A perturbation analysis of depth perception from combinations of texture and motion cues

Abstract

We examined how depth information from two different cue types (object motion and texture gradient) is integrated into a single estimate in human vision. Two critical assumptions of a recent'model of depth cue combination (termed moa!i@d weak fusion) were tested. The first assumption is that the overall depth estimate is a weighted linear combination of the estimates derived from the individual cues, after initial processing needed to bring them to a common format. The second assumption is that the weight assigned to a cue reflects the apparent reliability of that cue in a particular scene. By this account, the depth combination rule is linear and dynamic, changing in a predictable fashion in response to the particular scene and viewing conditions. A novel procedure was used to measure the weights assigned to the texture and motion cues across experimental conditions. This procedure uses a type of perturbation analysis. The results are consistent with the weighted linear combination rule. In addition, when either cue is corrupted by added noise, the weighted linear combination rule shifts in favor of the uncontaminated cue.

Integration of stereopsis and motion shape cues

“…However, their results could be explained by assigning weight to a default "flatness" term when only one cue is present, but giving no weight to flatness when two consistent cues are present. This term can be thought of as a default tendency towards seeing no depth when the information is poor (Young et al, 1993). This flatness explanation could be applied to the stereo and motion data at the 200 cm viewing distance (Fig.…”

Toward a new theory of stereopsis.

“…The induction of stereopsis under monocular viewing of a picture is ascribed to the elimination of the conflicting binocular information, resulting in a greater magnitude of perceived depth consistent with the monocular cues (Ames, 1925;Schlosberg, 1949;Koenderink, 1998; see also Young et al, 1993). This idea therefore links stereopsis to a greater coherence among depth cues and a greater magnitude of perceived depth.…”

“…Only one conventional cue 30 (shading) specifies a 3-D shape in the image shown in Figure 20, but it is a cue that is generally regarded to be significantly less reliable than binocular disparity (Bülthoff & Mallot, 1988;Vuoung, 55 . Moreover, recent statistical models of depth cue integration posit the existence of a "flatness prior" which claims that the visual system is biased towards inferring flat frontoparallel surfaces in the presence of unreliable or conflicting depth cues (Adams & Mammasian, 2004;Domini, Shah & Caudek, 2011;Watt et al, 2005;Young et al, 1993). Since the majority of cues, including the most reliable one (disparity), specify a flat surface, it would be difficult for cue integration models to adequately explain why we perceive a 3-D object instead of a flat patterned surface.…”

An algorithmic model of heading perception

“…The weak fusion model is successful for explanation of cue integration about depth (Young et al 1993;Landy et al 1995). In the weak fusion model, the visual system weighs depth cues according to the reliability and linearly combines the depth cues using the weights.…”