search

cdcseacave / openMVS

open Multi-View Stereo reconstruction library
http://cdcseacave.github.io
GNU Affero General Public License v3.0
3.19k stars 892 forks source link

how to DensifyPointCloud with a few image with known camera extrinsic and intrinsic parameters? #181

Open wguo68 opened 7 years ago

wguo68 commented 7 years ago

All SFM programs can't detect enough features and do feature matching. So OpenCV MVG can't work for these images. But I know the camera parameters for these images. how to use OpenMVS to do 3d reconstruction?

cdcseacave commented 7 years ago

You can't, cause you need also a sparse point cloud. Send me the images and the extrinsics and I'll try to create a project for you. In the future the code can be modified to work without sparse point cloud.

pmoulon commented 7 years ago

@wguo68 You can use OpenMVG to generate the initial point cloud and then use the OpenMVG2OpenMVS and then run OpenMVS tool.

You will need to initialize the 3d scene in the OpenMVG format (sfm_data.json) => Setup a view for each image, intrinsics and extrinsic. Then run the feature extraction and then run ComputeStructureFromKnowPoses that will look for the corresponding point accross your images and triangulate the good one.

cdcseacave commented 7 years ago

Once again, I need to see the images to give you an answer.