3D simple deformation reconstruction

asked 2018-04-30 09:58:51 -0500

DaluS gravatar image

Hello everyone,

I’m quite new in openCV, but OK with python and development, never coded in C++ but if it’s totally necessary I can try. I would like to reconstruct the deformation of an object in 3D, using multiple cameras.

My object look like this :


The movement can be more complex, but regular. It’s a continuous deformation. Instead of filming by the side it will be done in front.

I have never done any 3D reconstruction, and I’m quite lost in all the possibilities I could take to do such implementation. I don’t care how long the program has to run if it can be more precise. My goal is to get an evolution of the position of each points, kind of like this : https://youtu.be/qaCNiZncHY0

I saw a lot of various stuf with stereo depth map, feature detections, stereoBM… But I think my problem is not a general case, thus some methods might be more efficient/precise.

My questions are : _What is the best camera disposition ? (We will have an homogeneous light and a black background, with 60° angle camera, 4 of them) _What 3D reconstruction method should I use ?

edit retag flag offensive close merge delete



You can use sfm module in opencv_contrib or may be lib as openmvg to get 3d point. To track surface google is your friend : track deformable surface github

LBerger gravatar imageLBerger ( 2018-04-30 10:44:35 -0500 )edit