Home /Research /Design and calibration of a hybrid computer vision and structured light 3D imaging system
PERCEPTION

Design and calibration of a hybrid computer vision and structured light 3D imaging system

Tom Botterill, Steven Mills, Richard Green

Year
2011
Citations
8

Abstract

This paper describes a structured light system for generating a dense 3D reconstruction of crops viewed by a robot, as the robot moves. A line laser is used to project a plane of light into the image, which is viewed by a colour camera. The laser line is localised with subpixel accuracy despite variations in colour and texture of the surface being imaged, and challenging lighting conditions. A dense depth map with errors from laser line localisation of just 0.15mm is constructed. A simple calibration procedure is described, where a large number of constraints are automatically extracted from images showing where the laser line crosses a checkerboard pattern. Errors in calibration are estimated at 2.3mm.

Keywords

Subpixel renderingComputer visionStructured lightArtificial intelligenceCalibrationLine (geometry)Computer scienceLaserCamera resectioningRobot

Related papers

Browse all PERCEPTION papers