Operation manual for "Trace"

"Trace" determines the cartesian coordinates of the outline of a bright 
object against a dark background in a digital image with 256 grey-levels. 
Images can be captured with NIH-Image 1.60 (public domain, ftp-address: 
zippy.nimh.nih.gov), and preprocessed with Adobe Photoshop 3.0 on the 
Macintosh series of computers.


Included are:
The source code Trace5.txt, written in Fortran 77 by Absoft.
The executable program Trace5.exe, is compiled for running on a 
PowerMacintosh.


The experimental setup is a Leica MZ6 stereo binocular with a color CCD 
camera mounted and connected to a PowerMacintosh 8500/120MHz. The 
program is setup to microscope magnifications at 1x to 4x (having a 
objective lense inserted on the microscope of 1x, and no additional lense 
inserted in the photographic tubus).


Operation for outline determination:

1. Calibration for your specific environment:
Before measurements can be done the system must be calibrated in order 
to apply your specific hardware & microscope environments. For this take 
a micrometer-scale and grab images of the scale in both, vertical (=Y) and 
horizontal (=X) directions at several magnifications of the image.
Then determine the number of pixels per micrometer (=XPREC and YPREC) 
for both, X- and Y directions at each magnification (MAG). Produce plots of 
XPREC versus MAG and YPREC versus MAG. This results in two linear 
relationships of the form

	XPREC=AX*MAG+BX		(X-direction, in pixel/µm)
	YPREC=AY*MAG+BY		(Y-direction, in pixel/µm)

Determine the coefficients AX, BX, AY, and BY and insert the in line 9 of 
file CALIBRATION. After this calibration is performed the system is ready 
for routine measurements. The default condition is an image size of 
640x480 pixels. The threshold condition for outline detection of the 
object is at a grey-level of 49 (grey levels range from 0 to 255).




Image acquisition and pre-processing:
2. Grab image (256 grey-levels) with NIH (or another image capturing 
program) to obtain a 640 by 480 pixel image. If other image sizes are 
obtained, this must be specified in the first line of the accompagnying 
text file CALIBRATION. Sharpen and smooth with the tools given in NIH-
Image. Multiply the image by a factor of 1.25 (optional, but recommended). 
Store the image.

3. Re-open image with Adobe Photoshop.
Enhance contrast and reduce brightness to an optimum to distinguish 
between background and object.
Turn entire side, that is left to the object to black (use fill area tools in 
Adobe Photoshop).
Use enlargement and pencil tools to remove bright spots of the background 
that interfere with the object.
Filter entire image by applying interpolation-tool (radius of 4 pixels) in 
Adobe Photoshop (optional).
Store image in "raw" format.

4. Run "Trace"
For running "Trace" the input image must be stored in "raw" format and 
must be deposited in the same folder than "Trace".
Enter magnification used on the binocular. By default the program uses a 
grey-level threshold of 49 as a criterion for outline extraction. If other 
grey-levels are preferred this must be indicated in file CALIBRATION in 
line 5. Any integer number between 0 (=black) and 255 (=white) can be 
used.
Enter the name of the image file.
Enter whether the outline coordinates shall be given in pixel-coordinates 
(enter 1) or in micrometer coordinates (enter 2).

The program runs about 9 seconds to detect about 1200 outline points 
(less for smaller outlines) for a 640x480 pixel image on a PowerMac 
8500/120MHz.

Result is a file called "OUTPUT", which contains the cartesian X,Y-
coordinates of the outline.

4. "OUTPUT" can be opened and edited with any text-editor. For graphical 
display any graphical software capable of displaying X and Y coordinates 
(for example Cricket Graph) can be used.


Reference:
Knappertsbusch, M.: A simple Fortran 77 program for the outline detection 
of digitized microfossils. Manuscript submitted to "Computers & 
Geosciences" August 8, 1997.

Author's address:
Michael Knappertsbusch
Natural History Museum Basel
Geology Department
Augustinergasse 2
4001-Basel, Switzerland

Tel.: +41-61-266 55 64
Fax: +41-61-266 55 46
Email: Knappertsbus@ubaclu.unibas.ch