depth image does not corresponds to color image
I have bought a realsense d435 camera. When viewing in intel RealSense viewer the co ordinates of depth image is not matching with co ordinates of color image. I e It seems like the field of view of depth camera is around 20 percent more than that of RGB camera. Is this a calibration issue?. What can be done?
-
Hi Jeswinmp44 The D435 and D435i camera models have a smaller field of view on the color sensor than the depth sensor. This is intentional and not a bug or calibration problem.
The 2D mode of the RealSense Viewer tool does not have a function for aligning the two stream types but the 3D point cloud mode does. If you left-click on the 3D option in the top corner of the Viewer to enter point cloud mode and then enable depth first and RGB second, the color should automatically map to the depth coordinates to generate a textured point cloud.
If you wish to align depth and color but do not wish to do so with a point cloud, the RealSense SDK provides example programs that demonstrate how to do so in the C++ and Python languages.
C++
https://github.com/IntelRealSense/librealsense/tree/master/examples/align
https://github.com/IntelRealSense/librealsense/tree/master/examples/align-advanced
Python
The SDK provides pre-built versions of the C++ alignment examples that can be launched immediately. On Windows the examples can be found by right-clicking on the launch shortcut of the RealSense Viewer on the Windows desktop and selecting the Open file location menu option to navigate to the SDK's Tools folder..
On Ubuntu, if the examples have been successfully built after the compilation of the librealsense SDK then you should find them in the usr/local/bin folder. The alignment examples, if they have been built, may also be launched by going to the Ubuntu terminal and inputting rs-align for the 'align' example or rs-align-advanced for the 'align-advanced' example.
A stream that is being aligned to another type of stream will have its field of view size resized to that of the stream that is being aligned to. So when aligning depth to color, the depth FOV will be resized to the color FOV size. On camera models with a smaller FOV on the color sensor, this may result in the outer regions of the depth image being excluded.
The opening section of the documentation of the rs-align example provides more information about the principles of alignment.
https://github.com/IntelRealSense/librealsense/tree/master/examples/align
Please sign in to leave a comment.
Comments
1 comment