A fast 3D reconstruction method based on curve segment of binocular vision

Abstract

A fast 3D reconstruction method based on piecewise quadratic curve is proposed, aiming at the low efficiency of conventional methods. Firstly, a multi-line structured light binocular vision measurement system was adopted to obtain the two-dimensional image. Through the two-dimensional image, the centre line and feature points of the laser strip are fast acquired. Then, the acquired feature points were used as the given data points of curve interpolation, and then the NURBS curves are used to interpolate the feature lines of the image. According to the perspective invariance principle of plane curve, the zero curvature points and cusp points on the curve were detected as segmentation points. Using the segmentation results, several points on each NURBS curve segment were taken to fit as the quadratic curve. Finally, the three-dimensional reconstruction of the feature lines was accomplished via intersection of the right and left space straight taper surfaces. Taking the black shoe sole as the experimental object, the results show that the method proposed in this paper can fast complete the three-dimensional reconstruction of the inner and outer feature lines of the shoe sole, which can provide the basis and guarantee for the follow-up shoe sole in-process monitoring and quality inspection.