The Color-Code Technique for 3D Scanning

We present a new method of image coding in structured light 3D scanning called Color-Coding. The method reduces the number of projected patterns needed to generate the code used to retrieve depths of an object being scanned to only three slides, each one containg RGB values that identify the vertical image stripes correctly. The technique converts the RGB values to the well-known Gray-Code algorithm values, making easier and faster the scanning process. Acquiring three dimensional shapes by computer is becoming commercially important and among the enormously diversified methodologies for this purpose, active ilumination methods are a reasonable alternative concerning to both robustness and cost. Using color-code instead of simple gray-code methods allow us to greatly reduce the number of projections needed to retrive depths of an scanned object, making it a viable technique for real-time scanning. In this work we developed a color-code technique that converts RGB values of the images containing the scanned object to the well-known Gray Code algorithms values. This procedure not only makes the scanning process faster but also mantains its robustness. Due to the inherent problem of blending created by the low-cost acquisition system used and since the object own color affects the obtained pattern coding, appropriated techniques had to be applied in order to correctly identify the projected stripes. a) Both the colors projected and the thresholding process used were customized for the object to be reconstructed. Complementary colors to those of the object are preferred. b) Background and object codes are non-decreasing functions along an image line. Therefore, filters had to be designed to ensure this property holds. c) Artifacts created by the quantization process are also removed by suitable filtering. We used an Epson 50c Power Lite multimedia projector with maximum resolution of 800X600 and a Sony Handycam NTSC camera. The images where captured by the camera and sent to the computer (Pentium II, 500Hz) using a 640X480 frame-grabber. As can be seen in Figure 1, the results are acceptable, specially concerning the global shape acquisition and it’s precision in the X and Yaxis. An experimental result is presented below. Figure 1 The pattern and the reconstructed head. References[1] Rocchini, C. , Cignoni, P. , Montani, C. , Pingi, P. andScopigno, R.: “A low cost 3d scanner based on structuredlight”; Computer Graphics Forum; vol. 20, n. 3, 2001.[2] Rodella, R. and Sansoni, G.: “Fast digitization ofheritage objects by means of a 3D vision system based onthe projection of structured light”; Canada-ItalyWorkshopHeritage Applications of 3D Digital Imaging, Ottawa,Canada;1999.[3] Hall-Holt, O. and Rusinkiewicz, S.: “Stripe boundarycodes for real-time structured light range scanning ofmoving objects”, 8 IEEE. Int. Conf. on Comp. Vision;vol.II.; 359-366; 2001.[4] Davies, C. and Nixon, M.: “Sensing surfacediscontinuities via coloured spots”; Proc. IWISP,1996. Proceedings of the XV Brazilian Symposium on Computer Graphics and Image Processing (SIBGRAPI’02)1530-1834/02 $17.00 © 2002 IEEE