基于霍夫变换的结构光场3维成像方法

Structured light field 3-D imaging method based on Hough transform

为了在被测物体部分被遮挡的情况下,仍能完整地测量出物体的表面形貌,提出了一种基于霍夫变换的多线结构光标记的光场3维成像系统。通过提取亚像素的条纹中心坐标进行投票,分析霍夫参数空间中的投票分布,设计了自适应范围投票和自适应窗口策略,无需对条纹级次编码也能够准确地确定对极平面图像中多根直线的参数。结果表明,该系统拟合平面的平均偏差为0.0096 mm,标准偏差为0.0074 mm,并利用光场成像中不同视角下物体的遮挡关系不一致,准确地恢复了被遮挡物体的完整表面形貌。这一结果对于解决3维测量过程中遮挡问题是有帮助的,该研究为获取完整和高效3维数据的测量方法提供了参考。

In order to accurately recover the complete shape of the measured object with partial occlusion,a system for light field 3-D imaging based on multi-line structured light marking was proposed and a structured light field 3-D imaging method based on the Hough transform was proposed for the image processing problem in this system.The extracted sub-pixel fringe center coordinates were used to Hough transform,then the voting distribution in the Hough parameter space was analyzed,and adaptive range voting and adaptive window strategies were designed.Consequently,the parameters of multiple straight lines in the epipolar plane image can be accurately determined by the proposed method,which does not need to encode the fringe level.The experimental results show that the mean deviation and standard deviation of the fitted plane are 0.0096 mm and 0.0074 mm respectively.The complete shape of measured object is accurately obtained,which utilities the occlusion of different sub-aperture images is inconsistent in light field imaging.The result is helpful for solving the occlusion problem in the 3-D measurement process,and this work provides a reference for the measurement method to obtain complete and efficient 3-D data.