A Robust Depth Imaging Sensor Based on Signal Separation Coding

We present a depth imaging sensor based on signal separation coding. Since conventional approaches to structured light system depend on unrealistic assumptions about a workspace, such as absent of reflection or discontinuous, a depth image may collapse in a real home environment when the workspace has specular objects reflecting the active signal to the environment. In order to handle erroneous data caused by reflection and blurring, we used a two cameras and one projector system. The geometric constraint among them has been analyzed and applied to an error detection and correction procedure. The system achieves pixel correspondences not only within epipolar lines but also between epipolar lines efficiently. The experimental results have shown that the proposed system significantly enhances the robustness in depth imaging, compared to the conventional approaches. The proposed approach opens a greater feasibility of applying structured light based depth imaging in a cluttered workspace for home service robots.