We were very surprised and pleased as to how well
our scanner can read white-on-black. The Jack Daniel's logo was scanned from an
image the size of a full sheet of paper. However, the decorative lines in the
logo were much thinner than you can tell from the TV image. The white-on-black
actually seemed to be amplified by the photosensor as even small details were
picked up by the scanner. Scanning black-on-white was less reliable; we used 150
sized font for the initial testing of our scanner, which gave reliable results.
We scanned a test image of lines from 16 point font down to 10 points. While all
these lines were at some point detected by the scanner, the results were very
splotchy and unnacceptable. We recommend at least a font size of 50 points or
above to get a consistent image for black-on-white.
Our half-step option also worked pretty well. While there is not much noticeable
difference in the quality of the image, the image resolution is much larger.
The variability in the quality of our images can mostly be attributed to the
vibrations on the photosensor due to the stepper motors. Also, the surface of
the scanned image must be flat. Any rise in the elevation of the paper may
result in a black pixel being detected since the photosensor must rest a certain
distance from the image surface for optimal sensing.
Even our lab TA had doubts as to whether we would finish, but we were able to
get things working and started scanning images almost a week before the project
demo. We have since added various functionalities to the scanner such as 3 scan
modes, a half-stepping option, mid-scan TV display, zig-zag image compensation,
and software reset. We think our project has met if not exceeded expectations.
For the future, we would have liked to implement a USART connection to save our
image data as a bitmap on the computer. We would have also liked to implement
grey-scale imaging on the TV since our photosensor is actually taking grey-scale
image data. Currently, it is lost due to the black and white television code. We
would also have very much appreciated some machine and power tools for the
construction of our scanner.
The only standard we can think of that was used in this project was the NTSC
video standard. The code for this part was obtained from Prof. Bruce Land's
video code from Lab 4 on the ECE 476 web site.
That being said, intellectual rights of the video code belongs to Prof. Land. We
also used a ULN2003 motor drive circuit from the Postage Meter 2002 project web
page owned by Aksel Coskun and Hossein Sheini.