Help for BESTFOCUS

Purpose: To combine a set of images taken of a 3-d object
         at progressively different focus positions and create:
1. A single best focus image.
2. A relative depth map (topo or DTM).

Usage: bestfocus in=p1,p2,p3,...pn out=focus,topo parameters
where:

p1...pn are input images with incremental changes in focus. The object
must be a 3-d target.

'focus' is a composite image made from the best focus locations of 
all the input images. Same format as input.

'topo' is a scaled relative depth map or topo map (dtm map). See the
"scale" parameter for scaling. Without scaling the topo image values
range from 1 to nin where nin is the number of input images.
Format is REAL.
 
Restrictions:
The focus must monotonically increase (or decrease) in the order of the
input images.



METHOD:
Bestfocus computes the local standard deviation in a moving box nlw
lines by nsw samples across all input images. It then computes a
curve of standard deviation versus image input number at each pixel.
The image with the highest standard deviation is nearest the best focus
and this image is selected for the "focus" image at this pixel.
The peak of the curve is determined by quadratic fitting to the highest
point and it's two lower neighbors. This number from 1 to nin (the
number of input images) is then scaled by the "scale" parameter and
written to the "topo" image. output=topo*scale1+scale2 (see scale
defaults). The "inverse" parameter can complement topo before scaling.
If the highest standard deviation image is 1 or nin then no quadratic
fit is made and the result will be 1 or nin respectively. Quadratic fitting
results in non integer topo values but the limit will always be between
1.0 and real(nin).
Note a low pass median filter will improve the "topo" image.

HISTORY:
2-2004  J Lorre. 
COGNIZANT PROGRAMMER:  Jean Lorre