Hello All,
At a meet in HBSCE, before the meet, in the canteen, Krish was explaining to me his project about converting books into talking ones by scanning the pages and converting them into text and text to speech. He mentioned about the difficulty in getting a low cost hand-held scanner. At that time I thought of a an idea of using a megapixel camera/camera phone instead of a scanner in order to speed up the scanning process. When we met Dr. Nagarjun, we mentioned this to him too and he found it feasible.
A few days ago a client of mine asked me to checkout the 'MotoMing' cell phone as it had a business card reader along with some other good features. Today as I visited the websites and saw a flash demo, I was pleasantly surprised to know that it runs on linux and they use the same technique of clicking photos of the cards and extracting the text through OCR.
So Krish can try out this method for his project too and this also gives me a positive feeling that the bar code based software transfer technique through a mp camera should be feasible in the sign posts for VI people idea that was discussed some time back. We only need to create our own algorithm for the OCR.
Regards,
Rony. Send instant messages to your online friends http://uk.messenger.yahoo.com
On 10/17/06, Rony ronbillypop@yahoo.co.uk wrote:
So Krish can try out this method for his project too and this also gives me a positive feeling that the bar code based software transfer technique through a mp camera should be feasible in the sign posts for VI people idea that was discussed some time back. We only need to create our own algorithm for the OCR.
OCR and text-to speech software is hard to get right. Also Why re-invent the wheel? Contribute and add modules to opensource OCR software such as Tessaract OCR[1] (which google uses) or GOCR [2] (see in action[3]). Integrate with low-cost camera and profit.
Better yet make a camera which has OCR and text-to-speech such as festival[4]. You could build a organisation around this. The components are already these. Someone needs to customise them and bundle them together. (This is harder than it sounds). Also costs of digicams is coming down drastically.
-- Vinayak
References: [1]. http://sourceforge.net/projects/tesseract-ocr [2]. http://jocr.sourceforge.net/ [3]. http://blog.eberly.org/2006/10/12/worlds-worst-use-of-a-jpeg/ [4]. http://www.cstr.ed.ac.uk/projects/festival/
if cameras can do the kind of scanning needed for ocr and if those gajets are using gnu/linux, then nothing like it. we can use some good ocr softwares from the foss products, and festival will just fill in the place for TTS system. at least for time being, festival will just do right. all I need to know is what exactly rony has found out from his site surfing. it is a great news for this project. thanks rony for the info and vinayak for suggesting exactly what I was thinking. that more than confirms my idea. thanks again. Krishnakant.
krishnakant Mane wrote:
all I need to know is what exactly rony has found out from his site surfing.
This is the link for the flash demo. http://in.motorola.com/motoming/index.asp
In order to scan a business card (visiting card) the camera clicks its photo and converts image to text. It does away with contact scanning.
Regards,
Rony.
___________________________________________________________ Copy addresses and emails from any email account to Yahoo! Mail - quick, easy and free. http://uk.docs.yahoo.com/trueswitch2.html
-
Dinesh Joshi -
krishnakant Mane -
Rony -
Vinayak Hegde