So there is a market for personal receipt and document scanners of the quick(er), feed in variety.  Neat Receipts is an awesome product that combines a small (non-flatbed) scanner with character recognition software and quickly turns your documents into spreadsheet data and retains a PDF backup.  Lately, I have done a lot of scanning and faxing right off of my smart phone.  With the right lighting and decent enhancements, text can be quickly taken off of the page and the finished result at the email or fax machine of the recipient is more than satisfactory.

So here I sit at the edge of my bed, looking to my desk.  Under the desk is a large amount of paperwork, and frankly I haven’t had the time to sit and scan everything into the scanner.  It is fast, but not that fast.  It’s hours of loading, and I’m just a small scale contractor currently.

The solution, as I envision it, is to place my cell phone on a small stand.  A very thin universal, clamp type holder that basically just holds the phone still, facing and about 12 inches from the surface of any desk.  It would be an arm, and some manner of a base, like a long necked version of one of those desk lamps with the flexible neck, but instead of a lamp at the top, there would be a cell phone clamp.  (If you can’t tell yet, I am writing this as the details come to me)  Why get rid of the lamp, lol.  Let’s put an LED lamp on each side of where the cell phone will reside (I’ll put a drawing up soon).  That way you could create a better environment of proper lighting for the “scanning.”

Now, the character recognition is rather demanding, and will slow down the process, so I think it should be done in stages.  You will want to be in capturing mode, and only do the OCR (optical character recognition) mode after you stop taking pictures.  The software could be set up to automatically just jump in and begin the conversion whenever you pause taking shots or it could wait for your go ahead.  So you would set the cell phone into the holder, position a pre-printed calibration sheet on the desk and allow the phone to calibrate itself for the light and focus.

As soon as the phone is ready you could begin.  The thing to understand is that the camera in your cell phone can be zoomed way outside the edges of the document and still have more than enough detail to fully recognize the text of even the smallest text on a sheet of paper.  This means that the document you are scanning does not have to be oriented or aligned in any particular way, as long as it is within the field of view of the camera.  This means speed.  A speed currently unrealized.

Using a finger anywhere on the page, all you would need to do is pass the paper through the field of view.  The camera would be set in focus and would automatically looking for geometric shapes to come into the field of view.  Upon recognition of the shape, and as it passed a particular threshold of the camera’s view, the camera would snap a single shot, quickly saving it to the camera’s memory.  Literally faster than a page a second, (limited only by what the camera could do) this process could be repeated through a stack of papers.

Upon completion of a group of documents, the second phase of the process would kick in, and the software would one at a time, take each picture and highlight the geometric shape (document) within it.  It would then crop out everything except the geometric shape, and lastly it would recognize the finger used to slide the paper across the field of view, and it would white it out from the image.  These images would be saved as PDF.

The third phase would kick in, and the OCR process would take all the data from the documents.  If it determines it is a receipt, it will save the data in an excel sheet.  If it determines it is a page from a book, then a rich text file will be generated.  So now you have a original source document PDF, and you also have a file with the data derived from the source.

Doing the process in batches and phases as I describe, would allow you to process a large group of files quickly.  It causes my mind to take it to the next step, which would be, camera permitting, a rapid version of the above.  Perhaps, a book reader, or movie speed (many frames per second)  Make no mistake, everything mentioned above is individually available right now.  I’m just suggesting we tie it all together.