基于计算机视觉的非朗伯表面三维重构

3D reconstruction of non-Lambertian surfaces based on computer vision

基于计算机视觉的三维重构方法已经广泛应用在各行各业中。目前的三维重构研究主要针对不透明的朗伯表面,且已经比较成熟,但对非朗伯表面仍然面临诸多问题。而实际场景中的物体表面大多是非朗伯表面,因而,随着实际应用的推广,非朗伯表面的三维重构问题在计算机视觉领域越来越受到关注。虽然本现状研究不能完全涵盖针对非朗伯表面三维重构的所有方法,但它包涵了三维重构每个步骤中的各种典型方法。文中按照图像获取过程中的照明方式和重构原理对现有方法进行了分类,并逐类进行了介绍。由于不存在公共测试网络平台和带有标准视差的非朗伯表面立体图像集,因而,很难对各种算法的计算效率和匹配质量进行比较,文中主要对非朗伯表面的现有三维重构方法的原理、特点、适用范围和最新研究方向进行了介绍,对非朗伯表面三维重构的现有问题和发展前景进行了讨论。

3D reconstruction method based on computer vision is widely used in various industries.Though the 3D reconstruction of opaque surfaces with Lambertian reflectance on which the current research is mainly concentrated is a well-studied issue,the non-Lambertian surfaces pose challenging problems for the 3D reconstruction.Unfortunately,most surfaces of the objects in the actual scene are non Lambertian,therefore,a growing body of recent work has concentrated on the real-world problems,and has been concerned primarily with non-Lambertian surfaces reconstruction from images in the field of computer vision.Although this state of the art report does not cover exhaustively the generic methods of 3D reconstruction for non-Lambertian surfaces,it does contain examples of all the significantly different approaches to the steps in this field.The works discussed in the survey are categorized and reviewed in order.The main categories of classification are defined in terms of the illumination used for capturing the image and the principle based on.Due to lack of the common test bench available on the web and stereo data sets of non-Lambertian surfaces with ground truth,it is difficult to compare the computational efficiency and matching quality of the various schemes.In that case,here we are concerned with only the principle and characteristics of the existing methods,the scope of application and the latest trends in the research.We conclude the survey by discussing open problems and future perspectives and hope to provide an entry point to this existing field.