Optical 3D imaging using VCSEL-based speckle scanning projection

Abstract

Speckle projection profilometry, as a non-contact 3D measurement technique, has been extensively employed in consumer electronics and 3D facial recognition systems due to its structurally simple configuration and high 3D measurement efficiency using a single-frame speckle patterns. However, when dealing with object surfaces with complex reflection characteristics, it remains challenging for single-frame speckle projection methods to obtain stable and reliable matching results. To address this issue, we propose an optical 3D imaging method using VCSEL-based speckle scanning projection. Firstly, a spatial-temporal speckle matching method based on stereo digital image correlation (DIC) is introduced. The initial disparity maps of spatial-temporal speckle images are obtained through spatial-temporal matching based on a zero-mean normalized cross-correlation (ZNCC), followed by sub-pixel disparity optimization integrated with the stereo DIC matching algorithm. Subsequently, an unsupervised spatial-temporal speckle matching network is further proposed to iteratively optimize and enhance the precision of disparity estimation and the quality of 3D reconstruction. Experimental results confirm that the proposed method achieves high-speed and high-precision 3D imaging for large-scale measured objects.