Simulation of Structured Light 3D Scanning using Blender

Many different optical setups and patterns are proposed for structured light 3D surface scanning. It is difficult and time consuming to experimentally assess all imaging options and to select the best one for a particular application. An efficient way of assessing imaging options is simulation of 3D scanning on a computer. We describe an efficient and low cost approach to simulating structured light scanning using Blender. Blender is an open source 3D computer graphics program primarily intended for 3D drawing and animation also supporting Phyton scripting to extend its capabilities. We describe how to simulate a simple scanning setup comprised of one camera, one projector, and one object. We also provide detailed description how to set both intrinsic and extrinsic parameters of projector and camera and discuss how to automate the simulation. To assess the correctness of the proposed simulation we simulate a real-world 3D surface scanner and we compare simulation results to the real-world results. We also perform experiments of scanning calibration objects. The proposed simulation method is correct and can be use to assess imaging geometry, to select structured light patterns, or to generate data for novel pattern decoders which use deep learning.