Robust 3D object recognition using complex amplitude information by holographic multiplexing

Abstract

We propose robust three-dimensional (3D) object recognition method using a complex amplitude based on the Fourier transform profilometry (FTP) and the numerical holographic multiplexing. In the proposed method, a reference data for correlation operation is produced by multiplexing of the complex amplitude derived from several rotated objects, taking into account that the complex amplitude derived from rotated objects is mutually uncorrelated with respect to the rotation angle. Due to the use of the multiplex reference data, the correlation between each reference data to be multiplexed and a target object is performed with single operation. Therefore, the proposed method has tolerance to rotation in addition to translation due to the shift-invariant property of the FTP-based correlator. The spatial position and the orientation of the 3D object can be also evaluated by the proposed method. Experimental results show that the proposed method successfully recognizes the rotated 3D object.