Optical three-dimensional shape measurement based on structured light and a binocular vision system

Abstract

Three-dimensional shape measurement based on structured light is affected by two factors: the number of fringe patterns and the phase unwrapping process. Although one-shot technology can get the wrapped phase, it is not suitable for measuring complex surface. Moreover, phase unwrapping also affects measurement speed and accuracy. To overcome these problems, a two-dimensional wavelet transform with binocular vision system is proposed. Wavelet transform is used to get the wrapped phase based on the Morlet wavelet. In order to get a three-dimensional shape without phase unwrapping, a binocular vision system is used. The increase matching accuracy, the preliminary disparity, and the sub-pixel optimization are calculated, respectively. Based on the calibration parameters, three-dimensional information can be obtained directly from the wrapped phase. In addition, the average phase is calculated based on ambient pixels to confirm wrapped phase boundary. Experimental results demonstrate the feasibility and advantage of the proposed method. Compared with traditional methods, both measurement accuracy and measurement speed can be increased.