3D data acquisition using active stereo based on spatial neighbourhood technique

Active stereo using spatial neighbourhood of a structured light pattern is a well-known technique for acquiring 3D information of an object or scene. By projecting a pattern onto a targeted object and taking the image, the structured light is measured as coded words and the corresponding points between a pair of stereo images can be found. Finally, the 3D structure of the object is reconstructed using the principle of triangulation. Most existing techniques require a large number of colors in order to get a large number of stripes and avoid repeated coded words. However, the limitation of a DLP projector and a camera that produce incorrect colors, so called the crosstalk problem, may become a drawback. An incorrect codeword may occur and cause the problem on codification process. This paper aims to reduce the number of colors to minimize the crosstalk effect while generating a maximum number of stripes using the De Bruijn Hybrid pattern. By selecting the suitable colors and intensities that clearly match to projector and camera used, 3D information can be acquired. Dynamic and static objects were tested with peak-based, edge-based strategies.