I was curious if there are guidelines for thermal cameras being utilized for detection related to the FOV coverage areas and PPF categories similar to standard cameras (Detection, Observation, Recognition, Identification, Inspection) and what those approximate PPF values are for. I realize that you are trying to both detect objects and be able to discern between temperatures as part of this, but what are the guidelines for determining how much area a thermal camera can effectively operate in?
Subscriber Discussion
Thermal Camera FOV
GC
Greg Cortina
Jul 20, 2017FLIR sales guy answer:
The combination of lens in mm, the pixel pitch in microns, the resolution of the camera (320x240 or 640x480) and the speed of the lens (f. Stop) all play a factor in the image that will be generated. Temperature, background and humidity can also factor.
From there, you now have to find out what the analytics processor you are using requires. In the old days with VideoIQ our cameras were defined by a 200' FOV on 640x480 and a 150' FOV on 320x240 for human detection and classification.
Others use from 2 Pixels on Target to 125 for recognition (classification) of a human walking.
All of this factors in to the acceptable level of false positives, nuisance alarms and probability of detection. (PD/FAR/NAR)
The Johnson Criteria really doesn't apply in today's detection and analytics requirements.
Start with your chosen analytics and work it backward from there is my point of view.
Actual temperature measurement can be accomplished with just a few pixels but requires a fully calibrated imager and an understanding of emissivity.
True temperature detection isn't used in any security detection algorithm I have used, maybe someone knows otherwise.
Greg
U
Undisclosed #1
Jul 20, 2017Actual temperature...True temperature...
Are you referring to the same thing or is there a distinction?
GC
Greg Cortina
Jul 20, 2017Fair question.
I didn't intend for them to be separate meanings, just late night phone messaging.
U
Undisclosed #1
Jul 20, 2017True temperature detection isn't used in any security detection algorithm I have used, maybe someone knows otherwise.
Though I remember hearing something about FLIR releasing "calibrated radiometric security cameras" years ago. Did that ever happen?
U
Undisclosed #2
Jul 20, 2017"Actual temperature measurement can be accomplished with just a few pixels but requires a fully calibrated imager and an understanding of emissivity"
Please tell me
How many pix do we need to measure temp? ( hor and ver )
thanks
U
Undisclosed #1
Jul 20, 2017How many pix do we need to measure temp? ( hor and ver )
Interested as well in the answer. I always assumed that each pixel measures temperature on its own.
U
Undisclosed #2
Jul 20, 2017GC
Greg Cortina
Jul 20, 2017That's good, I could give you two by myself. I believe we are answering different but similar questions.
Each pixel measures temperature. An image is built by assigning colors to the temperature values. Each pixel is calibrated to a constant at the factory and typically a flag in front of the lens on occasion is used to balance that in the field (flat field correction).
As for radiometric temperature measurement, a single pixel can provide a reading. Is that the recommended amount of pixels? Maybe, maybe not. That depends on the acccuracy required, environmental concerns and more.
For that, we do a great ITC class for Thermographers.
JK
Jim Krile
Jul 20, 2017Greg: I knew that there was a lot that went into this for Thermal, probably as much if not more than standard visible. Thanks for the input.
New discussion
Ask questions and get answers to your physical security questions from IPVM team members and fellow subscribers.
Newest discussions