Hello,
I am working on a image processing project where the goal is to extract the wall from the background using depth image from the D415 camera. But, the depth information of the wall is acting like a background and after the binarization the information loss is large from the wall and it is not possible to recover it. The high accuracy preset was used with edge preserving filters in Viewer.
Is there something wrong with the camera settings and use? Are there any appropriate settings for this kind of capturing?
Thank you.
A depth-based green screen solution for the D415 and D435 called WildKey may be helpful to you.
https://mixcast.me/ditching-green-screen-mixcast-gdc-2018
A more technical solution might be to segment out surfaces using SLAM in combination with RANSAC.
链接已复制
If you are aiming to extract the entire wall instead of just a piece of it then the link below has advice on some methods for setting a maximum observable depth distance for the camera.
Hi Marty,
I have tried this, but no success. I set the min to 80cm, max to 1m, the camera was on 90cm. Do you have another suggestion? The difference between the wall and window is about 5-6cm.
Kind regards
A depth-based green screen solution for the D415 and D435 called WildKey may be helpful to you.
https://mixcast.me/ditching-green-screen-mixcast-gdc-2018
A more technical solution might be to segment out surfaces using SLAM in combination with RANSAC.
