Review of sub-$100 Doc Camera for Book Scanning w. video

[jman]

Sad to say that three years after starting my scanner project (http://www.diybookscanner.org/forum/vie ... f=15&t=351), it's in a box in another country! I recently looked for low cost doc camera solutions, found one I wanted to try, and so contacted the company to inquire about a review. Happily IPEVO agreed to allow me to review their Ziggi HD document camera—I have no material connection to the company other than being given the product for review (I'm not an affiliate, etc.).

It's not as fun as building a scanner, but it is a cost-effective solution for me. In short, its strengths are mobility (I took into a library with my laptop in a computer bag), ease of use, speed (my max was 1 image/3 sec = 2 pgs/3sec), and cost (just under $100). Its weaknesses are auto-focus (occasionally un-crisp image) and slightly warped text (as with any scan taken from one camera above a book being held open). If you need hi-res images for 98%+ accurate OCR, this may not do the job. It certainly produces readable text, and I would estimate that quality allows for OCR that is about 85% accurate or better.

I've posted more information at my blog, including a HD youtube video that includes a screencast of the product in use and scantailor processing: http://www.joshualmann.com/ziggicamera/. I'm happy to respond to questions or comments here or there.

[victoriaaustralia]

Great write up, lovely video which was a great description of the product.

The recording of video from another device is an interesting illustration of the analogue hole.

Do you have any example pages to upload after scantailor processing?

On this thread I mention my post-processing method with freeware OCR using PDF-Xchange:
http://www.diybookscanner.org/forum/vie ... =19&t=2835

[jman]

I deleted the output files once I created the pdf--I should have saved a few for the demonstration. What I can show you is a pdf I produced. After processing in scantailor I created the pdf using Adobe Acrobat Pro XI and OCR'd using "clearscan" at 300 dpi. It gives the appearance of sharpening the text a bit (a lot more to it than that, as you know), so you can imagine the .tiff images produced by scantailor were a slightly more grainy. Here's a link to a 5 page pdf excerpt (for some reason the OCR went bad on it, but it will have to do): http://www.joshualmann.com/wp-content/u ... xcerpt.pdf

I have something to scan later this week, so I can come back with a few .tiff images from that.

[dpc]

jman,

Do you know if there's some way to fix the OCR errors introduced by Acrobat Pro when it generates a ClearScan PDF?

I ask because there are a number of places in that PDF you posted where multiple words are combined into one "phrase". For example, near the end of line number six, you're only able to highlight the three words "of secondary importance" as one, rather than the individual words in the phrase. If you try to search for one of those three words, or the entire three-word phrase, it won't be found in the document. IMO, this is bad and defeats the purpose of a searchable PDF.

I too use Acrobat Pro XI and produce ClearScan PDFs and hit the same problem and was wondering of there's some way to fix this, even if you have to proof/edit by hand?

[jman]

dpc,

So far as I know, you cannot reverse a clearscan OCR with Acrobat XI. I'm also unsure if this is a problem only with XI or previous versions, too. I haven't found much out there on this problem yet.

What you can do is export your pdf to .tiff images (or whatever you want) and then re-import them as a pdf and OCR again. I did this for one error-ridden pdf and OCR'd using clearscan once again. I have yet to encounter the same errors. Of course, the other option is to avoid clearscan and just OCR using one of the other two options. But clearscan is nice because it is pleasing to the eye AND makes for smaller files!

I suppose another option is to keep a copy of your non-OCR pdf in case an OCR version goes bad, then OCR again. That's not ideal either.

[cday]

I don't have have Acrobat XI -- you're lucky that software prices are much lower in the USA -- but I've been fascinated by ClearScan since Adobe introduced it a few versions back. If it worked well it would certainly be the ideal OCR method: beautiful text and very small file sizes. But apart from the editability question, an older thread commented on the large number of typefaces that are sometimes generated for a single typeface in the document; I suspect that could in part be the result of imaging at well below the 600dpi Adobe recommend.

Following a discussion on another thread last week, I downloaded the Acrobat XI user guides and looked to see if there was a way of editing the output to correct OCR errors. I couldn't find one, but it's a complex program and user guides aren't always as comprehensive or clear as they might be. It would seem quite surprising if there really is no way of correcting errors... As ClearType is a typeface, maybe it's as simple as clicking in the output and getting an insertion point as in Word?? But is the OCR output actually displayed?

As proof reading is such a basic function it might be worth asking the question on the Adobe forum: I don't have the link but think I've seen useful replies from Adobe personnel in the past.

[M@rtijn]

Some info is documented on http://acrobatusers.com. Perhaps the following links might be of some help:

Optimize scanned pdf document
Correct ocr errors

[jman]

Thanks for the links. It doesn't appear to me that the update fixes this error, but time will tell.