Adaptive spatiotemporal structured light method for fast three-dimensional measurement

Abstract

We present a high-speed 3-D spatiotemporal shape measurement technique by means of structured light. Current methods use a constant number of images that do not take into account the available temporal continuity of the measured object. That is, they focus on acquiring and processing as quickly as possible a fixed number of images to solve for the correspondence problem and later obtain the 3-D shape by triangulation. The number of images used imposes the use of some spatial support. The major contribution of our research is a new spatiotemporal scheme that, depending on the object's movement, adaptively uses the maximum number of projected images consistent with the local temporal continuity, therefore solving the correspondence problem with the minimum possible spatial support for each position. This is achieved by the use of a hybrid color pattern composed of an analog sinusoidal periodic code in the red channel and a digital binary spatial code in the blue channel that is cyclically displaced. No subpixel calculation is used and it is possible to implement error correction strategies that make the method fast and reliable, enabling dynamic online 3-D measurement of objects in movement.