ScenTraK™ Pattern Recognition
The output data collected by the microprocessor from an entire sensor array can be graphically
represented by 3-D plots, such as shown below. Each plot displays the temporal responses of the entire
sensor array to different odorant. Identification occurs by comparing the test sniff response pattern
with those patterns stored in memory and statistically determining the best match.
The output data collected by the microprocessor from an entire sensor array can be graphically
represented by 3-D plots, such as shown below. Each plot displays the temporal responses of the entire
sensor array to different odorant. Identification occurs by comparing the test sniff response pattern
with those patterns stored in memory and statistically determining the best match.
With ScenTraK's pattern recognition, positive identifications are made not only when many sensors
respond (ammonia) but also when few sensors respond (CS or “tear” gas and parathion, a pesticide).
The concentrations shown here are all at the OSHA standard PEL (Permissible Exposure Limits) for the
individual compounds.
The statistical method of pattern recognition allows the device to be trained to detect different vapor
phase targets by taking numerous sniffs of the target and then storing the array response as a target for
matching and identifying the sample.
respond (ammonia) but also when few sensors respond (CS or “tear” gas and parathion, a pesticide).
The concentrations shown here are all at the OSHA standard PEL (Permissible Exposure Limits) for the
individual compounds.
The statistical method of pattern recognition allows the device to be trained to detect different vapor
phase targets by taking numerous sniffs of the target and then storing the array response as a target for
matching and identifying the sample.
 | |  | |  | |  | |  |