Hello everyone.
I am currently using cv::rgbd::registerDepth() function from the rgb module to register a depth camera map (RealSense D415) with and an external RGB camera.
The function works very well !! But it is al little bit slow for my needs. It is for a video application, so registerDepth() is called on every frame. Since all my code already runs in cv::cuda. I am trying to find a faster version of this function that already runs in CUDA. If such port exists, can you tell me where it is ?
Also, I am newbie on modern computer graphics, so any other suggestions on the correct use of registerDepth() for a video feed are welcomed. I was thinking to call registerDepth() only in a fraction of the frames and apply the same map on the rest of the frames... Is it a good idea ?