Single page scanner - Glare of page

[mindtrapper]

Hello friends.

I have run into a problem after building my "scanner" (basically just a cradle with a stand). It is, of course, single page scanning with the camera on a tripod and doing left-right pages separately. A piece of 3mm plexiglass is used to keep the page flat(ish). I want to scan mostly paperback books that range in the thousand pages or so, having completed one so far.

This is an image out of the camera (Canon SX210 IS):

As you can see the image quality is mediocre and the focusing sharper in center. My lighting doesn't help much, a single desk lamp at non-optimal position. But my biggest problem is the opposite page that is pretty clearly visible near the "spine".

This is the output i get in Scan Tailor in BW and Color modes.

The BW version is clearly better, with the only problem being a bit of washed out text where the reflection was. The color output is bad. For this book (Internal Medicine textbook, maybe 5 pictures in 1020 pages) i used Photoshop Levels and Curves as well as Shadows/Highlights to even the light and make the reflection disappear. But this caused problems with DPI and much fuss down the line.

Now i am about to scan another textbook with much more color and photos, thus BW or Mixed setting in ST will not be an option. How can i set things up to make the reflection disappear? My angle is at 100 degrees. Is using a black piece of paper on the opposite page the only (slow) option?

Thanks.

[kalyan]

Change the lighting position as given in the image and check.

[TomHorsley]

I'm doing a similar single page scanner (which I'm working on making far too elaborate because it is fun . Instead of a plastic "platen", I got a cheap piece of leftover museum glass from a framing shop and it does seem to cut way down on reflections. You can see an example photo near the bottom of this web page: http://home.comcast.net/~tomhorsley/har ... l#hardware

[dpc]

How far away is your camera from the platen surface?

If you're using a platen angle of 100 degrees, a bit of trig tells me that the camera would have to be positioned about 26" away from the platen surface when capturing a 9" wide page (just assumed your setup will be doing a max of 8-1/2 x 11 inches), in order to not see the reflection of the adjacent page.

The formula for determining proper theoretical camera height is:

camera_height = (page_width/2) / tan(platen_angle - 90)

(platen_angle must be greater than 90 degrees)

Of course if you have a set camera height that you can't easily change (i.e. a scanner build with a fixed camera position), you can reorder the equation above and solve for platen_angle or page_width to determine how to get rid of the artifact by changing those constraints. Depending on your camera's ability to zoom in tight on the page, you may have to alter the camera height a bit as some of the point-n-shoot cameras have zoom steps instead of the smooth zoom of a DSLR lens.

Note to self: I really need to jot all of this down in a sort of "scanner design handbook" sometime.

[mindtrapper]

Thank you all for you input. I will work out my situation soon, but because i am a bit short on space (I am doing all of this on my desk and space is tight), i temporarily solved my problem by opening my cradle (thankfully one side of the cradle is movable on rails). I suppose this created a bigger angle and took care of the reflection problem but the book was moving a lot (damn paperbacks).

By the way, is the equation correct? With your example I am getting:

height = (9/2) / (100-90)
height = 4.5 / 10
height = 0.45

[dpc]

By the way, is the equation correct? With your example I am getting:

height = (9/2) / (100-90)
height = 4.5 / 10
height = 0.45

You forgot to take the tangent of the angle...

height = (9/2) / tan(100-90)
height = 4.5 / 0.176327
height = 25.52