Posted: April 3, 2016
Single-camera 3D image reconstruction
Computer vision has become popular for observing remote environments, whether monitoring undersea habitats, or inspecting industrial equipment. Recovering 3D geometry of a scene using 2D images requires special camera calibration.
University of Alberta researchers have developed a new technology to calibrate and reconstruct underwater 3D images using a single camera, exploiting the triple wavelength dispersion phenomenon. This technology does not require a calibration object and can estimate camera parameters more accurately than any other existing method reported. Using a single camera instead of two cameras in stereo is less expensive and encounters fewer calibration hurdles.
Reconstructed 3D model of Folger Passage underwater seafloor observations.