2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR)最新文献_第2页

Experiment study on utilizing convolutional neural networks to recognize historical Arabic handwritten text 利用卷积神经网络识别阿拉伯历史手写体文本的实验研究

2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR) Pub Date : 2017-04-01 DOI: 10.1109/ASAR.2017.8067773

Reem Alaasam, Berat Kurar Barakat, M. Kassis, Jihad El-Sana

{"title":"Experiment study on utilizing convolutional neural networks to recognize historical Arabic handwritten text","authors":"Reem Alaasam, Berat Kurar Barakat, M. Kassis, Jihad El-Sana","doi":"10.1109/ASAR.2017.8067773","DOIUrl":"https://doi.org/10.1109/ASAR.2017.8067773","url":null,"abstract":"Deep learning is a form of hierarchical learning, it consists of multiple layers of representations that gradually transform data into high level concepts. Deep learning has been providing the state of the art results for various computer vision problems. However, a typical deep leaning algorithm needs a large amount of data to train a deep model and guarantee the models ability to generalize. It is not easy to generate large labeled datasets and it is one of the main barriers to apply deep learning for many problems. Data augmentation schemes were introduced to overcome this limitation, by extending small available labeled datasets. In this work we experiment with extending a small labeled dataset of Arabic continuous subwords by an orders of magnitude. The labeled dataset, which consist of handwritten Arabic subwords is used to synthesize a large collection of labeled dataset. The synthesized subwords are based on one or multiple writing styles from the original labeled dataset. We also experiment with generating various printed forms of subwords. We include only Naskh font, as most of the Arabic historical manuscripts were written in this type of font. We train several convolutional neural networks using handwritten, printed and synthesized datasets and obtain encouraging results.","PeriodicalId":224611,"journal":{"name":"2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131753343","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}

引用次数: 13

Decision trees based on perceptual codes for on-line Arabic character recognition 基于感知码的决策树在线阿拉伯字符识别

2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR) Pub Date : 2017-04-01 DOI: 10.1109/ASAR.2017.8067778

Hanen Akouaydi, Slim Abdelhedi, S. Njah, M. Zaied, A. Alimi

引用次数: 14

Ligature-based font size independent OCR for Noori Nastalique writing style Noori Nastalique书写风格的基于ligature的字体大小独立OCR

2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR) Pub Date : 2017-04-01 DOI: 10.1109/ASAR.2017.8067774

Q. Akram, S. Hussain

{"title":"Ligature-based font size independent OCR for Noori Nastalique writing style","authors":"Q. Akram, S. Hussain","doi":"10.1109/ASAR.2017.8067774","DOIUrl":"https://doi.org/10.1109/ASAR.2017.8067774","url":null,"abstract":"In this paper, a font size independent Optical Character Recognition (OCR) system for Urdu document images is presented. Urdu documents are written using Noori Nastalique writing style with different font sizes of normal text and headings. Most of current state of the art techniques of Urdu OCRs support recognition of text having single font size. The presented study deals with the recognition of Nastalique text having 14 to 28 font sizes. Three recognizers at three font sizes(called pivot) including 14, 16 and 22 are developed. Urdu document images having remaining font sizes such as 18, 20, 24, 26 and 28 are resized to the nearest pivot font size using Nearest Neighboring interpolation technique so that it can be recognized. The detailed analysis has been carried out to compute optimal scaling factor of each font size to improve recognition results. It has been observed that recognizers perform better at resized images by applying optimal scaling factors instead of simple computed scaling factors. The system is developed and matured on 1,965 main body classes covering 59,974 high frequent Urdu words. After maturation, system has 97.20%, 97.08%, 95.13%, 95.65%, 96.26%, 96.52%, 95.78%, 96.38%, 96.66% main body recognition accuracy for 14, 16, 18, 20, 24, 26, 28 font sizes respectively.","PeriodicalId":224611,"journal":{"name":"2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114490517","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}

引用次数: 10

Enabling indexing and retrieval of historical Arabic manuscripts through template matching based word spotting 使索引和检索历史阿拉伯手稿通过模板匹配基于词定位

2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR) Pub Date : 2017-04-01 DOI: 10.1109/ASAR.2017.8067760

Tayyeba Faisal, S. Al-Maadeed

{"title":"Enabling indexing and retrieval of historical Arabic manuscripts through template matching based word spotting","authors":"Tayyeba Faisal, S. Al-Maadeed","doi":"10.1109/ASAR.2017.8067760","DOIUrl":"https://doi.org/10.1109/ASAR.2017.8067760","url":null,"abstract":"We present a holistic segmentation-free query by example word spotting technique based on template matching. We have applied this technique to a dataset of historical Arabic handwritten manuscript images. First, the documents as well as query word images are pre-processed for separating text from the noisy background and converting to their binary equivalents. Then a pixel based approach is used for computing the similarity between the pre-processed template query word and document images by using the Correlation similarity measure. Slight variations in font sizes are tolerated by adjusting the threshold of similarity. Our robust pre-processing algorithm significantly enhances the performance of the learning-free template matching based word spotting approach. The proposed technique is simple as well as efficient as it does not involve any time-consuming learning steps. Experiments with a historical Arabic dataset yield promising results. This technique can generate locations of occurrences of query word images which is the fundamental step towards building searchable indexes for historical manuscripts.","PeriodicalId":224611,"journal":{"name":"2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129102395","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}

引用次数: 6

WAHD: A database for writer identification of Arabic historical documents WAHD:阿拉伯历史文献作者识别数据库

2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR) Pub Date : 2017-04-01 DOI: 10.1109/ASAR.2017.8067761

Alaa Abdalhaleem, Ahmed Droby, Abedelkadir Asi, M. Kassis, Reem Al Asam, Jihad El-Sana

{"title":"WAHD: A database for writer identification of Arabic historical documents","authors":"Alaa Abdalhaleem, Ahmed Droby, Abedelkadir Asi, M. Kassis, Reem Al Asam, Jihad El-Sana","doi":"10.1109/ASAR.2017.8067761","DOIUrl":"https://doi.org/10.1109/ASAR.2017.8067761","url":null,"abstract":"A comprehensive Arabic handwritten text database is an important resource for Arabic handwritten text recognition research. It is essential for training text recognition algorithms and vital for evaluating the performance of these algorithms. In this paper, we present a database that includes manuscripts from the Islamic heritage project (IHP), consisting of 333 historical manuscripts written by 302 different writers, 23 from them are known. The database contains 54 manuscripts, whose writers are known, from 13 sources. Among these known writers, 11 have written multiple manuscripts. The total number of pages in the entire database is 36,969. Each manuscript in the database accompanied with metadata that include various properties of the manuscript, such as title, creator, subject, language, copyist name, etc. To enrich our database we added twenty historical books scanned from the National Library(NLJ), in Jerusalem. The books have different number of pages and different writing styles. In addition, we present a number of experimental results on the database using two classifiers, The GMMS System and The OBI/SIFT System. The database is made freely available to researchers worldwide for research in various handwritten related problems such as text recognition, writer identification, verification, forms analysis, pre-processing and segmentation.","PeriodicalId":224611,"journal":{"name":"2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR)","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115751604","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}

引用次数: 10

Off line writer identification for Arabic language: Analysis and classification techniques using subwords features 阿拉伯语脱机写作者识别:使用子词特征的分析和分类技术

2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR) Pub Date : 2017-04-01 DOI: 10.1109/ASAR.2017.8067777

Makki Maliki, Naseer Al-Jawad, S. Jassim

{"title":"Off line writer identification for Arabic language: Analysis and classification techniques using subwords features","authors":"Makki Maliki, Naseer Al-Jawad, S. Jassim","doi":"10.1109/ASAR.2017.8067777","DOIUrl":"https://doi.org/10.1109/ASAR.2017.8067777","url":null,"abstract":"Handwritten text in any language is believed to convey a great deal of information about writers' personality and identity. Handwritten documents are frequently used as evidences in forensic tasks. Handwriting skills is learnt and developed from the early schooling stages. Research interest in behavioral biometrics was the main driving force behind the growth in research into Writer Identification (WI) from handwritten text, but recent rise in terrorism associated with extreme religious ideologies spreading primarily, but not exclusively, from the middle-east has led to a surge of interest in WI from handwritten text in Arabic and similar languages. This paper is the main outcome of extensive research investigations conducted with the aim of developing an automatic identification of a person from handwritten Arabic text samples. For the dependent 8-dimensional WI scheme, we identify the best performing set of subwords (best 22 subwords out of 49 then followed by best 11 out of these 22 subwords). We established the validity of our hypothesis for different versions of subwords based Writer identification WI schemes by providing empirical evidence when testing on a number of existing text-dependent. The database consisted of 20 text samples from 95 writers. In these experiments we used two different similarity functions, one where we considered the Euclidian distance function while the other one combines the Dynamic Time Warping (DTW) for the projection attributes with Euclidian for the other features. The experimental results confirmed beyond any doubt the validity of our hypothesis and outperformed existing word-base WI, which only achieves high accuracy at rank 10 nearest neighbours. The 11 subwords based schemes with the DTW related similarity function achieved the higher accuracy in top rank around 98.12% accuracy rate of WI.","PeriodicalId":224611,"journal":{"name":"2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132457703","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}

引用次数: 4

Arabic manuscript author verification using deep convolutional networks 使用深度卷积网络的阿拉伯手稿作者验证

2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR) Pub Date : 2017-04-01 DOI: 10.1109/ASAR.2017.8067750

A. Boiarov, A. Senov, A. Knysh

引用次数: 2

Segmentation of offline handwritten Arabic text 离线手写阿拉伯语文本的分割

2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR) Pub Date : 2017-04-01 DOI: 10.1109/ASAR.2017.8067757

Hashem Ghaleb, P. Nagabhushan, U. Pal

{"title":"Segmentation of offline handwritten Arabic text","authors":"Hashem Ghaleb, P. Nagabhushan, U. Pal","doi":"10.1109/ASAR.2017.8067757","DOIUrl":"https://doi.org/10.1109/ASAR.2017.8067757","url":null,"abstract":"Arabic script is cursive in both printed and handwritten forms. This intrinsic nature of cursiveness renders the segmentation task challenging. An Arabic word generally consists of multiple parts known as Parts of Arabic Words (PAWs) or simply sub-words. Sub-words share the same vertical space quite frequently which makes vertical projection segmentation technique inefficient. Several Arabic letters have annexed parts (diacritics) which are located above or below the main parts of the character. The relative positions of the annexed parts and main parts vary a lot in handwritten text. In this paper the task of segmenting offline handwritten Arabic text up to character level is taken up. Firstly, graph-theoretic modeling is utilized to extract connected components of word image. These components are subjected to a thorough analysis to facilitate the segmentation of input image into sub-words. In the sequel diacritics are removed. Then, large number of candidate segmentation points is identified based on two strategies that utilize stroke thickness as a heuristic. Final segmentation points are obtained using a set of rules on the candidate segmentation points. Finally, each sub-word is segmented and diacritics are brought back to their respective segments taking into account the issue of diacritics displacement. Experimentation is conducted on a set of handwritten images of Arabic text drawn from IFN/ENIT dataset. The results obtained are encouraging.","PeriodicalId":224611,"journal":{"name":"2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123885154","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}

引用次数: 8

Arabic/Latin and handwritten/machine-printed formula classification and recognition 阿拉伯/拉丁和手写/机印公式分类和识别

2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR) Pub Date : 2017-04-01 DOI: 10.1109/ASAR.2017.8067766

Kawther Khazri Ayeb, A. Kacem, A. Belaïd

{"title":"Arabic/Latin and handwritten/machine-printed formula classification and recognition","authors":"Kawther Khazri Ayeb, A. Kacem, A. Belaïd","doi":"10.1109/ASAR.2017.8067766","DOIUrl":"https://doi.org/10.1109/ASAR.2017.8067766","url":null,"abstract":"In this paper we mainly introduce a method for mathematical formula script and type identification based on handcrafted features. Arabic/Latin scripts are discriminated by detecting specific symbols based on their pixel density and distribution. Once the formula script identified, we proposed to separate between machine-printed and handwritten formulas. For that, we analyzed the spaces between some specific symbols and their neighbors and the pixel density of some other symbols presenting discriminative differences in their form when they are handwritten or machine-printed. The use of such structural characteristics makes the complexity of the method considerably lower than learning based methods. To finally recognize formulas, two main steps are followed: symbol recognition and formula structure analysis. For the first step, we used a combination of statistical features and a Bayes Network classifier. For the second step, we proceeded by top-down and bottom-up parsing scheme based on operator dominance. A set of replacement rules is defined by a coordinate grammar. Formula parsing consists of applying, from the dominant operator and its context, the appropriate rule to divide the formula into sub-formulas which will be recursively analyzed by the same way. Carried experiments on various mathematical formulas, show the efficiency of both script and type identification and recognition proposed methods.","PeriodicalId":224611,"journal":{"name":"2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR)","volume":"359 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123217919","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}

引用次数: 1

Boosting feature based classifiers for writer identification 增强基于特征的分类器，用于编写器识别

2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR) Pub Date : 2017-04-01 DOI: 10.1109/ASAR.2017.8067768

Raid Saabni

引用次数: 2