基于梯度数据的三维面形重建方法

目前三维传感、机器视觉等领域的许多方法如阴影法、光度立体法,偏折术法等获得的数据为离散梯度.一些实际应用如波前重建、相位测量和三维面形测量等,需要进行面形重建.本文讨论了测得梯度的偏差对重建结果的影响,分析比较了不同偏差环境下基于局部积分技术的路径积分法与基于全局积分技术的傅立叶变换积分法及其修正方法的重建能力,计算机模拟实验证明了根据实际梯度偏差选择适当算法的必要性.

Shape reconstruction methods from gradient field

Many methods developed in 3D-sensor and machinevision such as Shape form shading,Photometric Stereo, Defleotometry result in discrete gradient. Shape reconstruction from gradient field is very necessary in some application fields, e. g. wavefront reconstruetion, phase unwrapping, 3D shape measurement. In this paper , the influence of gradient field deviation on reconstruction precision was discussed, and then the performances of the path integration based on local techniques, classical and modified Fourier transform integration based on global techniques were compared, The simulation result shows that it is very necessary to choose proper methed for shape reconstruction according to actual gradient field deviation.