DJI MAVIC 2 EA thermal sdk misalignment

Completed
Follow
Carl

Trying to calibration my IR camera and I found the following results. The first image is the checkerboard corners we found in a .jpg image. The second is the corresponding results in the .raw image using dji_thermal_sdk_v1.4_20220929. It is not a common issue. Can anyone kindly tell me why this happens?

0

Comments

13 comments

Date Votes
  • DJI Developer Support
    Could you upload the original image to this address for us to analyze? https://pan-sec.djicorp.com/s/ZSq7RE9CeF49GBc password：dji123
    0
    Comment actions Permalink
  • Carl

    Thanks. I have uploaded the IR image and .raw file. I am using dji_thermal_sdk_v1.4_20220929 for extracting the temperature field.

    0
    Comment actions Permalink
  • DJI Developer Support
    Thank you for your cooperation. I have reproduced this issue, and we are currently investigating the cause. Please wait for some time.
    0
    Comment actions Permalink
  • DJI Developer Support
    I apologize for keeping you waiting for so long. The situation shown in the photo is normal. This can happen when there is high-speed movement of the aircraft or camera, causing a discrepancy between the colors corresponding to the infrared sensor data and the original photo.
    0
    Comment actions Permalink
  • Carl

    Thanks for your feedback. I am not sure I follow your meaning. Does it mean the two photos coming from two sensors? I understand there is a discrepancy between visual and infrared images. And we have observed this phenomenon. But the second image was generated using the infrared image. So I don't think it is caused by high-speed movement. Or can you give me more details about how your SDK works? For now I think the infrared camera records the temperature field, and display it in fig 1, but in fig 1 there is only relative values. By using your sdk, we can get the temperature field. So there shouldn't be a discrepancy here.

    0
    Comment actions Permalink
  • DJI Developer Support
    You can understand it as two different sensors because temperature data must be obtained by the infrared sensor, while the original photo not only uses the infrared sensor.
    0
    Comment actions Permalink
  • Carl

    Still hard to understand it. I see this phenomenon might be caused by the two different sensors. But why there are two different sensors? The temperature data is extracted from the original photo. So no matter how many sensors here, the temperature data and photo should use the same group of sensors, does it?

    0
    Comment actions Permalink
  • DJI Developer Support
    The issue with M2EA on the thermal camera is indeed a design problem, so it cannot be resolved through software.
    0
    Comment actions Permalink
  • Carl

    Thanks very much for your explanation. We are using it for research purpose. If possible, may I ask what exactly this design problem it is?

    In addition, we are considering an upgradation. Has it been solved in the M3T?

    0
    Comment actions Permalink
  • DJI Developer Support
    I believe there won't be this issue on M3T. I am verifying this with the engineers, and I will reply here again after receiving confirmation.
    0
    Comment actions Permalink
  • DJI Developer Support
    Thank you for your patience. We have confirmed with the engineers that the M3T will not encounter this issue. The M3T and M2EA use different infrared lenses.
    0
    Comment actions Permalink
  • Carl

    Many thanks for feedback. It is very useful.

    0
    Comment actions Permalink
  • DJI Developer Support
    I am happy to help you.
    0
    Comment actions Permalink

Please sign in to leave a comment.

Didn't find what you were looking for?

New post