3D Full Field Measurement Based on Micro-geometrical Texture Patterns - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2006

3D Full Field Measurement Based on Micro-geometrical Texture Patterns

Résumé

The work presented is based on the laser scanning device that is now currently used in the reverse engineering or tri-dimensionnal inspection domains. The interest of this technique is to measure directly the point clouds in 3D coordinates. That permits to reduce the approximations due to the different correlations made in grey level images stereo-correlation. The sensor is mounted on a CMM that moves it and gives it a good orientation in the measurement space.The pattern is determined by the natural or artificially created micro-geometry (roughness) of the surface product. A depth image is obtained after filtering of the shape of the object. This filtering can be carried out by a least square method applied to analytic equations for simple objects (cylinder, cone, ...) or by NURBS approximation for complex parts. The depth image giving the micro-geometry is then regularised by B-spline interpolation to compare two configurations (initial and displaced) with classical method of correlation. Several constrains or limits exist: - The necessity to move the laser sensor to have scanning: here, we use a Coordinate Measurement Machine (CMM) that permits to reach a given position with a chosen orientation. The orientation is given by a motorised head PH10 whose uncertainty of repositioning is not negligible. - The limit of resolution: the density in the laser plane direction is given by the sensor maker (e.g. at best 24 points/mm). So, the roughness pitch doesn't be smaller than 0.08 mm in respect of Shannon theorem about sampling. In the case of machined mechanical part, this condition is not always established. In the orthogonal direction, the resolution is dependent on the motion of the MMT, so it can be controlled. - The measurement uncertainties: in direction (Laser axis see Fig 1), the uncertainty depends on several parameters notably the distance between the sensor and the measured surface, and the two incidences angles between the laser plane and the normal direction of the surface. In good conditions, it can be of the order of +/- 15 m with a 6  confidence but it can reach the double for some distances and orientation [6]. In the others directions, it cannot be easily quantifiable. - The signal-to-noise ratio (S/N): to have a good ratio, we have estimated that the roughness must be in the order of 60 m. however, it depends on the optical properties of the surface. - The speckle phenomenon: used to create the pattern in DSPI or in DSP, the speckle becomes harmful in the present case. It produces aberrant points which can be eliminated when they are far from the surface or which creates noise in the other cases. It will be all the more bothering when metallic objects will be considered and will have a great influence when the surface reflectivity will be high. It is possible to apply a spray that gives a matt surface and so reduces this phenomenon
Fichier non déposé

Dates et versions

hal-00803934 , version 1 (24-03-2013)

Identifiants

  • HAL Id : hal-00803934 , version 1

Citer

Jean François Fontaine, Halima Isheil Bubaker, Tadeus Sliwa, Christine Zimmer. 3D Full Field Measurement Based on Micro-geometrical Texture Patterns. Photomechanic's 2006, Jul 2006, CLERMONT FERRAND, France. ⟨hal-00803934⟩
188 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More