Ok, basically I've looked all over the internet and I've been at a loss for any kind of useful information.

The project:
I need to be able to build a linear CCD camera array and be able to use it like a document scanner you'd find at any best buy. We're scanning massive documents though and we need things at true 1200-5000 dpi.

Essentially, I have no idea where to start. I need to be linking 4 or 20 or 50 together in an array, depending on the ccds . . .

This is all from scratch, I'll need to build the array, mount it to a step motor and be able to actually scan something with software thats already out there, or that will have to be created.

Any advice or tips you guys could give me would help immensely, as I'm completely lost.

Thanks much.

 

This is a hugely complicated project for an inexperienced project builder to take on.

How much time can you take to build this? Is there a fixed deadline?

 

Call NASA, fire up the supercomputers, this is going to be a late night boys! WOW you picked a whopper. First, take an existing one apart, study it, learn everything about it, then, and this is important, put it back together. If you can do that then you will have a good understanding of what the next step will be.

Good Luck

 

We don't necessarily have a deadline, but of course asap. I'm more concerned with being able to understand all the components in their entirety. I'm willing to put in the time and effort, but again, I'm quite inexperienced.

Since most of the off-the-shelf kind of things are proprietary, it seems if you take it out of the casing it just wont work. Where would be a good place to start?

 

You can always start with the basics - http://computer.howstuffworks.com/scanner1.htm

You may notice that a scanner uses the same basic means of converting light to a voltage, but the CCD array is considerably different from a camera's. A conventional scanner can offer up to 9600 DPI.

 

One way to start might be to divide the project into separate subsystems.

One subsystem might be the operating of the CCD array by a computer - clocking the cell charges out to an ADC and reading and storing the values.

One might be making the moveable document tray, with smooth linear bearings, a motor, a glass surface, limit switches, and a program that can drive the motor and detect the limits.

One might be the optics - focusing an image thru a lens onto the array, including lighting the image.

Then you need to combine the CCD readout program with the tray movement program to scan an image.

Finally, you need to store the raw values into some image format. There are dozens of commonly used ones to choose from, but each has plenty of details you'd need to learn.

The nice thing about this method is that each part is much less complicated and is easier to test and debug.

 

Our stepping motor has the ability to step 1/1000th of an inch so that should help dpi. Standards for what we scan goes up into 5000 dpi, with minimal interpolation. We're mapmakers, so that's why it has to be so picky.

Honestly, if I could figure out how to just use one camera and have it step a section of the document one row, move down to the next, then repeat and produce one massive image, that would be ideal. Then I can start building on to the array and scan a large area every step.

I don't want to use a scan arm because the documents are up against a suction board and the documents will range in size. Or would a scan arm be the best way?

Basically, picture a document thats 6 feet by 6 feet and you have to capture every detail perfectly up to 1/1000th. Thats usually what we do.

Right now we have a standard flatboard scanner that we have on a stepping motor. It does a pass and scans an area, makes a file, steps and repeats for the entire thing, then we hand assemble the pieces together.

 

I agree Davebee and I think subsystems will be the way we go.

 

Look here for one example (they built this in the 80's): http://www.nso.edu/general/computing/fms.html
Also, contact Dalsa for linear CCDs and their TDI technology: http://www.dalsa.com/mv/knowledge/TDIprimer.aspx