Reusing image poses from previous reconstruction for static images

[Alex Klimov]

Hello, everyone!

I have a setup with static cameras. I use colmap pipeline to get image/camera poses.

To speed up iterations I would like to skip as many steps as possible in colmap pipeline.

Can pipeline be simplified when cameras are static?

1. Extract features from new set of images.

2. Match features in new set of images.

3. Use previous camera poses to position images.

4. Run bundle adjustment to fix any minor errors, if some camera moved slightly between reconstructions.

Do I need to run mapper between steps 3 and 4?

What is the easiest way to import image poses into colmap db?

Do I need to write to database directly with python script?

[Wayne Cochran]

Wayne Cochran

way...@quintar.ai

On Nov 15, 2022, at 11:15 AM, Alex Klimov <alex....@gmail.com> wrote:

Hello, everyone!

I have a setup with static cameras. I use colmap pipeline to get image/camera poses.

To speed up iterations I would like to skip as many steps as possible in colmap pipeline.

Can pipeline be simplified when cameras are static?

1. Extract features from new set of images.

2. Match features in new set of images.

3. Use previous camera poses to position images.

4. Run bundle adjustment to fix any minor errors, if some camera moved slightly between reconstructions.

Do I need to run mapper between steps 3 and 4?

Typically you will use the incremental mapper to register new images and perform BA.

Note that the poses of all your images can change due to adjustment and

renormalization of t he coordinate system.

What is the easiest way to import image poses into colmap db?

Image poses are typically stored in the reconstruction, not the db. 

That is because multiple reconstructions may use the same db and place the

cameras at different locations and orientations.

Pose information can be stored in the db — usually prior info is stored there.

Do I need to write to database directly with python script?

Or you can use the C++ interface in base/database.{h,cc} — that’s what I do.

Or you can use sqlite directly.

--
You received this message because you are subscribed to the Google Groups "COLMAP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to colmap+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/colmap/7e29267c-601b-4af5-a6d8-f981d88f5246n%40googlegroups.com.

[Alex Klimov]

Thank you for reply! Allow me to explain my setup a bit better.

In my process - I have the scene changing, while cameras remaining in place. So somehow, I need to input to colmap both intrinsic and extrinsic parameters of the cameras. And hopefully allow colmap to use that information to speed up / skip reconstruction process for the new scene and jump to bundle adjustment right away.

Is that possible?