vttrack − detect and track a set of templates in an image sequence

vttrack if=<infile> [ig=<edgefile>] [gf=<vfile>] [of=<outfile>] t1=<tfile> [th=<threshold>] [t2=<tfile>] [th2=] [is=] [ol=] [-v]

Vttrack is a tracking program which determines the locations of a set of predefined templates in the first image of an image sequence and tracks them as long as possible through a set of consecutive frames. The current version of vttrack is limited to a maximum of two explicitly defined templates.

The strongest template matches are first determined by an identify exhaustive scan for all templates of the first image frame. Then an exhaustive local search is made in each new frame to track a detected template; the search is constrained to a "radius" distance which is currently defined as half the x extent of the narrowest template. Every five frames an exhaustive search of the whole image frame is made to identify any newly emerged template matches. To ensure reliable tracking, overlapping of tracked templates is not, in general, allowed. In more detail the centers of two tracked templates cannot be closer than two times the "radius". When this situation occurs during tracking the trajectory with the least support is terminated.

Normally, both an intensity image sequence and its corresponding edge image sequence are read by the program. The default match criteria is to sum the pixels in the edge image that correspond to true values of the template. An exact location of a match is determined by a local maximum; only one match is permitted in a distance of "radius". The recorded match value is normalized by the number of elements in the template so that comparisons between different template matches can be made. A threshold determines when correlations are too small. The threshold is set as a percentage of the total elements in the template; the default threshold is 30%.

Vttrack can generate three output files. The output file specified by the of= parameter is an annotated image set in which the location of each feature is marked with the template superimposed in the matched location. The second output format is a vector file (suitable viewing with v3d) specified with the gf= parameter. In this file the temporal trajectories of each feature are stored. The third trace file is called "loc" and is output when the -l flag is specified. It contains tracking information in ascii format.

The input images must be of type unsigned byte. Only two templates are currently allowed. When a template is not located within the search window it is not tracked further.

if=

Specifies the input file which must be a VisionX byte image

ig=

Specifies an edge input image which should be binary valued (true=255). Typically this is created with vsobel -d | vedgex.

of=

Specifies the name of the annotated image set output file.

gf=gfile

specifies the vector output file which contains the trajectories of the templates.

t1=

specifies the first binary valued template image.

t2=

specifies the second binary valued template image.

th=

specifies the threshold (percentage), the default threshold is 30%.

th2=

specifies the threshold (percentage) for the second template if different than that of the first template.

gt=

specifies a grey level pixel threshold for identifying the class (independent of which template is the best match). The grey level of the object is determined by the average grey value of the central 3x3 region of the template. If the mean value is below the specified threshold then class one is indicated otherwise class two.

i=

Set the frequency that identify scans are made. (default is every 5 frames) ol= Set the permitted overlap between detected templates. (default is 1 pixel)

-v

specifies a verbose mode of operation.

-l

specifies that the ascii trace file called "loc" is to be generated.

M. Y. Chen and A.P. Reeves