08-18-2021 09:40 AM - edited 08-18-2021 10:23 AM
Hello,
I have an FLIR T series camera and I have connected it through USB mini-B cable with my PC but I cannot see it in MAX (Desktop webcam is visible in MAX).
Even IMAQdx does not appear in MAX but I can see the VIs of vision acquistion in the blockdigram by right clicking and going to Vision & Motion palette.
I have measurement and automation version of 2021 but my LabView and IMAQ, IMAQdx has 2017 version, is it because of that I cannot see the camera ?
08-19-2021 12:09 PM - edited 08-19-2021 12:10 PM
@Pukhtun_Yum wrote:
I have measurement and automation version of 2021 but my LabView and IMAQ, IMAQdx has 2017 version, is it because of that I cannot see the camera ?
No. The FLIR Cameras use a proprietary USB protocol that IMAQdx doesn't (and can't) know. You need to install a driver from FLIR that makes your camera appear as a DirectShow device in Windows, then you can access it as such in NI-MAX and LabVIEW IMAQdx.
Supposedly most FLIR USB-2 cameras are based around a sensor from Point Grey and you will liley need to install the FlyCapture driver from them, but as you have already seen in the other thread you started, this seems to be not an easy solution and you are even then not able to do more than just grab some images. And that is not so surprising since the DirectShow driver interface is mostly meant to be used for web cameras, where all you do is grab some pics. It usually does not allow to change many, if any, settings of a camera at all.
08-23-2021 03:41 AM
The camera is now appearing in MAX but as you said I cannot do anything else except grabbing pictures.
The problem is I want to get the value of temperature ( Spot 23.0 shown in picture attached ).
Can I do it by detecting the numeric value in the picture through OCR in vision development module ? What other options do I have to detect this temperature and show it as a numeric in LabView ?
08-23-2021 03:57 AM
Well, there is no other specific standard to transfer temperature values directly. IR cameras work the same as any camera, but except that they transfer numeric values for RGB, they transfer numeric values for IR light.
Your image that you get from the camera does unfortunately encode the temperature into RGB values. Great for direct human interaction, a little more cumbersome for automated machine analysis.
Basically you have to determine the RGB value at the point in question (which doesn't quite require OCR but simply determining the geometric location of the point of interest) then interpolate that into a scale that is setup according to the values in the bar at the right side of your image. And yes you could go and start to try to detect that bar, the numbers at the top and bottom and then build the interpolation array from the colors in between, but that will start to get a major project of its own, with very shaky results whenever the slightest change in the camera provided image happens.
Professional automated IR measurements use industrial IR cameras that directly can provide you a simple 8 or 16 bit grey-scale image that corresponds to the temperature you want to know according to preset min and max temperature limits. Maybe some of the higher end FLIR cameras provide that too, their lower end consumer market oriented devices are meant for quick human visual inspection, not automated measurements.
08-23-2021 04:27 AM
Why cant I simply take the temperature of the spot through OCR which is at the top left side of the picture ?
In order to take the bar ( to the right ) and perform interpolation, I still have to know the low and high temperature numbers of the bar, how would LabView recognise those numbers ?
08-23-2021 04:50 AM
You can, but depending on the actual temperature at that point, the color that forms the background will be often very different and that makes OCR rather tricky.
And what is the point of automating the measurement when you still have to position the point of interest manually?
08-23-2021 05:12 AM
The point of interest will be fixed and will not change from the start of experiment to the end.
Based on the image I have attached, how can I train the OCR so that it reads this number every time from different pictures ?
I read on internet to train the OCR but it is very confusing.