Conference Paper

A line integration based method for depth recovery from surfacenormals

Dept. of Inf. Electron., Tsinghua Univ., Beijing
DOI: 10.1109/ICPR.1988.28301 Conference: Pattern Recognition, 1988., 9th International Conference on
Source: IEEE Xplore

ABSTRACT A method for constructing a depth map from surface normals is described. In this depth recovery method, an arbitrary depth must first be preset for a point somewhere in the image, and then path-independent line integrals are computed to get the relative depths at every point in the image. The validity of the proposed method is discussed and its efficiency is tested using surface normals obtained by shape from the shading algorithm. A comparison to previous methods is made. Theoretical analysis and experimental results show that the present method is both powerful and easy to implement

0 Bookmarks
 · 
142 Views
  • [Show abstract] [Hide abstract]
    ABSTRACT: This paper addresses the issue of regularization in the surface reconstruction from gradients problem in Industrial Photometric Stereo. Regularization of the solution is a necessary step in an industrial environment, where algorithms must cope with non-Gaussian noise, such as outliers, or non-Lambertian textures such as corrosion. Introducing Tikhonov regularization into the global least squares solution suppresses the influence of outliers in the reconstruction. Viable methods should both minimize a global least squares cost function and also introduce some form of regularization into the solution; state-of-the-art methods to this end are grossly inefficient and are severely limited in the size of surface they can reconstruct. We present a new algorithm which can reconstruct a surface of 1200×12001200×1200, (i.e., greater than 1 M-pixel) in a few seconds. This is orders of magnitude faster than state-of-the-art methods incorporating regularization, and hence presents the first method viable for regularized reconstructions in practical applications.
    Computers in Industry 12/2013; 64(9):1221–1228. · 1.46 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We introduce an example-based photometric stereo approach that does not require explicit reference objects. Instead, we use a robust multi-view stereo technique to create a partial reconstruction of the scene which serves as scene-intrinsic reference geometry. Similar to the standard approach, we then transfer normals from reconstructed to unreconstructed regions based on robust photometric matching. In contrast to traditional reference objects, the scene-intrinsic reference geometry is neither noise free nor does it necessarily contain all possible normal directions for given materials. We therefore propose several modifications that allow us to reconstruct high quality normal maps. During integration, we combine both normal and positional information yielding high quality reconstructions. We show results on several datasets including an example based on data solely collected from the Internet.
    Proceedings of the 11th European conference on Trends and Topics in Computer Vision - Volume Part II; 09/2010
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We describe a robust method for the recovery of the depth map (or height map) from a gradient map (or normal map) of a scene, such as would be obtained by photometric stereo or interferometry. Our method allows for uncertain or missing samples, which are often present in experimentally measured gradient maps, and also for sharp discontinuities in the scene’s depth, e.g. along object silhouette edges. By using a multi-scale approach, our integration algorithm achieves linear time and memory costs. A key feature of our method is the allowance for a given weight map that flags unreliable or missing gradient samples. We also describe several integration methods from the literature that are commonly used for this task. Based on theoretical analysis and tests with various synthetic and measured gradient maps, we argue that our algorithm is as accurate as the best existing methods, handling incomplete data and discontinuities, and is more efficient in time and memory usage, especially for large gradient maps.
    Computer Vision and Image Understanding 08/2012; 116(8):882–895. · 1.36 Impact Factor