Shape Recovery from an Image Sequence and a Gyro Sensor

Our study is the recovery of 3D shape from an image sequence. For example, your favorite building can be reconstructed and seen from a new viewpoint in a computer. In order to improve accuracy, we attached an acceleration-gyro sensor to the camera and used sensor fusion technique.

Please see slides for explanation in English or in Japanese (0.9MBytes).

Please see our demo movie (5.4MBytes, mpeg format).

References:

  1. Toshiharu Mukai and Noboru Ohnishi: "The Recovery of Object Shape and Camera Motion Using a Sensing System with a Video Camera and a Gyro Sensor," Proceedings of the 7th IEEE International Conference on Computer Vision (ICCV'99), Kerkyra, Greece, September 20-27, 1999 (accepted). PostScript file (gzipped, 1.5Mbytes)

  2. Toshiharu Mukai and Noboru Ohnishi: "Sensor Fusion of a CCD Camera and an Acceleration-Gyro Sensor for the Recovery of Three-Dimensional Shape and Scale," Proceedings of the 2nd International Conference on Information Fusion (FUSION'99), pp.221-228, California, USA, July 6-8, 1999. PostScript file (gzipped, 486Kbytes)
[Picture
Our sensor system consists of the CCD camera and the acceleration-gyro sensor

[Original
An image from the original image sequence

[Recovered
The recovered structure seen from a new viewpoint

[Up]