Akademik Veri Yönetim Sistemi
COMPUTER JOURNAL, cilt.54, sa.5, ss.738-753, 2011 (SCI-Expanded)
Many industrial applications use object range data. Therefore, various range scanners, based on different working principles are developed. Among these, scanners using laser stripe-based triangulation are the most promising ones. Unfortunately, obtaining the range data of shiny objects using these scanners is problematic under ambient light. For specular objects, the laser stripe causes multiple reflections on the object. Even if this problem is solved, the ambient light also affects the scanning problem by introducing highlights on the (specular or shiny) object surface. To solve this problem, we develop two stripe-based scanners in this study. Our first scanner is based on line laser and mechanical rotation parts. Our second scanner is based on a projection machine (providing multicolor line stripes). For both systems, our main contribution is eliminating the affect of highlights (originating from ambient light) in detecting the line stripe (from line laser or projection machine) in range data extraction. To do so, we introduce a novel color invariant. We test and report the performance of our novel stripe detection method under various controlled experiments. We also test and report the overall range data extraction performance of our scanners on several shiny and matte objects. These objects have different color, texture and shape characteristics. We compare our range scanners with a commercial one as a benchmark. Extensive testings indicate the success of our range scanners, especially for scanning shiny surfaces under ambient light.