3-D Measurement Of Automotive Glass Using Reflective Fringes

There has been much interest in the automotive industry in developing non-contact techniques for measurement of reflective surfaces to provide in-line glass shape quality control system. Research work is focused on developing techniques for the measurement of non full-field reflective surfaces of automotive glass by using a reflective fringe pattern technique.

Physical properties of the measurement surfaces do not allow us to apply optical geometries used in existing techniques for surface measurement based upon direct fringe pattern illumination. However, this property of surface reflectivity can be used to implement similar ideas from existing techniques in a new improved method. In other words the reflective surface can be used as a mirror to reflect illuminated fringe patterns onto a screen behind. It has been found that in the case of implementing the reflective fringe technique, the phase shift distribution depends not only on the height of the object but also on the slope in each measurement point. This requires the solving of differential equations to find the surface slope and height distributions in the x and y directions and development of the additional height reconstruction algorithms.

The main focus has been made on developing a mathematical model of the optical sub-system and investigating ways for its practical implementation including calibration routines (camera calibration by TSAI technique, reference calibration, height calibration etc), and possible problems which may arise during real measurement processes. A video projector was used in the system with a resolution 1024x768 pixels. A progressive scan camera was applied for image recording. The digitisation resolution is 768x576 pixels. Figure 1 shows some key stages during the process of specular surface measurement. A piece of automotive glass (800 x 500mm) has been measured. A number of fringe patterns are grabbed shifted in both directions, see figure 1 (a) and (b) and the wrapped phase is calculated for the x and y directions, as shown in figures 1 (c) and (d). A masking algorithm separates the background area from the measurement object. The unwrapped phase is obtained by applying the unwrapping algorithm to the wrapped phase distribution. Calculated surface height map distribution and its scaled profile view are shown in figure 1 (e) and (f) respectively.