Proceedings Article | 18 December 2019
KEYWORDS: Speckle, Sensors, Speckle pattern, Spatial light modulators, Digital holography, Stochastic processes, Image sensors, Holograms, Diffraction
Measurement systems based on laser spot triangulation are widely used in many fields of distance-, surface and 3D measurements. Currently, strong interest in applications like laser-range detection (LIDAR) in automotive driving and robotics also promote further research in improving the technique. Laser-based ranging and triangulation techniques suffer from the great advantage of one of the lasers properties - its coherence. A coherent laser light ray allows projecting a spatially small spot onto an object compared to a broadly illuminating LED or bulb. A detector, e.g. camera or photodiode, detects the reflected measurement spot precisely with its position on the object. The downside of the coherence appears in combination with a rough reflecting surface as speckles. Speckle is the noise of the spot due to the rough object surface and the resulting random phases. In some measurement settings, the speckle noise has a large negative effect and influence on the measurement uncertainty to determine the precise center-of-gravity of the measurement-spot's position. In this contribution, we investigate a method introduced by us in an experiment to reduce the speckle noise and therefore improve the measurement uncertainty in triangulation. The proposed multi-spot-scanning method decorrelates speckle noise patterns through dynamical holography of the laser-spot by varying the position microscopically. For the dynamical scanning of the spot, we use a spatial light modulator (SLM) being addressed with digital holograms. SLMs are part of many measurement systems and other optical applications, especially projection-based systems. The holograms, illuminated by the laser, microscopically translate the measurement spot position in the object plane. These so-called "miicro-variations" are in fact translations of the first diffraction order using a grating on the SLM with additional overlaid Zernike-polynomials. The grating has a carrier frequency calculated according to the SLM's pixel size. Due to the micro-variations, the speckle noise patterns vary and decorrelate. Integrating or averaging over many patterns on the camera then allows predicting the spot's position with higher certainty through its better uniformity. Additionally, we examined how well the speckle decorrelation affects the uncertainty reduction especially in a triangulation sensor setup. Our experiments, utilizing the multi-spot-scanning method, showed promising first results. The microvariations in our experiments corresponded well to the simulations. Using different Zernike-coefficients, we are able to control the spot position to roughly a couple of micrometers while having a maximum field size for micro-variation of up to 16 mm in diameter. The best micro-variation-spot radius was between 75 μm and 85 μm considering a spot size of FWHM = 53 μm. We also found out, that with the method we could decorrelate speckle patterns to a certain degree. This degree can reduce the noise while averaging by a factor of around 40% compared to the original noise. Overall, the statistical measurement uncertainty improved by the reduction ratio of approximately R = 3 in axial direction. Further investigations on more spot-variations (7-42 spots) lead to solely slight improvements, because the denser the spots are, more overlap is present and less microvariation and decorrelation is possible. Furthermore, we have investigated different Zernike-polynomials (coma, defocus and astigmatism) and their influence on combined variations. In conclusion, we describe a method for laser-triangulation sensor systems to reduce laser speckle noise and improve measurement uncertainty. With this method, we can increase the axial resolution of the triangulation with the disadvantage of losing a small amount of lateral resolution due to the averaging and variation of the spot. We hope to be able to also expand our findings to other scenarios and setups as well as improve the findings with further research.
