search

nmoehrle / uavmvs

UAV capture planning for MVS reconstructions
Other
54 stars 19 forks source link

About the trajectory about NY-1 #10

Open yilinliu77 opened 5 years ago

yilinliu77 commented 5 years ago

I took the trajectory following makeTraj.sh successfully and get a trajectory file ny1_opti.traj. However, I was confused about the content.

In makeTraj.sh, the overlap is set to 80/80. Which generate 140+ cameras in nadir. And optimize_trajectory will optimize the 140+ cameras. But in the SyntheticProxyAndTrajectory.zip, the trajectory file for NY-1 contains 400+ cameras. Does the spline calculation or tsp calculation generate new cameras? But after I run interpolate-trajectory, I get 2000+ cameras which have keys 0 or 1. I was quiet confused about it.

ngsmith commented 5 years ago

Hi, The amount of images you get from 80/80 overlap also is determined by the altitude you set. A lower altitude which results in a higher GSD will have more images to get 80/80 overlaop. We targeted I believe for NY-1 an altitude of 30m which led to around 400+ images to cover the area. The spline calculation generates interpolated positions between each camera position. The output of 0 is a spline interpolated position and 1 is the camera position, this way you can separate between the two. -Neil

On Thu, May 2, 2019 at 8:26 AM whatseven notifications@github.com wrote:

I took the trajectory following makeTraj.sh successfully and get a trajectory file ny1_opti.traj. However, I was confused about the content.

In makeTraj.sh, the overlap is set to 80/80. Which generate 140+ cameras in nadir. And optimize_trajectory will optimize the 140+ cameras. But in the SyntheticProxyAndTrajectory.zip, the trajectory file for NY-1 contains 400+ cameras. Does the spline calculation or tsp calculation generate new cameras? But after I run interpolate-trajectory, I get 2000+ cameras which have keys 0 or 1. I was quiet confused about it.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/nmoehrle/uavmvs/issues/10, or mute the thread https://github.com/notifications/unsubscribe-auth/AAMILLYJDGJOAOBVORFL5XDPTJ3RZANCNFSM4HJ3PSPQ .

-- Neil G. Smith, Ph.D. King Abdullah University of Science and Technology Research Scientist, GMSV neil.smith@kaust.edu.sa

yilinliu77 commented 5 years ago

I get it! Thanks for your explanation!

Does the qx,qy,qz,qw in the trajectory file mean the rotation quaternion? So I can multiply the quaternion with (0,0,-1) which means the original orientation?

And about the evaluation. The paper shows that it evaluates the result by error and completeness. About these two metrics, how can I ensure these two point clouds are at the same density? Or the density will not affect the result? How can I control the density of reconstructed mesh?

rebecca0011 commented 1 year ago

I took the trajectory following makeTraj.sh successfully and get a trajectory file ny1_opti.traj. However, I was confused about the content.

In makeTraj.sh, the overlap is set to 80/80. Which generate 140+ cameras in nadir. And optimize_trajectory will optimize the 140+ cameras. But in the SyntheticProxyAndTrajectory.zip, the trajectory file for NY-1 contains 400+ cameras. Does the spline calculation or tsp calculation generate new cameras? But after I run interpolate-trajectory, I get 2000+ cameras which have keys 0 or 1. I was quiet confused about it.

I have successfully build the project and download the data, but have no idea about how to change the file path in makeTraj.sh. Could you plz give me some advice?