|
exeDERARegTrack
|
| Centre for Vision, Speech & Signal Processing |
The task is to detect small and indistinct targets in image sequences and track them through consecutive frames. This is achieved using a multi-stage approach which combines image segmentation, motion estimation and data fusion. The algorithm is described as a note on the DERA Small Target Detection page which is in the Applications subtopic of the AMMA class documentation.
$(NAME)$(NUMBER).pgmwhere $(NAME) is the sequence name and $(NUMBER) is the current frame number which should not have leading zeros. All frames must be the same size.
Assuming that the frame numbers run from $(START) to $(FINISH) then the executable is run with the following compulsory command line options:
DERARegTrack -Start $(START) -End $(FINISH) -i $(NAME)Output will be generated to the standard output stream and result image files will be created with the following names:
$(NAME)$(NUMBER)_F.ppmThe output formats are described here.
For example: Consider the sequence with images b2_20_40_0.pgm to b2_20_40_10.pgm. This sequence would be processed using the following command line:
DERARegTrack -Start 0 -End 10 -i b2_20_40_
Usage: /vol/vssp/local/trees/amma-4/linux/bin/DERARegTrack [options] -optload () [] Load options from a file. -optsave () [] Save options to a file. -Start (0) [0] First frame number -End (0) [0] End frame number -i () [] Path and filename for images -v (FALSE) [FALSE] Dump all images -s (500) [500] Maximum segmented region size -minss (2) [2] Minimum static region size -maxss (200) [200] Maximum static region size -avcont (3) [3] Minimum average contrast -grcont (3) [3] Minimum gradient contrast -maxcomp (9) [9] Maximum compactness (perimeter/area) -maxmt (2) [2] Maximum acceptable motion error -hist (4) [4] Length of history to examine -help (TRUE) [FALSE] Print usage information. Dependancies: -Start is compulsory. -End is compulsory. -i is compulsory.
| Author: Robert Crida, Created: 17/6/1998, Generated by DocExe: March 20, 2001 |