how to use multiple volume in Kinect fusion C#?

Question

i am creating map using kinect fusion. I am moving camera scan the area I want to know how to use use multiple volumes in kinect fusion?

any one can help me?

Answer 1

I am doing something similar.
Using multiple volume is very simple in C#, you can simply just do new ColorReconstruction() then you have another volume.

Problem is that you need to do alignment between each volume.

My first approach is to create a new volume when camera pose is translated or rotated cross some certain threshold (like 0.3 meters and 30 degrees). Then record current camera pose, and stream the old volume's mesh into disk with a inverted camera pose transform.

but, switching from one volume to another causes big offset (how big is depend on how stable you hold the kinect) since the new volume may not start to do reconstruction at the recorded camera pose.

so i do a alignment between the first frame after i create a new volume and old volume. this limits the offset to like 5~10 cm. still too big.

now i am checking a approach presented in this paper , they do a method called volume shift by moving the data from old volume to new volume with a camera pose transform.