Hi,

Has anyone tried using a digital camera, I'm thinking of PowerShot A5 or PDR-M1, for text and then use OCR software that comes with a scanner to convert the text?

Also, has anyone tried doing the same thing using a digital camcorder? I've been thinking about this ever since the Snappy came out but the analog camcorder simply doesn't do the trick. Thanks for any enlightenment.

I haven't read through all the messages here. So sorry if someone already addressed the question somewhere.

-- David Wu (dwu@vt.edu), November 01, 1998

Answers

I don't think anybody has addressed this yet...

I haven't tried this myself, suspect the biggest problem is going to be getting enough area at a high enough resolution for the programs to work. I'd guess a PDR-M1 at 1.5 megapixels would work with typical OCR programs up to about 1/4 of a page of reasonably large text.

For comparison, most OCR programs need about 200 dpi on the page to work reasonably well with 10-12 point type. If we used the PDR-M1, at 1280x1024, this would translate to about 5x6 inches. You'll also have to watch out for rotation and any geometric distortion from the lens though - most OCR programs are very sensitive to distortions in the straight lines of text...

How about it? Anyone out there actually tried (or willing to try) this? Let us know if you have/do!

-- Dave Etchells (web@imaging-resource.com), November 02, 1998.

Thanks for the tip! I had received a message from Luca who had experience on this. I then threw more questions for him. I hope he won't mind my taking the liberty of posting our exchanges here to both potentially benefit the others and to solicit more answers.

Date: Mon, 02 Nov 1998 13:48:57 +0100 From: Luca To: David Subject: OCR and Digital cameras

the problem is how large is the page you have to get in the OCR. for a A4 size, 21x30 cm, to get a 300Dpi resolution (ok for OCR) you need a 2000x3000 pixel resolution camera. (around 10.000 USD).

I have tried with a digital camera with 1024x768 pixels and it works only in small area. For an A4 page my camera has only 85 dpi. Too little.

A4 page = 30 cm height 30 cm= 30/2.5=12 so 12 inches 1024(pixels) / 12 (inches) = 85.

Minolta has a device to make OCR and image acquisition for museum books, it has a high resolution and can correct even the bending of the page of a book. It cost around 15.000 USD.

Date: Mon, 02 Nov 1998 13:48:57 +0100 From: David To: Luca Subject: OCR and Digital cameras

Thanks so much for the information! Well, I guess, for quite sometime to come, I'll just have to settle for the old way: Photocopy what I need in the library and scan them for OCR if absolutely necessary. Otherwise, photocopying-handtyping is still the most practical way.

What type of OCR are you using? Is it good enough to pay for (vs. the free limited edition of OCR. normally bundled with a scanner)? How fast is it?

Is that Minolta device a camera or a scanner (or as fast as a cam or as slow as a scanner)?

Also, do you happen to know how the digital camcorder's 500-line resolution corresponds to the dpi?

-- David Wu (dwu@vt.edu), November 02, 1998.