Here is a sample of my research publications. Please feel free to email me with any questions.
Visual inspection and search are important tasks in many fields, including quality control, security surveillance and medical diagnosis. We investigate whether it is better to visually inspect a moving image as opposed to a series of equivalent static images using the challenging problem of locating individuals lost in a wilderness. Wilderness search and rescue may be approached with a systematic aerial search assisted by an unmanned aerial vehicle (UAV) whose camera relays the terrain below for human inspection. We investigated two presentation modes of simulated UAV video feeds. The first mimics the live video from the downward facing camera. In the second, ‘serial visual presentation’ (SVP) mode, a static image remains in view until replaced by a new image at a rate equivalent to the live video mode. We established a statistically significant improvement in the number of detected targets in the SVP mode when compared to the Moving mode. However, these improvements were accompanied by an increase in the number of incorrectly identified targets in SVP mode. UAV speed has a significant effect on target identification in both modes, presumably due to the extra time available for viewing at lower speeds. We found no significant interaction between speed and presentation mode.
If a person is lost in the wilderness it is increasingly normal for the area in question to be over-flown by an Unmanned Aerial Vehicle (UAV) whose on-board video camera transmits a view of the terrain below. It is then the task of a human operator to visually inspect that view by means of a visual interface specifically designed to enhance the likelihood of the missing person being located.
We investigate a novel approach to the visual inspection of the terrain image: that of presenting small segments of that image for very short periods of time, though commensurate with the speed at which the UAV flies.
Participants took part in an investigation in which the challenging task was to identify the presence, in typical terrain images, of human beings. Six representative terrain maps were involved, and the six degrees of segmentation explored were such as to provide individual terrain image viewing times between 3.9 s and 108 ms.
We report the result of investigating the proposed segmentation approach to visual inspection in the demanding and realistic context of Wilderness Search and Rescue. Our investigation reveals a clear and distinctive change of visual search strategy as segmentation increases, equating to a shift between well-established notions of serial attentive search and parallel (pre-attentive) recognition.
An important task in many fields is the human visual inspection of an image. Those fields include quality control, medical diagnosis, surveillance and Wilderness Search and Rescue (WiSAR). The latter activity, triggered by an individual becoming lost, is the context within which this work proposes and evaluates a new approach to the task of human visual inspection.
Rapid Serial Visual Presentations (RSVPs) mimic the riffling of a book’s pages and are widely employed as a way of gaining familiarity with a collection of images or selecting images of interest from a collection. In a typical RSVP a number of images within a collection enter a display in (say) the lower left-hand corner and move towards the opposite corner where they disappear (‘diagonal RSVP’). If the rate at which images appear on the display is high, the speed with which images traverse the display may be such that the images overlap to some degree. As a consequence there is a need to know how much overlap can be tolerated without seriously affecting image recognition. Diagonal RSVP was implemented for four different levels of overlap and four different image speeds in such a way as to separate the effects of these two parameters. Participants were required to identify theme images (such as ‘ships’ or ‘cars’) in a series of image sequences at various combinations of speed and overlap. In addition to recording their performance, participants also completed a questionnaire to gauge their opinion on perceived presentation speed and ease of recognition. Results showed a significant effect of both overlap and speed on the percentage of correctly identified images. On the basis of the experimental results suggestions are made concerning interaction design for RSVP.