3D scanning of objects by projection of three incoherent RGB lines with Gaussian profile

Abstract

The projection of one light line is a widely used technique for the 3D scanning of objects; however, its main disadvantage is the large number of stored images and the time it takes to captured them. In this work, it is proposed 3D scanning of objects with colored surface by projecting of three incoherent RGB lines with Gaussian profile which is generated numerically using a multimedia projector. Each color line (red, green and blue) simultaneously scans a third of the object to be reconstructed thus reducing the number of stored images, favoring the optimization of the capture time compared to the projection technique of one light line. To carry out our proposal, algorithms of light line segmentation by color and algorithms for the extraction of the skeleton with sub-pixel resolution based on Bezier polynomials were implemented. The height measurements obtained on a test object were compared with a coordinate measuring machine (CMM); an average accuracy error of 0.2192 mm was achieved. The good results obtained in the 3D scanning of different objects support our conclusions.