3D vision methods and selected experiences in micro and macro applications H. Hügli, J. Mure-Dubois Institute of microtechnology, University of Neuchâtel, Rue Breguet 2, CH-2000 Neuchâtel, Switzerland ABSTRACT The paper provides considerations relative to the application of 3D vision methods and presents some lessons learnt in this respect by presenting four 3D vision tasks and discussing the selection of vision sensing devices meant to solving the task. After a short reminder of 3D vision methods of interest for optical range imaging for microvision and macrovision applications, the paper enumerates and comments some aspects which contribute to find a good solution. Then, it presents and discusses the four following tasks: 3D sensing for people surveillance, measurement of stamping burrs, sorting burred stamping parts and finally, hole filling algorithm. Keywords: machine vision, optical range imaging, 3D vision, 3D sensing, vision applications, microvision, vision software, depth from focus, time-of-flight, interferometry 1. INTRODUCTION The world in which we are living is three-dimensional and to evolve in it, most animals have developed means to sense their environment. Thus nature has developed different types of 3D vision systems, most based on optical sensing but some also using ranging by acoustic or electric signals. The predominant method, optical triangulation is found implemented in several flavors like stereovision, optical flow perception, parallax viewing, etc. Nowadays, non-contact sensing can be performed by using different bands of energy waves. Ultrasonic waves are used in sonar systems and microwaves represent the active signal used for synthetic aperture radar (SAR) imaging. On the other side of the electromagnetic spectrum, x-rays are used for imaging the inside of bodies, producing directly classical x-ray images or providing the input for the reconstruction of 3D images by computed tomography (CT). Surrounded by x-rays and microwaves on the electromagnetic spectrum, light waves are finally the dominant support for 3D sensing. A large number of techniques are available for the purpose of measuring 3D surfaces by optical means and there is not a unique way to classify them. A taxonomy proposed in 1 is close to the domains of application in the sense that it puts some needs of the user in the foreground. The main division is the classification in active and passive systems. The active systems are then divided according to basic physical differences, namely in triangulation and time-of-flight methods. Passive methods on the other hand, all based on geometry, are divided into several classes (stereo, silhouettes, shading, photogrammetry, focus) that differ basically in the reconstruction method they apply. A different taxonomy 2 is proposed to address the performance analysis and physical limitations and is therefore much closer to the physics of light. It proposes a division into triangulation, interferometry and time-of-flight. From the application point of view, often the choice of the adequate 3D vision method is not obvious, not to mention the choice of a specific vision system. Each method comes with a number of specificities that suit the application more or less and it is not always clear a priori what the right perspectives or compromises are. This paper is devoted to the presentation of work carried on in the perspective of matching 3D vision methods and systems with the requirements of some given tasks. Next section provides a reminder of main 3D vision methods of interest for macro and micro range imaging. Next follows a section giving some considerations for applying them. To illustrate the work in a more specific way, we present in a final section, a selection of four specific 3D vision tasks we investigated in the past. 2. 3D VISION METHODS This section provides a reminder of main 3D vision methods of interest for macro and micro range imaging. Published in Two- and Three-Dimensional Methods for Inspection and Metrology IV (Proceedings of SPIE) 6382, issue 10, 638209- 638216, 2006 which should be used for any reference to this work 1