Vision and Expertise in Remote Sensing Surveying

Abstract

Remote sensing surveyors are tasked with extracting information from aerial landscape images to build maps and geospatial models of landscapes. This thesis focuses on four broad hypotheses related to this task. First, that the unfamiliar aerial viewpoint is more difficult to process than the familiar ground viewpoint, but surveyors are better than novices at processing this viewpoint. Next, surveyors are experienced with using binocular disparity cues in stereoscopic aerial images, and thus make better use of this cue. Further, surveyors also adapt to the aerial imagery, and this can alter perceptual priors for shape-from-shading. Finally, surveyors develop expertise, and this can in part be explained by perceptual learning. An initial study established that expert surveyors have a superior ability to process the aerial viewpoint, and better recognise aerial-view features, compared to novices from the general population. Next, depth perception in stereoscopic aerial images was explored with two specific depth cues: binocular disparity, and luminance cues used to interpret shape-from-shading. This study required the innovation of a novel version of the classification image technique, that can estimate the simultaneous use of binocular disparity and luminance cues. Experts and novices classified stereoscopic aerial images, and group differences showed that: 1) Experts are better at prioritising and sampling binocular disparity cues, and 2) experts may have adapted to diminish the conventional lighting-from-above prior following experience with counter-conventional light source directions in aerial images. Finally, as a hallmark of expertise is better processing of binocular disparity cues, the classification images were employed to explore stereoscopic perceptual learning in novices. This study found evidence of learning that characterises how observers improve to better sample disparity cues in stereograms. This thesis provides novel evidence on the mechanisms involved in interpreting stereoscopic aerial images, and characterises expertise in remote sensing surveyors.

Date of Award	Jul 2023
Original language	English
Supervisor	Andrew Schofield (Supervisor) & Tim S. Meese (Supervisor)