RGB laser speckles based 3D profilometry

摘要

Laser speckle is becoming popular for encoding objects for 3D profile reconstructions. However, the measurement accuracy of laser speckles based stereophotogrammetry is often not high due to the subjective speckles formed on the sensor plane of the camera. In this letter, we propose a method to enhance the measurement accuracy by encoding the surface of the object using red, green, and blue (RGB) laser speckles which are generated when a combined RGB laser beam passes through a ground glass diffuser. A 3CCD color camera is utilized to capture the object from two different perspectives. Interestingly, there is no crosstalk between different channels. Hence, the proposed system is practically equivalent to three synchronized monochrome vision systems having separate red, green, and blue laser speckle projections. Correspondences located by the RGB speckle image correlation are more accurate, because the structured information in the color image is three times that in the monochrome image. Experimental results validate the higher accuracy of the proposed method against the conventional method using monochrome laser speckles. Moreover, the motion robustness of the proposed method is verified on the measurement of the moving objects.