topic Re: Camera intrinsics mismatch in Items with no label

Camera intrinsics mismatch

JMois — Fri, 13 Jan 2017 03:04:07 GMT

I've found that when I use the Calibration interface to query camera intrinsics (focal length and principal point), I get a different result than when I use the Device interface.

For example, with my SR300, using Intel::RealSense::Calibration:

calibration->QueryStreamProjectionParameters(StreamType::STREAM_TYPE_DEPTH, &depthCalib, &depthTransform);

reports focal length of {x=475.89, y=475.89} and principal point of {x=313.28, y = 240.76}.

But when I using Intel::RealSense::Capture::Device:

auto depthFocalLength = device->QueryDepthFocalLength();

auto depthPrincipalPoint = device->QueryDepthPrincipalPoint();

I get focal length of {x=463.88, y=463.88} and principal point of {x=320.00, y=240.00}.

Why do these two interfaces report different results? Which is accurate?

Re: Camera intrinsics mismatch

idata — Fri, 13 Jan 2017 19:37:59 GMT

Hello JonathanMoisantThompson,

Thanks for reaching out!

I'd like to share the links for the documentation of QueryStreamProjectionParameters, QueryDepthFocalLength, QueryDepthPrincipalPoint:

https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?querystreamprojectionparameters_pxccalibration.html

https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?querydepthfocallength_device_pxccapture.html

https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?querydepthprincipalpoint_device_pxccapture.html

If you read the description of the QueryStreamProjectionParameters function, it is mentioned that it returns the calibration and transformation parameters of the specified stream type.

Meanwhile on the documentation of QueryDepthFocalLength and QueryDepthPrincipalPoint it is mentioned that they return their respective parameters values but they also make an important note, the property value is the model fixed value, and not the device instance calibrated value hence the difference in the outputs.

I hope this helps.

-Peter.

Re: Camera intrinsics mismatch

idata — Thu, 19 Jan 2017 20:06:06 GMT

Hello JonathanMoisantThompson,

Do you have any updates about this?

-Peter.