database. The proposed method consists of two stages. In the first stage, we apply an unsupervised clustering method to create a smaller number of groups of handwritten Tamil character classes. In the second stage, we consider a supervised classification technique in each of these smaller groups for final recognition. The features considered in the two stages are different. The proposed two-stage recognition scheme provided acceptable classification accuracies on both the training and test sets of the present database.","PeriodicalId":279268,"journal":{"name":"Ninth International Conference on Document Analysis and Recognition (ICDAR 2007)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131686054","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}
A novel approach for the Arabic handwriting recognition is presented. The use of a planar hidden Markov model (PHMM) has permitted to split the Arabic script into five homogeneous horizontal regions. Each region was described by a 1D-HMM. This modeling is based on different levels of segmentation: horizontal, natural and vertical. Both holistic and analytical approaches have been tested for the description of the median band of the Arabic writing. We show finally that a hybrid approach conducted to the improvement of the whole system performances.
{"title":"A hybrid approach for off-line Arabic handwriting recognition based on a Planar Hidden Markov modeling","authors":"Sameh Masmoudi Touj, N. Amara, H. Amiri","doi":"10.1109/ICDAR.2007.14","DOIUrl":"https://doi.org/10.1109/ICDAR.2007.14","url":null,"abstract":"A novel approach for the Arabic handwriting recognition is presented. The use of a planar hidden Markov model (PHMM) has permitted to split the Arabic script into five homogeneous horizontal regions. Each region was described by a 1D-HMM. This modeling is based on different levels of segmentation: horizontal, natural and vertical. Both holistic and analytical approaches have been tested for the description of the median band of the Arabic writing. We show finally that a hybrid approach conducted to the improvement of the whole system performances.","PeriodicalId":279268,"journal":{"name":"Ninth International Conference on Document Analysis and Recognition (ICDAR 2007)","volume":"56 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134096675","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}
Hidden Markov models (HMM) have long been a popular choice for Western cursive handwriting recognition following their success in speech recognition. Even for the recognition of Oriental scripts such as Chinese, Japanese and Korean, hidden Markov models are increasingly being used to model substrokes of characters. However, when it comes to Indie script recognition, the published work employing HMMs is limited, and generally focussed on isolated character recognition. In this effort, a data-driven HMM-based online handwritten word recognition system for Tamil, an Indie script, is proposed. The accuracies obtained ranged from 98% to 92.2% with different lexicon sizes (IK to 20 K words). These initial results are promising and warrant further research in this direction. The results are also encouraging to explore possibilities for adopting the approach to other Indie scripts as well.
{"title":"Hidden Markov Models for Online Handwritten Tamil Word Recognition","authors":"A. Bharath, S. Madhvanath","doi":"10.1109/ICDAR.2007.131","DOIUrl":"https://doi.org/10.1109/ICDAR.2007.131","url":null,"abstract":"Hidden Markov models (HMM) have long been a popular choice for Western cursive handwriting recognition following their success in speech recognition. Even for the recognition of Oriental scripts such as Chinese, Japanese and Korean, hidden Markov models are increasingly being used to model substrokes of characters. However, when it comes to Indie script recognition, the published work employing HMMs is limited, and generally focussed on isolated character recognition. In this effort, a data-driven HMM-based online handwritten word recognition system for Tamil, an Indie script, is proposed. The accuracies obtained ranged from 98% to 92.2% with different lexicon sizes (IK to 20 K words). These initial results are promising and warrant further research in this direction. The results are also encouraging to explore possibilities for adopting the approach to other Indie scripts as well.","PeriodicalId":279268,"journal":{"name":"Ninth International Conference on Document Analysis and Recognition (ICDAR 2007)","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133060010","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}
In this paper we present Patra - an integrated document architecture which incorporates handwritten illustrations captured and rendered in a temporal fashion synchronized with audio, video, text, and image data. The architecture of Patra permits non-linear growth in the form of multiple hierarchically organized play streams. Semantic metadata is also an integral part of Patra which serves a useful purpose of organizing such documents in a collection. We have developed an email application in which the users are provided with an authoring and rendering environment to compose, view, and reply to messages in the form of Patra.
{"title":"Pàtrà: A Novel Document Architecture for Integrating Handwriting with Audio-Visual Information","authors":"Gaurav Harit, V. Mankar, S. Chaudhury","doi":"10.1109/ICDAR.2007.204","DOIUrl":"https://doi.org/10.1109/ICDAR.2007.204","url":null,"abstract":"In this paper we present Patra - an integrated document architecture which incorporates handwritten illustrations captured and rendered in a temporal fashion synchronized with audio, video, text, and image data. The architecture of Patra permits non-linear growth in the form of multiple hierarchically organized play streams. Semantic metadata is also an integral part of Patra which serves a useful purpose of organizing such documents in a collection. We have developed an email application in which the users are provided with an authoring and rendering environment to compose, view, and reply to messages in the form of Patra.","PeriodicalId":279268,"journal":{"name":"Ninth International Conference on Document Analysis and Recognition (ICDAR 2007)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132187379","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}
Wafa Boussellaa, Abderrazak Zahour, B. Taconet, A. Alimi, A. BenAbdelhafid
This paper presents the new system PRAAD for preprocessing and analysis of Arabic historical documents. It is composed of two important parts: pre-processing and analysis of ancient documents. After digitization, the color or greyscale ancient documents images are distorted by the presence of strong background artefacts such as scan optical blur and noise, show-through and bleed-through effects and spots. In order to preserve and exploit this cultural heritage documents, we intend to create efficient tool that achieves restoration, binarisation, and analyses the document layout. The developed tool is done by adapting our expertise in document image processing of Arabic ancient documents, printed or manuscripts. The different functions of PRAAD system are tested on a set of Arabic ancient documents from the national library and the National Archives of Tunisia.
{"title":"PRAAD: Preprocessing and Analysis Tool for Arabic Ancient Documents","authors":"Wafa Boussellaa, Abderrazak Zahour, B. Taconet, A. Alimi, A. BenAbdelhafid","doi":"10.1109/ICDAR.2007.209","DOIUrl":"https://doi.org/10.1109/ICDAR.2007.209","url":null,"abstract":"This paper presents the new system PRAAD for preprocessing and analysis of Arabic historical documents. It is composed of two important parts: pre-processing and analysis of ancient documents. After digitization, the color or greyscale ancient documents images are distorted by the presence of strong background artefacts such as scan optical blur and noise, show-through and bleed-through effects and spots. In order to preserve and exploit this cultural heritage documents, we intend to create efficient tool that achieves restoration, binarisation, and analyses the document layout. The developed tool is done by adapting our expertise in document image processing of Arabic ancient documents, printed or manuscripts. The different functions of PRAAD system are tested on a set of Arabic ancient documents from the national library and the National Archives of Tunisia.","PeriodicalId":279268,"journal":{"name":"Ninth International Conference on Document Analysis and Recognition (ICDAR 2007)","volume":"10 2","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2007-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114038748","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：