Your Location Within Site:
Machine Vision
IRL’s key research interests in the field of machine vision include the rapid creation of 3D content for objects and scenes; the development of calibration techniques for imaging equipment; and 3D sensing and processing techniques.
3D content creation
We are conducting research into the rapid creation of 3D content for objects and scenes - in particular untextured 3D models and photo-realistic 3D models - focusing on techniques that can capture complex scenes and geometry that are not well catered for by existing technology. Applications include engineering refits, as-built surveying, art, heritage, games, movies, forensics, accidents, advertising, bioengineering. Read about our scene-scanner device, a fast hand-held 3D scanner.

A multi-camera rig
Calibration
We have experience developing calibration techniques for a wide range of imaging equipment. Some examples are:
- flexible camera intrinsic calibration using multiple views of a 2D calibration target.
- Camera lens distortion calibration from images based on "straightening" distorted lines.
- Robust calibration of the relative poses of multiple cameras.
- Camera shading calibration to remove effects such as vignetting, lens
- Shading without requiring expensive flat field light sources.
- Laser Line scanner pose calibration.
- Rapid calibration of 3D position of multiple targets.
-
Accurate calibration of temporally encoded structured light systems.

An image from an uncalibrated camera
3D sensing/processing
We have a strong interest in 3D sensing and processing techniques such as pose estimation, triangulation, meshing, surface geometry estimation. We are interested in the application of Clifford algebra (Geometric algebra) in 3D processing and have developed a Matlab toolbox (GAMatlab) to assist with rapid prototyping of algorithms. See a case study on our timber scanning technology (timber measuring tool for sawmills).

