Prashant, Nice work, and if the paper is not yet finished, then I would like to add some other things.
First, Around 7 years back, I had attended a conference organized by Ministry of Technology where IBM Research Lab had shown their Davanagari OCR, it was quite good at that time, and I hope the project would still be on, but I don't have any contact information. Secondly, I would not say that 70 percent accuracy is good, as the English OCRs that we use usually have more than 99 percent accuracy with varying fonts. 70 Percent accuracy means that if a page has 3000 characters, 900 are wrong. Third, Font dependent recognition is also not practical, as it is almost not feasible for users to train them, it may be only good in mass recognition situations. Regards Dinesh Kaushal blog at dineshkaushal.blogspot.com -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Prashant Naik Sent: Sunday, August 03, 2008 8:44 PM To: [email protected] Subject: Re: [AI] Fw: White Paper: OCR Softwares for Indian languages i have tested the link it is working fine. again i am giving it below. http://www.malayalamresourcecentre.org/Mrc/products/nayana.html prashant On 8/3/08, Krishnan <[EMAIL PROTECTED]> wrote: > > hai friends,,it is a good presentation the malayalam ocr software is > not available by that link. > bye, > krishnan.----- Original Message ----- > From: "Prashant Naik" <[EMAIL PROTECTED]> > To: <[email protected]> > Sent: Sunday, August 03, 2008 1:12 PM > Subject: [AI] White Paper: OCR Softwares for Indian languages > > > > Dear Access India Members, > > > > > > > > During the Daisy Forum of India meeting held in Mumbai on 11th and > > 12th April 2008, I was given the responsibility to find information > > on the status of OCR Softwares for Indian languages. So here I am > > presenting the findings that I am able to research. I have prepared > > a White Paper on it which I posted in the PDF format on the daisy > > forum of India's mailing list 3 > days > > back. But for benefit and awareness of others I am pasting content of > > it > > below this message. This will also help those who had posted queries > > on A I regarding this. > > > > > > > > White Paper: OCR Softwares for Indian languages > > > > Date: July 31st, 2008 > > > > Introduction : > > > > OCR softwares are available for English and other foreign languages > > but what is > > > > the status of OCR software availability for Indian languages? > > > > During the Daisy Forum of India meeting held in Mumbai on 11th and > > 12th April > > > > 2008, I was given the responsibility to find information on this. So > > here I am > > > > presenting the findings that I am able to research. > > > > > > > > Definitions : > > > > OCR: - Optical character recognition, usually abbreviated to OCR, is > > the > > > > mechanical or electronic translation of images of handwritten, > typewritten > > or > > > > printed text (usually captured by a scanner) into machine-editable text. > > > > OCR Software: - OCR Software converts paper documents into > > electronic data, > > > > so that you can handle the information (electronic text) in your > > computer system. > > > > Indian Languages: - Indian Constitution recognizes Hindi in > > Devanagari script > > > > as the official language of the central government India the > > Constitution of India > > > > recognizes 22 languages, spoken in different parts of the country, > > > > {All definitions source is "Wikipedia") > > > > > > > > Findings : > > > > As per the research on the web highlighted one workshop / seminar > > organized by > > > > Rediff Centre for Indian Language Content Management > > > > On the theme of "Brainstorming Workshop on OCR for Indian Languages" > > on > > > > 16-17 March, 2007, at Hotel Regalis, Mysore. > > > > Reference. Link: http://www.isim.ac.in/RCILCM/index.htm > > > > Further research on Access India (mailing group for the blind) > > querying more on > > > > this and contact with NAB Karnataka to get more info on this theme > > did > not > > throw > > > > up anything significant. > > > > > > > > Visit by Mr. Venki, rediff.com Technical Head : > > > > During a meeting with Mr. Venki at the XRCVC in the month of June > > 2008, > > > > Some more information about the conference was secured. This was > > because > > > > Mr. Venki himself was a one of the members of the organizing team > > from rediff. > > > > He made the following observation. "Overall the conference was good. > > Speakers > > > > had shared new ideas on developing Indian OCR." > > > > However further following up with regard to this conference it seems > > no > > > > significant progress have been made thereafter. > > > > > > > > Chennai "Print Access" Seminar Findings" > > > > Our XRCVC team member Neha learned about many technological > > > > developments from the "Print Access" conference which was held at > > Chennai on > > > > April 19th, 2008. She shared lot of information, contacts and links. > > > > E.g. Acharya website (http://acharya.iitm.ac.in) TTS translator in > > 22 languages, > > > > Ravi TTS for Telgu, C-DAC softwares like Mantra, Shruti Drishti, > > Shrut Lekhan > > > > and very important lead on Indian OCR software developed by C-DAC Pune. > > > > > > > > Visit to C-DAC Pune : > > > > On May 14th and 15th, the XRCVC team visited C-DAC Pune. The visit > > was very > > > > fruitful. A fully developed off-the-shelf product for Hindi-Devnagri > > Indian > > > > language software named as CHITRANKAN developed by GIST Development > > > > Team, C-DAC, Pune, Maharashtra. They demonstrated the product. The > > result > > > > was very good. CHITRANKAN is commercially used by 2-3 organizations > > in > > > > Pune. > > > > Other C-DAC resources : > > > > OCR softwares in Hindi called CHITRANKAN, in Marathi called > > > > CHITRAKSHARIKA and in Malayalam called NAYANA. > > > > > > > > About NAYANA : > > > > Source: > > http://www.malayalamresourcecentre.org/Mrc/products/nayana.html > > > > NAYANA is a product that enables the user to convert printed > > Malayalam > > > > documents to editable computer files. This system is very simple to > > use and > > > > requires no prior expertise. > > > > FEATURES > > > > - NAYANA processes all types of printed Malayalam Documents. > > > > - Supports TIFF and BMP image formats. > > > > - Supports document Images with resolution 300 dpi and above. > > > > - Detection and correction of document skew of -5o to +5o. > > > > - The output document can be stored in both ISCII and ISFOC form. > > > > - The output document can be saved as TXT, RTF, HTML or ACI file formats. > > > > - User friendly interface. > > > > - Recognition speed of 50 char /sec. > > > > - Conversion of printed documents to editable text. > > > > - Optical Character Recognition combined with TextToSpeech > > technology can > > > > be used for text reading system. > > > > EXPANDABILITY > > > > - A layout analyzer can be added to the system to reproduce the > > input document > > > > in its original layout. > > > > - Can be expanded to cater to hand writ ten and old document. > > > > The linguistic resource generation tools such as Prabandhika and > > Vishleshika > > > > Source: http://delnet.nic.in/news-naclin-report.htm > > > > > > > > About CHITRANKAN : > > > > Source: http://www.cdac.in/html/gist/products/chitra.asp > > > > CHITRANKAN - the first OCR (Optical Character Recognition) system > > for Indian > > > > Languages. > > > > The OCR process involves: > > > > Conversion of printed matter into an electronic image - the > > printed matter can > > > > be converted into an image using Scanner or a Digital Camera > > > > Electronic Image Processing - this involves identifying text > information > > by > > > > analyzing the image for noise and skew. Once text information is > available > > > > another algorithm reads and recognizes the printed matter > > > > Storing the extracted text information as an electronic data: the > > recognized > > > > input is converted to a standard format, which can be opened in any > > word > > > > processing application, facilitating the user to edit the text data. > > > > Chitrankan archives Indian Language content in electronic form > > through OCR. > > It > > > > enables the user to take a book, magazine or printed text in an > > Indian Language, > > > > feed it directly into an electronic computer file, and edit the file > using > > a > > word > > > > processor. Once the data is in the form of electronic text it can be > > searched, > > > > sorted and indexed. > > > > Chitrankan saves the user the effort of typing an entire document. > > > > Chitrankan scans a document to screen by recognizing the text and > > other images > > > > as objects. These scanned images are flawless and can be stored or > printed > > time > > > > and again. > > > > Exceedingly user-friendly with features that can edit, move, resize > > or duplicate > > > > the scanned document, Chitrankan also provides a spell check facility. > > > > The potential of Chitrankan is enormous as it enables users to > > harness > the > > power > > > > of computers to access printed documents in Indian Languages. > > > > Software Advantage: > > > > Recognizes Hindi and Marathi languages along with Embedded English Text. > > > > Skew detection and correction for input image upto ± 15° > > > > Grabs images directly from the scanner for processing > > > > Automatic Text and Picture region detection > > > > Supports all TWAIN compatible scanners and digital cameras > > > > Supports 256 grayscale/color, .bmp/.tiff images scanned at 300 dpi > > as input > > > > image for recognition > > > > Ideal for font sizes between 10 pt. and 36 pt, and all popular fonts. > > > > Saves scanned/modified images as .BMP files > > > > Saves recognized text in ISCII format or exporting as .RTF for > > editing using > > > > GIST range of software > > > > Uses advanced DSP (Digital Signal Processing) algorithms to remove > > "Noise" > > > > and "Back Page Reflection" > > > > Enables printing both - the input image as well as the recognized text. > > > > Provided with inbuilt Flip, Rotate and Negate options for Input > > Image > > > > User Advantage: > > > > Allows deletion of associated pictures from the image by using the > > ERASE > > > > option > > > > Provides painting tools to join the breaks in the characters to get > > good results > > > > Allows OCR to be applied on an image rotated by 180° or flipped > > > > Applies OCR to image having text in reverse by using INVERT option > > > > Provides inbuilt spell checking facility > > > > Provides editing tools like cut, copy, paste, find and replace > > options for use on > > > > recognized text > > > > System Requirements: > > > > Minimum Configuration: > > > > Pentium II with 64 MB RAM > > > > Virtual Memory requirement 300 MB (Swap File Space in Hard Disk) > > > > Recommended Configuration: > > > > Pentium III with 128 MB RAM and above > > > > Virtual Memory requirement 400 MB > > > > Operating Systems Supported: > > > > Window NT ver. 4.0, Service Pack 6.0 and above/ Windows 9X and > > above, > > > > Windows 2000 and Windows XP. > > > > Price: CHITRANKAN Single user license for CHITRANKAN Rs. 10,000/- > > > > > > > > Contacts: channel partner list URL - > > http://www.cdac.in/html/gist/ch_part.asp > > > > CHITRANKAN demo can be downloaded from > > > > http://www.cdac.in/html/gist/down/chtri_d.asp > > > > File Size: 45 MB > > > > Experimenting with CHITRANKAN at the XRCVC findings : > > > > At the XRCVC demo of CHITRANKAN was installed and put through tests. > > > > The Rajyasabha website webpage were used for testing Hindi-Devnagri > script > > > > which uses Yogesh font typeset. Its accuracy can be described as > > good > > > > approximately 70%. This can be improved by using font training mode. > > > > Additional documents in Hindi and Marathi were tested. Results from > > those were > > > > fair amount approximately 40% accuracy level. The font training > > module > > > > however can increase the accuracy. > > > > The software supports the Yogesh Hindi font by default. Mare fonts > > can be > > > > added on by training OCR using font recognition module. > > > > Font training modules enable the user to train the software to > > decipher > > > > documents in particular fonts. To make the software even more > > useful, > > > > CHITRANKAN incorporates a set of application program interfaces > > (APIs) which > > > > allow software developers the flexibility to build features from > > CHITRANKAN into > > > > their software application. > > > > You can save recognized output in RTF format and even choose > > recognition > > > > language as either Hindi or Marathi. > > > > > > > > Screen reader access with Chitrankan - > > > > Graphical User Interface of Chitrankan is very friendly with menus > > and shortcuts > > > > are available for all important options. > > > > In the workspace area it has mainly three windows such as input > > image window, > > > > recognized output text window and digitized image windows. However > > screen > > > > reader (SAFA) is not able to read the recognized text. > > > > > > > > Conclusion : > > > > One can definitely contribute to the development of the Indian > > language OCR > > > > through download, testing and the feedback can be given to C-DAC > > that would > > > > help in product enhancement. Those who are familiar with Malayalam > > would do > > > > well to test the NAYANA OCR software. > > > > Prashant Naik > > > > The Xavier's Resource Centre for the Visually Challenged (XRCVC) > > > > St. Xavier's College, Mumbai. > > ---- > > VISION WITHOUT ACTION IS MERELY A DREAM, ACTION WITHOUT VISION JUST > > PASSES THE TIME, VISION WITH ACTION CAN CHANGE THE WORLD. > > Join Access India convention: For updates on it visit: > > http://accessindia.org.in/harish/convention.htm > > Registration is now open! > > > > To unsubscribe send a message to > > [EMAIL PROTECTED] > > with the subject unsubscribe. > > > > To change your subscription to digest mode or make any other > > changes, please visit the list home page at > > > http://accessindia.org.in/mailman/listinfo/accessindia_accessindia.org > .in > > > Join Access India convention: For updates on it visit: > http://accessindia.org.in/harish/convention.htm > Registration is now open! > > To unsubscribe send a message to [EMAIL PROTECTED] the subject unsubscribe. > > To change your subscription to digest mode or make any other changes, > please visit the list home page at > http://accessindia.org.in/mailman/listinfo/accessindia_accessindia.org > .in -- VISION WITHOUT ACTION IS MERELY A DREAM, ACTION WITHOUT VISION JUST PASSES THE TIME, VISION WITH ACTION CAN CHANGE THE WORLD. Join Access India convention: For updates on it visit: http://accessindia.org.in/harish/convention.htm Registration is now open! To unsubscribe send a message to [EMAIL PROTECTED] with the subject unsubscribe. To change your subscription to digest mode or make any other changes, please visit the list home page at http://accessindia.org.in/mailman/listinfo/accessindia_accessindia.org.in Join Access India convention: For updates on it visit: http://accessindia.org.in/harish/convention.htm Registration is now open! To unsubscribe send a message to [EMAIL PROTECTED] with the subject unsubscribe. To change your subscription to digest mode or make any other changes, please visit the list home page at http://accessindia.org.in/mailman/listinfo/accessindia_accessindia.org.in
