Browsing by Subject "Photogrammetry"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Photogrammetry for 3D Reconstruction in SOLIDWORKS and its Applications in Industry
    (2019-08) Potabatti, Nikhil S.; El-Mounayri, Hazim; Chen, Jie; Anwar, Sohel
    Close range, image based photogrammetry and LIDAR laser scanning technique are commonly utilized methodologies to snap real objects.3D models of already existing model or parts can be reconstructed by laser scanning and photogrammetry. These 3D models can be useful in applications like quality inspection, reverse engineering. With these techniques, they have their merits and limitations. Though laser scanners have higher accuracy, they require higher initial investment. Close-range photogrammetry is known for its simplicity, versatility and e ective detection of complex surfaces and 3D measurement of parts. But photogrammetry techniques can be initiated with comparatively much lower initial cost with acceptable accuracy. Currently, many industries are using photogrammetry for reverse engineering, quality inspection purposes. But, for photogrammetric object reconstruction, they are using di erent softwares. Industrial researchers are using commercial/open source codes for reconstruction and another stand-alone software for reverse engineering and mesh deviation analysis. So the problem statement here for this thesis is to integrate Photogrammetry, reverse engineering and deviation analysis to make one state-of-the-art work ow. xx The objectives of this thesis are as follows: 1. Comparative study between available source codes and identify suitable and stable code for integration; understand the photogrammetry methodology of that particular code. 2. To create a taskpane add-in using API for Integration of selected photogrammetry methodology and facilitate methodology with parameters. 3. To demonstrate the photogrammetric work ow followed by a reverse engineering case studies to showcase the potential of integration. 4. Parametric study for number of images vs accuracy 5. Comparison of Scan results, photogrammetry results with actual CAD data
  • Thumbnail Image
    Item
    Registration and Localization of Unknown Moving Objects in Monocular SLAM
    (IEEE, 2022-03-23) Troutman, Blake; Tuceryan, Mihran; Computer and Information Science, School of Science
    Augmented reality (AR) applications require constant device localization, which is often fulfilled by visual simultaneous localization and mapping (SLAM). SLAM provides realtime camera localization by also dynamically building a 3D map of the environment, but the functionality of SLAM systems generally stops here. Useful applications of AR could make great use of additional information about the environment, such as the structure and location of moving objects in the scene (including objects that were not previously known to be separate from the static points of the map). We present an approach for solving the visual SLAM problem while also registering and localizing moving objects without prior knowledge of the objects’ structure, appearance, or existence. This is accomplished via analysis of reprojection errors and iterative use of the ePnP algorithm in a RANSAC scheme. The approach is demonstrated with the accompanying prototype system, LUMO-SLAM. The initial results achieved by this system indicate that the approach is both sound and potentially viable for some practical applications of AR and visual SLAM.