基于参数域映射及B样条插值的三维重构方法

利用单目CCD图像进行物体表面非接触测量的核心是基于单幅图像的三维重构技术,常采用由阴影恢复形状(SFS)的方法实现三维重构。当图像分辨率较低时,通过由阴影恢复形状的方法重构的三维表面模型其分辨力较差,无法满足实际要求。为此提出了一种基于参数域映射及B样条插值的三维重构方法。采用B样条插值技术对图像进行放大处理,通过像素的参数域映射减小图像的失真及高频信息的丢失,根据放大后图像的灰度信息重构物体的三维表面模型。实验表明,基于参数域映射及B样条技术的图像插值方法很好地保护了图像的细节,利用该方法进行三维重构能够有效改善重构模型的分辨力和光顺性,为提高三维表面非接触测量精度创造了条件。

Three-Dimensional Reconstruction Method Based on Parameter Mapping and B-Spline Interpolation

The key technology of single CCD image based non-contact measurement is three-dimensional (3D) reconstruction. The 3D surface model is usually obtained directly through shape from shading (SFS) method. But when the resolution of an input image is low, the resolution of the 3D surface model reconstructed by the above way would be poor and can not meet practical demmand. To resolve this problem, an improved 3D reconstruction method is proposed. First, the B-spline interpolation method is used to improve the image resolution, and the parameter mapping method of pixels is applied to avoid the image distortion and high frequency missing; then the SFS method is adopted to reconstruct the amplified image according to the gray gradient of image. The experimental results demonstrate that the proposed method can conserve the detail of an image, and improve the resolution and smoothness of the reconstruction model effectively. It will help improve the precision of non-contact measurement based on single CCD image.