A Novel Robust and Material Adaptive 3D Imaging System

Y.-H. Yeh, I.-C. Chang, C.-L. Huang, and W.J. Hsueh (Taiwan)

Keywords

Color structured light, 3D imaging, depth calculation, modeling.

Abstract

Color-structured-light based 3D Imaging techniques suffer from the problem of color shifting. When the color pattern is projected on the surface with complex reflectance, it is very difficult to recover correct color pattern. Previous works limited the number of colors in the color pattern to 3 or 4 in order to differentiate them easily. This causes another problem: low resolution, because it is not able to code ample color sequence. Our proposal increases the number of colors to 6, arranges as much as 512 vertical color bars in the color pattern, and achieves very high resolution (extract at most 60,000 3D points in a single image). We also have developed the triple-filter algorithm, which is used to recover the correct color pattern, even if it is seriously distorted by objects' reflectance. Experimental results show that our approach can measure skin, ceramic ware, stone, plastic commercial product, etc. It also proves that our proposal looks after both sides of high resolution and robustness.

Important Links:



Go Back