VFINDER (cam, mode, dmode, how_many_total {, times[]}) ibr

Function

This instruction performs ObjectFinder recognition using the planning associated with the given virtual camera.

Parameter s

cam Real-valued expression indicating the virtual camera's planning to use and parameters to read.

mode Recognition operation to perform. Default value is 1.
1 = ObjectFinder recognition-normal recognition
2 = ObjectFinder recognition for multi-instance training-
how_many_total should be -1; see the description for
how_many_total below
3 = ObjectFinder learning mode-learns from the recognition
successes and failures to automatically improve the recognition
strategy for faster recognition
4 = ObjectFinder automatic multi-instance training-internal model
representations (feature weights) are refined as more part
instances are found

dmode Real-valued expression that specifies the display mode to use when displaying the border of the window:
-1 No draw
0 Erase
1 Draw solid (default)
2 Complement
3 Draw dashed
4 Complement dashed

how_many_total
Specifies the total number of objects to find. In the case of multiple models planned together, this is regardless of which models are found. DEFAULT = -1 (find as many objects as possible). If 0 is given, none will be found, but the image is still processed through feature extraction.

For mode 2 when how_many_total is -1, recognition will not stop until all confirmed proposals are checked and the best proposal is chosen.

times[] Array of max_time information for the recognition process. The whole array is optional, defaults are as if times[0] = 0.
[0] How many max_time parameters to follow.
[1] Max time to first part (defaults to V.MAX.TIME[cam]).
[2] Max time to subsequent parts (defaults to times[1]).

ibr Integer value specifying the image buffer region within which to search. The AOI must be a nonrotated rectangle. For ObjectFinder, the AOI width is reduced, if necessary, to a multiple of four pixels. Image buffer regions specify both a size and a virtual frame buffer (see the description of VDEF.AOI).

NOTE: There is no optional shape specification: It must be an image buffer region (ibr).

Details

This instruction performs ObjectFinder recognition using the planning associated with the given virtual camera. As shown in the next section, several of the same switches and parameters apply as for prototype recognition.

The ObjectFinder learning stages are invoked by the values 3 and 4 for the mode argument in this instruction. During each VFINDER in learning mode, it will reorder the features classes after all recognition is complete. In automatic multi-instance learning mode (mode value 4), if the add instances ok flag in the model is FALSE, or if the maximum number of instances to accumulate has been reached, then any new instances will be ignored. The maximum number of instances is 17.

To set the add instances ok flag in the model, use the VTRAIN.FINDER instruction. See VTRAIN.FINDER (cam, mode, dmode, arg, arg2, arg3) $model_name, ibr = value, value2, value3 for details. To read the value from the model, use VSHOW/VFEATURE. See VFEATURE (index) for details.

Input parameters using system switches and parameters indexed by virtual cameras

NOTE: All switches and parameters should use the default settings, except as noted here.

The following settings are required:

V.MIN.AREA[vc] = 4

V.MIN.HOLE.AREA[vc] = 4

The following settings are suggested:

V.MAX.PIXEL.VAR[vc] Same as for models trained.

V.EDGE.STRENGTH[vc] Same as for models trained, assuming that neither F-stop nor illumination has changed significantly since the model was trained.

V.MAX.TIME[vc] = 2.0 Used if the times[] array is omitted.

V.MAX.VER.DIST[vc] = 5.0

The following switches are available for enabling visual feedback:

V.SHOW.FEATS[vc] Shows final features used for ObjectFinder.

V.SHOW.VERIFY[vc] Same as for prototype recognition. The lines and arcs of the model are compared to the edges in the image, and they are drawn in the Vision display window.

V.SHOW.RECOG[vc] Overlays the models on the found instances.

Syntax

VGETCAL-I