Motion Capture application
Captures video from multiple cameras. Output 3d points tracked from reflective markers attached to a human form.
* uses off the shelf usb cameras. will be validated with PS3 eye camera.
* uses at least 3 cameras with 16 maximum
* uses OpenCV to capture input
* calibrate intrinsic camera parameters - focal, distortion, etc. - with a chessboard pattern
* simple 3d extrinsic camera calibration. Ideally moving a single point light through the capture volume. See [login to view URL]~svoboda/SelfCal/
* tracks multiple reflective markers attached to a human - approximately 34 on average
* output stream of 3d points in c3d file format: [login to view URL]
* gui to drive calibration, capture process, and to display 3d point results
* may use c, c++. python is fine only for prototyping
* must run in linux. Mac OSX or Windows optional. Ideally Ubuntu 11.04
* delivery of full source. site all references
similar to [login to view URL] except no skeleton solver, uses off the shelf components.
I'm a senior engineer with 15 years experience in gaming and motion capture. I will provide the skeleton solver. I wish for you to solve the calibration, capture, and output of 3d points.
It would greatly increase my confidence in your bid if you can solve the problem:
given 8 3d points:
(0, 2, 0), (36, 2, 0), (36, 58, 0), (0, 58 ,0), (0, 0, 36), (0, 0, 72), (36, 0, 72), (36, 0, 36)
are projected onto an 640x480 image in the pixel locations:
(255, 327), (372, 342), (303, 446), (137, 413), (255, 206), (254, 79), (378, 74), (377, 212)
what is the approximate 3d translation of the camera? +- 12 units. Assume a perfect camera with no distortion.
I will attach a visual aid in the form of an image that provided the data.
10 freelancers are bidding on average $3710 for this job
Hello, I have experience in software development under/for Ubuntu OS in C/C++/Qt. I have academical background;dev. setup at my place. Please see PMB for additional details/proposition. Thank you.