Outdoor tracking using computer vision, xenon strobe illumination and retro-reflective landmark

This thesis investigates light conditioning methods for improved object recognition of outdoor structured landmarks using low-cost machine vision. The research's main objective is to produce a high contrast image of the structured target at distance up to a quarter mile. The target's contrast is enhanced by reducing the background noise associated with the outdoor environment. The conditioning method consists of an electronically shuttered vision system and a xenon strobe tube. The high contrast image with low background noise eases object recognition thus improving robustness and system response times. Tracking landmarks outdoors is important for farming, construction, and highway automation, as well as practical low-cost system for smaller scale outdoor automation tasks.;Two different setups using a xenon strobe are investigated both analytically and experimentally; a xenon strobe illuminating a retro-reflective landmark and a xenon strobe used as a beacon. The two different signal are compared to a noise surface consisting of a white diffuse surface illuminated by the sun. The analytical analysis shows the S/N ratio can be considered a ratio of the irradiances at the CCD detector with the CCD sensitivity considered a filter for different wavelengths. The S/N ratio is primarily affected by the strobe intensity, the retro-reflector efficiency and the distance to the target.;Numerical analysis shows the retro-reflector can be detected to 530 meters with a S/N ratio of 2 for the nominal case of a 1 Joule xenon flash focused to a 5 degree cone angle with a flash duration of 100 $mu$s. The analysis also shows an unfocused strobe beacon can be detected to 620 meters with a S/N ratio of 5 for the same strobe energy. The target position can be determined to within 17 millimeters at 400 meters using sub-pixel edge detection and a vision system with a 24 mm lens and 10 $mu$m square pixels.;The outdoor conditioning methods allow simple object recognition and tracking algorithms to be used with increased robustness. Future improvements can be made by improving the CCD's image shuttering capabilities, lowering the flash duration, and focusing the strobe for the strobe beacon setup.