Andrew E. Johnson

Carnegie Mellon University, Pittsburgh, Pennsylvania, United States

Are you Andrew E. Johnson?

Claim your profile

Publications (6)0 Total impact

  • Source
    Andrew E. Johnson, Martial Hebert
    [Show abstract] [Hide abstract]
    ABSTRACT: By combining techniques from geometric hashing and structural indexing, we have developed a new representation for recognition of free-form objects from three dimensional data. The representation comprises descriptive spin-images associated with each oriented point on the surface of an object. Constructed using single point bases, spin-images are data level shape descriptions that are used for efficient matching of oriented points. During recognition, scene spin-images are indexed into a stack of model spin-images to establish point correspondences between a model object and scene data. Given oriented point correspondences, a rigid transformation that maps the model into the scene is calculated and then refined and verified using a modified iterative closest point algorithm. Indexing of oriented points bridges the gap between recognition by global properties and feature based recognition without resorting to error-prone segmentation or feature extraction. It requires no kno...
    11/1999;
  • Article: Unknown
    Andrew E. Johnson, Martial Hebert
    [Show abstract] [Hide abstract]
    ABSTRACT: A common representation in 3-D computer vision is the polygonal surface mesh because meshes can model objects of arbitrary shape and are easily constructed from sensed 3-D data. The resolution of a surface mesh is the overall spacing between vertices that comprise the mesh. Because sensed 3-D points are often unevenly distributed, the resolution of a surface mesh is often poorly defined. We present an algorithm that transforms a mesh with an uneven spacing between vertices into a mesh with a more even spacing between vertices, thus improving its definition of resolution. In addition, we show how the algorithm can be used to control the resolution of surface meshes, making them amenable to multi-resolution approaches in computer vision.
    11/1999;
  • Andrew E. Johnson, Martial Hebert
    [Show abstract] [Hide abstract]
    ABSTRACT: We present a 3-D shape-based object recognition system for simultaneous recognition of multiple objects in scenes containing clutter and occlusion. Recognition is based on matching surfaces by matching points using the spin-image representation. The spin-image is a data level shape descriptor that is used to match surfaces represented as surface meshes. We present a compression scheme for spin-images that results in efficient multiple object recognition which we verify with results showing the simultaneous recognition of multiple objects from a library of 20 models. Furthermore, we demonstrate the robust performance of recognition in the presence of clutter and occlusion through analysis of recognition trials on 100 scenes. This research was performed at Carnegie Mellon University and was supported by the US Department of Energy under contract DE-AC21-92MC29104. 1 1 Introduction Surface matching is a technique from 3-D computer vision that has many applications in the area of robot...
    08/1998;
  • Source
    Andrew E. Johnson, Raul E. Valdes-Perez
    [Show abstract] [Hide abstract]
    ABSTRACT: The quantitative description of cell structures in light microscope images is an important task in biological research. The inclusion of digital image processing techniques and fluorescent markers into light microscope imaging has recently made this task feasible. In this paper, we present a method for detection of filamentary structures in cell images that have been highlighted with fluorescent markers. The method has three stages. First, pixels belonging to fiber contours are extracted from the image. Then these pixels are grouped together based on proximity by a minimal spanning tree. Finally, the fiber contours are determined as sub-trees of the minimal spanning tree. Once the pixels belonging to individual fibers contours have been determined, quantitative statistics describing the fibers in the cell can be calculated.
    09/1993;
  • Source
    Andrew E. Johnson, E. Vald
  • Source
    Andrew E. Johnson, Martial Hebert
    [Show abstract] [Hide abstract]
    ABSTRACT: Abstract We propose an algorithm for the reconstruction of elevation and materialproperty maps of the seafloor using a sidescan sonar backscatter image and sparse bathymetric points co-registered within the image. Given a path for the sensor; the reconstruction is corrected for the movement of the fish during the image generation process. To perform reconstruction, an arbitrary but computable scattering model is assumed for the seapoor backscatterer: The algorithm uses the sparse bathymetric data to generate an initial estimate for the elevation map which is then iteratively refied to jt the backscatter image by minimizing a global error functional. Concurrently, the parameters of the scattering model are determined on a coarse grid in the image byjtting the assumed scattering model to the backscaner data. The elevation surface and the scattering parameter maps converge to their best fit shape and values given the backscatter data. The reconstruction is corrected for the movement of the sensor by initially doing local reconstructions in sensor coordinates and then transforming the local reconstructions to a global coordinate system and performing the reconstruction again. The algorithm supports different scattering models, so it can be applied to different underwater environments and sonar sensors. In addition to the elevation map of the seafloor, the parameters of the scattering model at every point in the image are generated. Since these parameters describe material properties of the seafloor; the maps of the scattering model parameters can be used to seg-