Motion perception (Vision)

Vision is a critical sense for many species, with the perception of motion being a fundamental aspect. This aspect often provides richer information than static images for understanding the environment. Motion recognition is a relatively simple computation compared to shape recognition. Many creatures can discriminate moving objects quite well while having virtually no capacity for recognizing stationary objects.
Traditional methods for collision-free navigation require the reconstruction of a 3D model of the environment before planning an action. These methods face numerous limitations as they are computationally expensive and struggle to scale in unstructured and dynamic environments with a multitude of moving objects.
This thesis proposes a more scalable and efficient alternative approach without 3D reconstruction. We focus on visual motion cues, specifically ’visual looming’, the relative expansion of objects on an image sensor. This concept allows for the perception of collision threats and facilitates collision-free navigation in any environment, structured or unstructured, regardless of the vehicle’s movement or the number of moving objects present.

Prior research has explored the counterchange model of motion detection in terms of counterchanging information that originates in the stimulus foreground (or objects). These experiments explore counterchange apparent motion with regard to a new apparent motion stimulus where the necessary counterchanging information required for apparent motion is provided by altering the luminance of the background. It was found that apparent motion produced by background-counterchange requires longer frame durations and lower levels of average stimulus contrast compared to foreground-counterchange. Furthermore, inter-object distance does not influence apparent motion produced by background-counterchange to the degree it influences apparent motion produced by foreground-counterchange.

The effects of the onset, offset, and sustained presence of inducing lines on the perceived position of test lines were independently investigated in a vernier alignment task. For spatial separations larger than 2.3 min, repulsion effects were always observed. For the smallest spatial separation, 2.3 min, the effect of the inducing lines was attraction for 0 and 195 SOAs. Minimal attraction was observed for a 3000 SOA. However, when the offset effect was isolated using the 3000 SOA and a 0 ISI, a large repulsion effect was observed for the 2.3 min spatial separation, as well as for the larger spatial separations. These results indicate that the temporal separation between visual elements is as important in determining perceived position as their spatial separation, which has been demonstrated in earlier studies (Badcock & Westheimer, 1985). A differential gradient model is proposed which accounts for these findings in terms of excitatory and inhibitory interactions within an ensemble of position-sensitive units.

The effects of adaptation on motion were investigated using a modified apparent motion display. Unlike the classical apparent motion display, a BRLC (background relative luminance contrast) apparent motion display consists of two visible dots, each of a different luminance, which remain in the same position but exchange luminances on successive frames. This forms a bistable stimulus; stationarity-flicker or motion may be perceived, depending on the value of the BRLC. There was a significant interaction between condition (baseline or adaptation) and BRLC when testing motion perception following adaptation to a moving stimulus, a flickering stimulus and a static stimulus. Additionally, adaptation to flicker decreased motion perception at high BRLC values and increased it at low BRLC values. Our results reflected the presence of strong inhibitory competition between the mechanisms concerned with the perception of motion and stationarity which restricted adaptation effects to certain values of BRLC.

Anstis, et al. (1985) have reported that under certain conditions the visual system adapts and the perception of apparent motion breaks down. The present research indicates that breakdown is actually a result of same-place mechanisms successfully competing with motion-detecting mechanisms. Thus, the perception of stationarity (with flicker) can occur at the start of a trial and spontaneously switch to the perception of motion, or vice versa. The response of same-place mechanisms depends on the zero-hertz energy at each location of an apparent motion stimulus, whereas the response of motion mechanisms depends on the time-varying energy. Average luminance, luminance contrast, the temporal symmetry of the apparent motion display, and relative phase are manipulated to investigate competition between same-place and motion-detecting mechanisms.

This thesis studies the 2-D-based visual invariant that exists during relative motion between a camera and a 3-D object. We show that during fixation there is a measurable nonlinear function of optical flow that produces the same value for all points of a stationary environment regardless of the 3-D shape of the environment. During fixated camera motion relative to a rigid object, e.g., a stationary environment, the projection of the fixated point remains (by definition) at the same location in the image, and all other points located on the 3-D rigid object can only rotate relative to that 3-D fixation point. This rotation rate of the points is invariant for all points that lie on the particular environment, and it is measurable from a sequence of images. This new invariant is obtained from a set of monocular images and is expressed explicitly as a closed form solution.

In this thesis we describe a local-neighborhood-pixel-based adaptive algorithm to track image features, both spatially and temporally, over a sequence of monocular images. The algorithm assumes no a priori knowledge about the image features to be tracked, or the relative motion between the camera and the 3-D objects. The features to be tracked are selected by the algorithm and they correspond to the peaks of '2-D intensity correlation surface' constructed from a local neighborhood in the first image of the sequence to be analyzed. Any kind of motion, i.e., 6 DOF (translation and rotation), can be tolerated keeping in mind the pixels-per-frame motion limitations. No subpixel computations are necessary. Taking into account constraints of temporal continuity, the algorithm uses simple and efficient predictive tracking over multiple frames. Trajectories of features on multiple objects can also be computed. The algorithm accepts a slow, continuous change of brightness D.C. level in the pixels of the feature. Another important aspect of the algorithm is the use of an adaptive feature matching threshold that accounts for change in relative brightness of neighboring pixels. As applications of the feature-tracking algorithm and to test the accuracy of the tracking, we show how the algorithm has been used to extract the Focus of Expansion (FOE) and compute the Time-to-contact using real image sequences of unstructured, unknown environments. In both these applications, information from multiple frames is used.

Subjects judge motion direction for an apparent motion stimulus with competing perceptual organizations: Vertical vs. horizontal motion. The two patterns are coupled. When one is perceptually instantiated the other remains active in memory, resulting in sudden changes in perceived motion direction under constant stimulus conditions. The probability of change from an initially horizontal to a vertical pattern remains constant over time, showing that perceptual satiation is insufficient to explain the occurrence of spontaneous perceptual changes. It is proposed that spontaneous changes also occur because the pattern active in memory attracts the percept away from the currently instantiated pattern. The attraction hypothesis specifies that the activation of the memory pattern (and hence its attractive strength) increases as a result of previous experience. It is supported by evidence that the likelihood of changing, say from horizontal to vertical motion, is increased if the motion pattern was previously vertical.

A row of dots is presented in a series of alternating frames; dots in each frame are located at the midpoints between dots of the preceding frame. Although the perceived frame-to-frame direction of motion could vary randomly, cooperativity is indicated by the emergence of two coherent motion patterns, one unidirectional, the other oscillatory. Small increases in the time between frames are sufficient for the bias, which maintains the previously established motion direction (unidirectional motion), to be reversed, becoming a bias which inhibits that direction (oscillatory motion). Unidirectional motion, which predominates for small dot separations, and oscillatory motion, which predominates for large separations, are associated with short-range and long-range motion (Braddick, 1974) by manipulating the shape of the dots, their luminance, and the luminance of the inter-frame blank field. Pulsing/flicker emerges as a third perceptual state that competes with unidirectional motion for very small dot separations.

Six experiments were performed to examine the adequacy of detection/computation models for understanding the perception of relational motion, and in particular, the perception of three-dimensional motion in two-dimensional displays. The stimuli were a pair of dots which moved relationally (i.e., the relative location of the dots changed). Three-dimensional motion was seen when a contraction of the stimulus preceded an expansion (i.e., the dot separation first decreased, then increased), the angular difference between the pattern orientation and the direction of movement was small, and the spatial separation between dots was small. Neither the activation of higher-order, relational feature detectors, nor the construction/computation of relational motion from the detected motion of individual dots, can adequately explain the perception of three-dimensional motion.