search

VladimirYugay / KinectFusion

3D geometry estimation from RGB-D data using Kinect Fusion approach
52 stars 9 forks source link

ICP question #8

Open linrui123 opened 1 year ago

linrui123 commented 1 year ago

Hello! Why the corresponding points become 0? image

lin-name commented 1 year ago

Hello, i faced the same question of the corresponding points =0, and when it came out, the code would abort. Have you solved it?

linrui123 commented 1 year ago

I'm sorry. I can't solve it. Have you ask the code writer of this code?

------------------ Original ------------------ From: @.>; Send time: Tuesday, Sep 6, 2022 11:01 AM @.>; @.>; @.>; Subject:  Re: [VladimirYugay/KinectFusion] ICP question (Issue #8)

Hello, i faced the same question of the corresponding points =0, and when it came out, the code would abort. Have you solved it?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

lin-name commented 1 year ago

Hello, i have tried to adjust the parameter，for example, incerase the MIN_POINT and the MAX_POINT for the TUM rgbd_dataset_freiburg1_xyz to set a more bigger voxel range to include the points which are probable out of the source voxel range, as followed: the source: 111

after modified: 112

The question of the corresponding points =0 has solved, and the output is right. But for my own rgb and depth data, the output is not total registered, maybe icp is still not enough for the depth image registration.

linrui123 commented 1 year ago

Thank you very much!

------------------ Original ------------------ From: @.>; Send time: Friday, Sep 9, 2022 2:31 PM @.>; @.>; @.>; Subject:  Re: [VladimirYugay/KinectFusion] ICP question (Issue #8)

Hello, i have tried to adjust the parameter，for example, incerase the MIN_POINT and the MAX_POINT for the TUM rgbd_dataset_freiburg1_xyz to set a more bigger voxel range to include the points which are probable out of the source voxel range, as followed: the source:

after modified:

The question of the corresponding points =0 has solved, and the output is rigth. But for my own rgb and depth data, the output is not total registered, maybe icp is still not enough for the depth image registration.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>