Improving Laser Triangulation Sensors Using Polarization

J.Clark, E. Trucco and H-F. Cheung

International Conference on Computer Vision, 1995

Summary

  The paper is supporting the robustness of polarization-based vision addressing the robustness of laser triangulation range sensors. Basically these sensors are made of detecting a point or a line of pattern projected by a laser and not rarely there is the problem of highly specularly reflective surfaces, which cause reflection of the light, confusing the detection algorithms and mislead the measurements.

One possible solution, but not always applicable in an industrial environment, may be spray the components with matte pain, but basically it would be better to construct a system capable of distinguishing the primary reflection from the subsequent reflections.

The case studies is a combination of linear polarized (where light can be discussed only considering the electric field component) and unpolarized light (for which values of the electric field can only be determined statistically) and this partially linearly polarized light is described by the transmitted radiance sinusoid (TRS).

The fact that a change in orientation of the linearly polarized light changes upon specular reflection on metal (for which the example is taken), makes it possible to separate the inter-reflection from primary reflection; the amount of change is strictly dependent on the angle of incidence ψ.

The laser stripe can be observed by the CCD camera since laser light is reflected off planar microfacets created by the surface roughness, whose normal lie in the plane define by the camera’s viewing direction and the ray laser light itself.

This means that we can place the plan of laser light and the camera viewing direction at an optimal angle which may ensure, for significant orientation changes, independency from the surface normal.

The assumptions done have been tested using a CCD camera sensor, a laser and an infrared filter to drop wavelengths beyond 750 nm, at which light wouldn’t be efficiently polarized (at the CCD camera sensor).

Key Concepts

Distance measurement, Laser triangulation, Computer Vision

Key Results

The experiment shows that there is a wide application, since it captures a physical behavior of reflected light,

and reliability. It is not dependent on geometries of the observed surfaces.

Possible application fields are laser scanners, 3D industrial component’s design, medical, sporting, art products, video and film production, all cases where highly reflective specular surfaces can cause problems.

For the future the authors may implement a liquid crystal polarization camera, commercially available, which is more precise and would allow the avoidance of manual set up. Other kind of material should be investigated.