Laser Triangulation for shape Measurement

by Emanuele Riposi (for his Bachelor thesis)

The aim of this work is to reconstruct glass objects profile using a Laser triangulation sensor.

In this kind of sensor, a laser beam emitted from the semiconductor laser is applied to the target. The light reflected from the target is collected by the receiver lens and focused on the light-receiving element, in this case a CCD (Charge-Coupled Device). It’s possible to measure the distance to the target when it changes, at example when the object is moving: the position of the reflected light passing through the receiver lens changes, and the light is focused at a different portion of the light-receiving element.

If the object is moving (for example on a sled or conveyor belt) and the sensor is set on a fixed position, it’s possible to scan object’s profile. The configuration used for tests is the following:

The micrometric motorized sled allows the longitudinal movement, that can be set (speed, covered distance and step) thanks to the moving control software; transverse movement is manually set using a micrometric sled and it allows to shift the object for scanning profile throughout the perpendicular axis. Laser sensor is mounted in a fixed position on a plexiglass support structure.

Reflection of light can be divided into two types: specular and diffuse reflection. Specular reflection occurs when the angle of incidence is equal to the angle of reflection and the majority of the reflected light follows that path; in case of diffuse reflection, light reflects off the surface in every direction at varying intensities. Transparent objects, such as glass, have very little diffuse reflection, so most of the reflected light will travel the same path. For this reason, the sensor head is tilted, so specular reflection is allowed.

In case of glassware, using a Laser triangulation sensor for scanning profiles is necessary to consider the beam-material interaction: depending on type of glass, surface characteristics and sample geometry, the Laser return on CCD could be significantly affected.

If glass surface is reflective (for example a little LCD screen glass), the profile reconstruction is adherent to the original and do not present sharp discontinuities (outlayers):

On the contrary, when glassware is massive, or presents chamfers, fillets with low ray of curvature, holes or particular surface plots (like ornamental objects) the beam is distorted and the CCD acquisition is significantly affected:

In particular, scan presents outlayers and the measure of the distance is incorrect (the profile appears closer or more distant). A plot of the profile’s derivate shows a sharp discontinuity in correspondence of outlayers; these discontinuities are all equal to those found in the presence of vertical surfaces, but in some case they are more evident and they should not exist (like green-circled parts in previous image).

The reason for the incorrect measurement (and consequent outlayers) could be found looking at the amount of received light waveform available on Laser sensor control software; when sensor is acquiring, at example, the reference distance, a typical waveform print is:

There is only one peak and the lecture is unique. Using glassware, mostly when beam strikes chamfers, fillets, edges or a not well-reflective surface, waveform is very confused, presenting more peaks (2-3) or a distributed curve which presents peaks and other trends:

If more peaks are received, not always the correct one is detected: the “real” peak could be hidden among the others and the sensor cannot recognize him.

Sometimes happens that the beam is not reflected on CCD because it is diffuse by glass on over the room and outlayer occurs.

Some types of glasses, especially ones that presents bad surface reflection, let the beam pass completely through the material, detecting the reference distance directly. In case of inclined surfaces, scan could present steps rather than a continuous line. Farther, material used for this experimentation presents different properties when scan direction is moved. Repeating tests along several directions it is possible to identify where the profile scan is better.

Leave a Reply