Abstracted workflow framework with a structure from motion application

Abstract

In scientific and engineering disciplines, from academia to industry, there is an increasing need for the development of custom software to perform experiments, construct systems, and develop products. The natural mindset initially is to shortcut and bypass all overhead and process rigor in order to obtain an immediate result for the problem at hand, with the misconception that the software will simply be thrown away at the end. In a majority of the cases, it turns out the software persists for many years, and likely ends up in production systems for which it was not initially intended. In the current study, a framework that can be used in both industry and academic applications mitigates underlying problems associated with developing scientific and engineering software. This results in software that is much more maintainable, documented, and usable by others, specifically allowing new users to extend capabilities of components already implemented in the framework. There is a multidisciplinary need in the fields of imaging science, computer science, and software engineering for a unified implementation model, which motivates the development of an abstracted software framework. Structure from motion (SfM) has been identified as one use case where the abstracted workflow framework can improve research efficiencies and eliminate implementation redundancies in scientific fields. The SfM process begins by obtaining 2D images of a scene from different perspectives. Features from the images are extracted and correspondences are established. This provides a sufficient amount of information to initialize the problem for fully automated processing. Transformations are established between views, and 3D points are established via triangulation algorithms. The parameters for the camera models for all views / images are solved through bundle adjustment, establishing a highly consistent point cloud. The initial sparse point cloud and camera matrices are used to generate a dense point cloud through patch based techniques or densification algorithms such as Semi-Global Matching (SGM). The point cloud can be iii I would like to thank my advisor, Dr. Harvey Rhody, for his willingness to take me on as a research student and develop a thesis topic that was well-suited for my interests and background. I was fortunate to have taken many courses with Dr. Rhody, and it was a pleasure getting to know him as my thesis advisor throughout the course of our research. His personal and professional advice have been invaluable to me. I'm always amazed at how often I discover the applicability and foresight of his abstractions in practice. Thanks to my committee members, Dr. Carl Salvaggio and Dr. Derek Walvoord. After taking my first class with Carl, it became obvious that he is one of the exceptional teachers at RIT. He truly cares about his students and puts a great deal of effort into lectures and class material. He is student-focused and goes out of his way to make sure they succeed. Additionally, I am very grateful not only to have Derek on...

Estimating Soil Displacement from Timber Extraction Trails in Steep Terrain: Application of an Unmanned Aircraft for 3D Modelling

“…Image processing was conducted in compliance with photogrammetric workflow presented in Rossi, et al [24] and Turner, et al [29]. We used Agisoft Photoscan Professional ® software which employs full SFM-Multi View Stereopsis and estimates internal optical parameters and spatial positioning of the respective cameras [30].…”

“…This makes it possible to determine the scene geometry, camera calibration, position and orientation from an unordered, overlapping collection of images [22]. It makes use of Scale Invariant Feature Transform (SIFT) which is a feature extractor that is applicable to UAV imagery due to its robustness to changes in rotation, scale and translation between images [23][24][25]. SFM has been successfully applied in reconstructing objects from images coming from different sources [26].…”