Reconstructing partially visible models using stereo vision, structured light, and the g2o framework
Proceedings of the 27th Conference on Image and Vision Computing New Zealand
| Hi-index | 0.00 |
Machine vision lighting systems must be designed to evenly illuminate scenes, so that objects' appearances do not change depending on their position. This is challenging when imaging 3D scenes where the camera's field of view or depth of field are large. This paper describes a scheme to design effective lighting systems in such situations: lighting across the scene is modelled, and positions of lights are optimised to minimise variation in illumination levels. The scheme is used to design a lighting system for an agricultural robot, which reduces variation in the appearance of crops due to uneven lighting. The lighting for a compact medical imaging device is redesigned, which will result in a 53% reduction in the range of lighting levels in the scene.