Tile coding based camera modeling for 3D sensing

Recently, the 3D sensing technique using optical devices such as multiple cameras has been applied to various areas such as visualization, motion capturing, and so on. However, improvement of the camera model structure is indispensible for more precise measurement. In this study, we propose a camera modeling technique for 3D sensing based on tile coding (CMAC) structure. A distance between a sensing target and the camera is used to construct the camera model considering optical projection characteristics. In our approach, the least mean square error method is successfully applied considering the simple tile structure to formulate the camera model. Then iterative calculations for solving the inverse problem of the 3-D to 2-D projection by camera are performed to attain measured 3D coordinates. Through 3D sensing experiments of stereo vision measurement based on the proposed approach, we showed the proposed model drastically improved the modeling performance compared with the conventional modeling approach such as Open CV model and crisp partitioned model.