[accessibleimage] Re: SV: Re: embossing tactel images?

I am one who believes that a significant improvement in the quality of an
embossed bitmap image can often be achieved by using image processing
algorithms to extract the salient features of the image before throwing
away the color and converting to either gray scale or black and white.

The attached zip file contains six files that illustrate my point.

The file named girlsFace.jpg is a full color image of the author of the
textbook titled Introduction to Computing & Programming, A Multimedia
Approach, which I use to teach my students how to do this sort of image
processing. (Her image is contained on a CD in the back of the book.)

The file named threeFaces.jpg contains three versions of that image
converted to black and white and sized for embossing on 8.5x11 inch paper.
The first two on the left are sampled for 12 dpi, and the one on the right
is sampled for 24 dpi.

The image on the far left shows the result of simply applying an intensity
threshold to the image, making everything that exceeds the threshold black,
and making everything below the threshold white. As I understand it, this
is a typical approach that is used for embossers that don't provide
variable dot height. Different algorithms may adjust the threshold up or
down, but this simply increases or decreases the amount of black in the
final result and doesn't usually improve the quality of the embossed
image.  This is essentially what you would see using Amanda's Juliet
embosser at 12 dpi to emboss the file named 1.sig, which is included in the
zip file.

The middle image shows the result of extracting the salient features of the
image using the full color information in every original pixel, and then
converting those salient features to black and white at a sampling rate
consistent with 12 dpi. This is essentially what you would see using
Amanda's Juliet embosser at 12 dpi to emboss the file named 2.sig, which is
included in the zip file.

The image on the right was produced using the same full-color algorithm to
extract the salient features of the image and then converting those
features to black and white at a sampling rate consistent with 24 dpi. I
believe this is pretty close to what you would see using one of the new 25
dpi Phoenix embossers. This is also what you would see if you were to use
Amanda's Juliet embosser at 12 dpi to emboss the two files named 3-0.sig
and 3-1.sig and then lay the embossed images side-by-side producing an
overall image size of 17x22 inches.

If you have a sig-compatible embosser and would like to see the embossed
result, feel free to emboss the four sig files contained in the zip file.
Don't forget to lay the last two side-by-side to see the total image. If
you happen to have access to a Phoenix, let me know and I will provide a
single sig file designed for one page at 25 dpi. I would be very interested
in hearing about the results.

Amanda and I are in the final stages of completing and publishing (for
free) a pair of programs that can be used to apply this algorithm to bitmap
images and to convert those images to sig files for embossing.

One program will be designed for blind users, will be fully accessible, and
will not require the use of a mouse. Some trial and error will be required
in some cases to achieve the best quality embossed output with that version
of the program.

The other program will be designed for sighted teachers and others who
assist blind users. This program will provide a mouse-driven slider along
with an onscreen display that can be used to optimize the output prior to
embossing.

(We will also be publishing similar free programs to support embossers that
provide variable dot height, for which improved results are also possible
through image pre-processing at the full color stage.)

Dick Baldwin


On Thu, Feb 9, 2012 at 11:08 AM, <fnugg@xxxxxxxxx> wrote:

> **
> Thanks, blush and skoal.
>
> Lisa
>
> Den 09.02.2012 16:43, skrev bmarek:
>
> Thank you for the complement, But it's thanks to YOU, Lisa, that we can
> get together and exchange ideas, discuss things and get things done! Let's
> all raise a toast to Lisa!
>
> Bob
>
>
>
> W dniu 09.02.2012 16:06, fnugg@xxxxxxxxx napisał(a):
>
> The future is very promising though with you, Amanda, Bob and many others
> working towards the development of new tools. Regards, Lisa
>
>
>
>
>
>


-- 
Richard G. Baldwin (Dick Baldwin)
Home of Baldwin's on-line Java Tutorials
http://www.DickBaldwin.com

Professor of Computer Information Technology
Austin Community College
(512) 223-4758
mailto:Baldwin@xxxxxxxxxxxxxxx
http://www.austincc.edu/baldwin/

Other related posts: