OT: Looking for JPG capture software

Google (or Topeka if its still 4/01) 'homebrew book scanner'

--
Paul Hovnanian     mailto:Paul@Hovnanian.com
------------------------------------------------------------------
God doesn't play dice. However, He does play a mean game of
3 card monte.

Before you commit to JPG format (lossy), you might want to do some empirical tests (to verify the desired level of detail is preserved).

Items that I have opted to scan (many thousands of pages!) I "disassembled" into discrete pages and fed them through a document feeder. 100 pages takes about 20 minutes (not bad considering).

You should also verify that the images you obtain are good enough for OCR post-processing (unless youo never want to go that extra step)

You also have to think carefully about whether you preserve color, greyscale, B&W, etc. (if all you have is a few cookbooks, the savings probably won't matter)

My preference is color. And "No", I'm reasonably certain I don't want to do any post- processing. This is just for me to save space. I'm not looking for perfection, just something I can read, and possibly index. Not even sure I will go the extra step of indexing. Depends on the program I end up using, I guess?

thanks for the info.

Will do. Thanks for the tip. I tried a couple other keywords on Topeka, but got jibberish back. I'm sure this has been beat to death, I just can't find it. I should mention my preference is freeware or shareware, but I could shell out if I have to.

-mpm

WinFast PVR does exactly that. It came with my tuner card. Art

Check out Freedom Scientific, they have scanner software for the blind that does what you want... if you have a few grand!

Just jpegs won't cut it if you want to actually read the pages later...

Charlie

I seem to recollect that my camera, when connected to a PC (running Linux) via the USB can be controlled by a mouse click to take a photo.

I understand your pain. I recently scanned a book, page by page, using a camera on a tripod. After each photo, I had to crouch down to turn the page.

After downloading, cropping and thresholding the photos then reducing the resolution, I converted the jpegs into a PDF document using PHP and FPDF.

I used the maximum resolution of the camera, a Canon S5 IS (which doesn't offer a raw format), the minimum Iris aperture and flash - but still one could see some blurring at the peripery of the images.

a raw format), the minimum Iris aperture and

Yeah, those image periperys will get ya every time.

Hahahaha.

Seriously, if you had a small aperture, then the shot/camera is not focused/focusing right. Doh! There is no aberration in the center of the lens.

On a sunny day (Thu, 1 Apr 2010 18:41:58 -0700 (PDT)) it happened mpm wrote in :

I can just use my Cannon A470 digital camera and flash each 2 pages.... Just as good as a scan, but problem is to make someting that keeps the paper flat. If you put glass over it to keep it flat you need to be careful with lighting so as not to get reflections, maybe not use the flash. When using flash you can hold the camera in your hand, no motion artefacts.

ting

s.

I was thinking to use an aperture of f/11 to f/16, good lighting, and a reasonable distance between the camera & book so that everything would be in focus even if the distance to page isn't perfectly flat. I'm a pretty good photographer (if I don't mind saying so myself), so I'm not worried about the imaging setup. If I used glass, it can be lighted at 45-degree angles, etc...

My main concern is mechanically wearing out a digital camera shutter. For example, my Nikon D-90 camera has a shutter lifetime "tested to

100,000" exposures. That's actually not that much. I wouldn't want to burn off 5%-10% (or more) of that copying books.

IDEALLY, I would like to take JPG's (or whatever) of a VIDEO STREAM entering the computer. i.e., image on screen, press spacebar (footswitch would be even better, JPG stored to HD. Repeat. That way, it wouldn't wear out any of my digital cameras. (The D90 can be controlled by software, but I don't want to use it.)

As for the comment regarding being able to read JPG's, if you image appropriately, this shouldn't be a problem. You must have enough pixels, good lighting, enough contrast, and not too much compression. I have another program that can quickly convert on-screen JPG's to text and it has no trouble keeping up. Plus, I've done much testing and can quite easily read JPG's, even though getting old(er) is a bitch and 20/20 is at least ten years ago!

That software (ABBYY Screenshot Reader) would require lots of mouse movement, otherwise, it would be a candidate for this project. BTW: For the money, Screenshot Reader is hard to beat. It just that for this project, I always want the whole page (graphics & images), and therefore don't need the extra mouse steps that A/S/R would involve.

If you used flash, an idea might be to polarise the light and arrange the angle at the Brewster angle in order to reduce reflections. One snag, maybe insignificant, might be that autofocus would focus on the glass rather than the page below (one strives for perfection).

On a sunny day (Fri, 02 Apr 2010 20:14:55 +0100) it happened richard wrote in :

Oh, no problem with my A470, I have chdk on it:

It can be programmed to take a picture every so many seconds, and do a lot more. That is how I made this timelapse of the stars (using a tripod):

You might even find Google's own fully automated super-fast gizmo:

If you have a hi resolution screen, and "see" it on your screen, a simple press of the print screen button, and a paste into paint allows it to be saved under any number of formats. It starts out as a full resolution (to your screen) pixel array.

I have a Samsung Syncmaster 2343bwx, so I get 2048 x 1152 screen caps. So any smaller portion of the screen still remains fairly hi-res.

If you are talking about capture from hard copy, then there is nothing better than a modern 6+ Megapixel camera for the job. Dual page shots allow the job to be done in half the number of shots. Then you have to split them, so you end up taking just as much time as if you had simply turned the page and took another shot.

Anyway, I will post a shot I grabbed off the net, and capped myself. I like capturing standards and symbols.

I will post an EIC logo in abse that I grabbed myself. Most found are all jaggy jpegs. These are hi-res precision symbols.

April fool was yesterday.

Crap. there are no photographic resolution screens on home computers.

rubbish. That's about 2.5MP, the photographic equivelent of a cell phone camera or in the days of film, a polaroid instant picture, not even close to the quality of a $5 disposable box camera.

Let me know when you have a 30,000x20,000 screen.

It's not very high resolution. It's not even good enough for printing a postcard at photographic resolutions. Maybe a wallet picture.

you just don't know any better. You're like a photographer who think's he's hot shit with polaroid.

postcard

You're a goddamned idiot. I know what the resolution level is. and I know what is available. He wanted a jpeg capture, which is NOT high resolution at all, you dumb fucktard! So my suggestion is fine for the purpose.

On a sunny day (Fri, 2 Apr 2010 20:04:16 -0700 (PDT)) it happened JeffM wrote in :

That is clever, one could program one for home use that uses a visible light grid for a moment.

Takes about 10-12s per page with a modern scanner if you concentrate on doing it and have no interruptions. A digicam is faster if you have someone else to turn the pages for you and keep it flat(ish). And if you have a steady hand it can be done handheld at a pinch.

How many pages do you need to do?

Tripod, camera and good light from all sides to avoid shadows near the spine. A video camera even at HD resolution ~2Mp will be a bit marginal for recording A4 books. OCR software will struggle at that.

Otherwise JPEG will be fine provided you use a sensible quality setting. It is a lot better than most people think. If you have coloured line art to photograph you might want to choose 1x1 chroma subsampling if the software offers that option.

Regards, Martin Brown

it

caps.

even

postcard

nothing

I have seen many people dis Polaroid without remembering that the exposed= frame=20 is quite a bit larger than 35 mm. Nor was the actual grain size that = bad. OTOH 2.5 MPixel is still a bit shy of 110 film (about 16 mm).