Official RTAB-Map Forum - Re: Losing odometry when doing fixed point scanning

Re: Losing odometry when doing fixed point scanning

Posted by matlabbe on
URL: http://official-rtab-map-forum.206.s1.nabble.com/Losing-odometry-when-doing-fixed-point-scanning-tp4336p4411.html

Hi,

yes this constructor is for that. Examples for 4 cameras:


4x RGB merged into single cv::Mat:
 _____ _____ _____ _____
|     |     |     |     |
|     |     |     |     |
|     |     |     |     |
|_____|_____|_____|_____|

4x Depth merged into single cv::Mat:
 _____ _____ _____ _____
|     |     |     |     |
|     |     |     |     |
|     |     |     |     |
|_____|_____|_____|_____|

With OpenCV, you can create a large image (rows, 4xcols), then use Mat.copy() to copy each image in the right ROI. Then for each cameraModel, set the appropriate localTransform. You can then use DatabaseViewer to see all projected cameras in the 3D View, with enabling frustums (right click on the view) to see the camera orientations.

cheers,
Mathieu