|
Image Algorithm Environment
| Contents |
| |
General
Downloads
Bugs/Limits
Screenshots |
| |
| General |
| |
I made this program, in 2001, for a company named
CamTek. It was originally suppose to
provide an unified environment for their huge amount of images, and to let
them run Image Processing Algorithms on these images.
The algorithms were of find-error-location type, so what the algorithms
report are points on the map, and so they are displayed on an image. as little
circles on an image.
For checking algorithm's performance there is an option to hand-point the
errors, and then compare it to the algorithm's points. you can view an image,
zoom in and out, place point on errors, delete points, and bring on-screen
various results from the algorithms.
|
| |
| Downloads |
| |
The main project, written in VB6, is
available here.
A DLL project, that helps the VB program where VB isn't good enough, and
handle all the connections with algorithms DLLs, is
available here.
A Sample Image Algorithm DLL, is
available here.
Additional files may be needed for running this. (namely
msvbvm60.dll and mscomctl.ocx,
zipped) |
| |
| Bugs/Limits |
| |
1. The program assumes that every image is of predefined
constant size. It shouldn't be problem at all, if you only have one
image/level, but if you have more then one, there will be gaps between the
images. I hope there won't be other affects.
2. Other limitation is that in a layer, the algorithm is being run on each
image separately .If the algorithm need to look on all the layer in the
processing, you'll need to change the program.
3. There is an option to set a width of a hand-defined error. through the
error's properties panel, there is a "width" button, and if you push it, the
panel disappear, and you goes back to the results panel, where you should
press, drag a line, and release to return to the properties panel, with the
"width" field if filled. the hole process isn't clear and simple enough, and
can result some weird results. this corner should be smoothed or disabled. |
| |
| Screen Shoots |
| |
 |
This is the Image Organizer, as you can see, it's organized
in a tree, and in the leafs there are two type of images: the about is
the image, and below there is the CADs, which are some master bitmap,
that the images above should look like. notice that every image and cad
have two-coordinates, so each "layer" can but a sum of images. |
| |
 |
Here you can see an image, with a hand-pointed errors, (the
little red circles) and algorithm-pointed error. (the little green
circles) the algorithm used is the sample algorithm, that just look for
specified shade of gray, so it's figured that the points that it found
are always between the somewhat white, and the black areas. |
| |
|
