How can I create the depth_frame using RealSense SDK (Camera D405)
Hello
I am having a holes problem in the depth_frame received from the SDK, because of the brightness in the surface.
I use the two infrarred sensor to build the disparity map with openCV but the result wasn't good enough.
I would like to know how can I build the depth frame with the SDK, using the two infrarred images... My idea is to preprocess the infrarred images to try to elimate the efects of the brightness, and after that build the depth frame, but using the methods from the SDK
Thanks
-
Hi F Ricardo Auto As recommended in a recent D405 about glare at the link below, placing a linear polarization filter product over the lenses on the front of the camera can help to negate glare.
https://github.com/IntelRealSense/librealsense/issues/11904
RealSense 400 Series cameras already use the left and right infrared sensors to construct the depth image. This is done in the camera hardware at the point of capture before the depth image is sent along the USB cable to the computer. It uses raw infrared frames to do so, not the Infrared and Infrared 2 streams.
The D405 is not equipped with a projector component, which on other 400 Series models such as D415, D435 and D455 aids depth analysis by projecting a semi-random dot pattern onto surfaces for the camera to analyze for depth information. This means that plain, smooth or reflective surfaces may be more dificult for the D405 to depth-sense, resulting in more holes on the depth image.
An alternative method for dealing with plain, smooth surfaces without a built-in projector is to use an external visible pattern projector to project a texture onto a surface that a D405 can see and depth-analyze. Intel's white-paper document about use of projectos with 400 Series cameras has information about visible pattern projectors.
https://dev.intelrealsense.com/docs/projectors#6-the-consumer-front-projector
Please sign in to leave a comment.
Comments
1 comment