Return to search

Single Camera Photogrammetry MATLAB Solver Developed for Automation of the Oil Interferometry Process

Over the last 20 years, Gregory G. Zilliac of the NASA AMES Research Center has been in continuous development of a fringe-imaging skin friction PC application used in oil interferometry analysis. This application, CXWIN5G, allows users to analyze propagation of oil smears across an aerodynamic surface using photogrammetry. The purpose of this thesis is to investigate the feasibility of increasing the level of automation currently found in CXWIN5G by developing a MATLAB solver capable of determining oil smear geometry with minimal user input.
There are two main automation goals of this thesis that are reflected in the core of the MATLAB solver: the determination of oil smear centerline propagation without user input and the calculation of fringe spacing without the use of fiduciary markings on the test surface. In CXWIN5G, oil smear propagation centerlines must be drawn by a user with their computer mouse. The MATLAB solver removes the necessity for this by utilizing the centroid location of each fringe as a reference for centerline propagation. The solver’s ability to calculate fringe spacing without the use of fiduciary markings is a result of its ability to accurately determine the physical dimensions captured in an image. This is done by separating the camera’s field of view into its pixel components and calculating the horizontal and vertical object length captured in each pixel.
Validation of the MATLAB solver’s ability to define fringe propagation and fringe spacing is performed at multiple different camera positions. When the camera location is not directly overhead an oil smear the camera is in a state of skew. Camera skew is measured in degrees, and can occur in the horizontal or vertical direction. Images analyzed in this thesis feature representative hand-drawn oil smears, as well as oil smears created in the Cal Poly 3’ x 4’ low speed wind tunnel.
The MATLAB solver’s ability to create accurate centerlines is accessed by comparing pixel coordinates of the MATLAB centerlines with pixel coordinates of centerlines created on an identical image in Microsoft Paint. During experimentation, 18 images were analyzed under both horizontal (X) and vertical (Y) skew camera conditions, with skew angles ranging from zero to 13.2 degrees. Under X-skew camera conditions the average position error between MATLAB and hand drawn centerlines is 0.6 %, while average position error under Y-skew camera conditions is 1.0 %. Fringe spacing accuracy is defined by how closely fringe spacing determined by the MATLAB solver is to fringe spacing measured by hand with a 1/16th inch ruler. Spacing analysis is performed on the same photos used in centerline determination. For X-skew camera positions, the average fringe spacing error is 6.1 %, while the average spacing error in Y-skew conditions is 4.3 %. As is discussed in later sections of this text, the X-skew fringe spacing error is artificially inflated due to human error during data collection.

Identiferoai:union.ndltd.org:CALPOLY/oai:digitalcommons.calpoly.edu:theses-3098
Date01 January 2018
CreatorsDunn, Hunter Michael
PublisherDigitalCommons@CalPoly
Source SetsCalifornia Polytechnic State University
Detected LanguageEnglish
Typetext
Formatapplication/pdf
SourceMaster's Theses

Page generated in 0.0018 seconds