Available at: https://digitalcommons.calpoly.edu/theses/1863
Date of Award
MS in Electrical Engineering
The goal of this thesis is to allow a user to see minute motion of an object at different frequencies, using a computer program, to aid in vibration testing analysis without the use of complex setups of accelerometers or expensive laser vibrometers. MIT’s phase-based video motion processing was modified to enable modal determination of structures in the field using a cell phone camera. The algorithm was modified by implementing a stabilization algorithm and permitting the magnification filter to operate on multiple frequency ranges to enable visualization of the natural frequencies of structures in the field. To implement multiple frequency ranges a new function was developed to implement the magnification filter at each relevant frequency range within the original video. The stabilization algorithm would allow for a camera to be hand-held instead of requiring a tripod mount. The following methods for stabilization were tested: fixed point video stabilization and image registration. Neither method removed the global motion from the hand-held video, even after masking was implemented, which resulted in poor results. Specifically, fixed point did not remove much motion or created sharp motions and image registration introduced a pulsing effect. The best results occurred when the object being observed had contrast from the background, was the largest feature in the video frame, and the video was captured from a tripod at an appropriate angle. The final program can amplify the motion in user selected frequency bands and can be used as an aid in structural analysis testing.