Perception and cognition of depth of field

Abstract

A common way to present 3D materials to human observers nowadays is by stereoscopic displaying on 3D TVs or head-mounted displays such as the Oculus Rift. However, not everyone can see three-dimensional solid shape from stereoscopic viewing and the three-dimensional images remain two-dimensional pictures to them rather than solid shapes. Scientists and artists have spent a lot of effort in finding ways to dissolve pictorial space into visual space. In other words, they want to figure out how to create a sense of stereopsis in an observer when he or she is looking at non-stereo and stereo pictures. There are many monocular depth cues in addition to binocular disparity that can enhance depth perception, and these cues can therefore also create an impression of stereopsis when the viewing conditions are right. One of these depth cues is depth of field. Depth of field is defined as the distance range in which objects are perceived as sharp. Depth of field is a popular photographic technique that effectively makes the main subject in a picture appear sharp and the foreground and background blurred. The goal of this thesis is to understand the roles of depth of field in pictures from a relatively low perceptual level to a relatively high cognitive level. This thesis mainly reveals how depth of field influences the impression of stereopsis and how binocular disparity influences the perception of depth of field. I start the thesis with a study on the discrimination thresholds of depth of field. Then, a subjective study is reported in which I investigated the effects of depth of field on depth perception for binocular viewing. The next two studies address the roles of depth of field on a more cognitive level. The first of these two studies explored the effects of depth of field on change detection in pictures. The other one was conducted to evaluate how depth of field influences the aesthetic appeal and overall quality in photographs. Four subjective studies were conducted to achieve our goals. The cornerstone of the work described in this thesis is that humans are much more sensitive to changes in small depth of field than in large depth of field. A second important finding is that stereoscopic viewing does not significantly affect discrimination thresholds. Based on the discrimination thresholds that were measured in this study, I selected five levels of depth of field that can be well discriminated under stereo viewing conditions in the second study to explore the effects of depth of field on depth perception. I found that the presence of strong depth cues, in this case binocular disparity, weakens the effects of depth of field as a depth cue when depth of field is small. In contrast, when depth of field is not small, perceived depth decreases with increasing depth of field irrespective of whether binocular disparity is zero or not. With respect to the effects of depth of field on change detection, I found that depth of field directs viewers' attention similarly under both non-stereo and stereo viewing conditions. Depth of field does however weaken the effect of binocular disparity on change detection. In the final study, I found that there are no common rules on how to manipulate depth of field to make pictures more beautiful or have higher quality. The role of depth of field on aesthetic appeal and overall quality varies across content categories.