Turning Corners into Cameras:         

Principles and Methods

Katherine L. Bouman,       Vickie Ye,      Adam B. Yedidia,      Fredo Durand,
     Gregory W. Wornell       Antonio Torralba       William T. Freeman

• Home
• ICCV Paper PDF
• Code & Dataset

We construct a 1-D video of an obscured scene using RGB video taken with a consumer camera. The stylized diagram in (a) shows a typical scenario: two people—one wearing red and the other blue—are hidden from the camera’s view by a wall. Only the region shaded in yellow is visible to the camera. To an observer walking around the occluding edge (along the magenta arrow), light from different parts of the hidden scene becomes visible at different angles (see sequence (b)). Ultimately, this scene information is captured in the intensity and color of light reflected from the corresponding patch of ground near the corner. Although these subtle irradiance variations are invisible to the naked eye (c), they can be extracted and interpreted from a camera position from which the entire obscured scene is hidden from view. Image (d) visualizes these subtle variations in the highlighted corner region. We use temporal frames of these radiance variations on the ground to construct a 1-D video of motion evolution in the hidden scene. Specifically, (e) shows the trajectories over time that specify the angular position of hidden red and blue subjects illuminated by a diffuse light.

Code

Matlab and C++ code

To run a demo of corner cameras download the data below and run test_corner.m in the 'matlab' directory. Make sure to change the 'datafolder' path to point to the downloaded example_video directory. To change which videos are being processed select a different parameter file under the comment "select the videos to process below".

To process data in a stereo corner camera setup run test_stereo.m. Again, make sure to change the 'datafolder' path to point to the downloaded example_video directory.

The code and executables are provided for non-commercial research purposes only. Contact the authors if you wish to use the code commercially.

Data

Download sample videos that can be used with the above code to recover the motion of people behind the corner. We include videos taken both indoors and outdoors. We have included parameters that can be used for the calibration step (homography and corner selection). To run your own calibration simply remove the .mat files from the video directories.

Acknowledgements

This work was supported in part by the DARPA REVEAL Program under Contract No. HR0011- 16-C-0030, NSF Grant 1212849, Shell Research, and an NDSEG Fellowship (to ABY). We thank Yoav Schechner, Jeff Shapiro, Franco Wang, and Vivek Goyal for helpful discussions.