Two-frame Motion Research Articles

Retinal ON and OFF pathways contribute to initial optokinetic responses with different temporal characteristics.

Visual information in the retina is processed via two pathways: ON and OFF pathways that originate from ON and OFF bipolar cells. The differences in the receptors that mediate signal transmission from photoreceptors imply that the response speed to light signals differs between ON and OFF pathways. We studied the initial optokinetic responses (OKRs) of mice using two-frame motion stimuli presented with interstimulus intervals (ISIs) to understand functional difference of these pathways. When two successive image frames were presented with an ISI, observers often perceived motion in the opposite direction of the actual shift. This directional reversal results from the biphasic nature of the temporal filters in visual systems whose characteristics can be estimated from the dependence on ISIs. We examined the dependence on ISIs in the OKRs of TRPM1-/- mice, whose ON bipolar cells are dysfunctional, as well as in those of wild-type control mice. Wild type and TRPM1-/- mice showed comparable OKRs in the veridical direction when no ISI was present. Both types of mice showed OKRs that decreased and eventually reversed as the ISI increased, but with a directional reversal at a shorter ISI in TRPM1-/- than wild-type mice. In addition, the temporal filters of TRPM1-/- mice estimated from dependence on ISIs were tuned for higher frequencies, suggesting that compared with wild-type mice, the visual system of TRPM1-/- mice responds to light signals with faster dynamics. We conclude that the ON and OFF pathways contribute to initial OKRs by providing visual signals processed with different temporal resolutions.

Analysis of the thermo-mechanical mechanism during ultrasonic welding of battery tabs using high-speed image capturing

Ultrasonic metal welding (USMW) has proven to be a suitable technique for joining electronic components such as battery cells. However, despite its widespread application, the USMW process still suffers from considerable quality fluctuations. The fluctuations largely originate in the lack of profound scientific knowledge and understanding of the interdependency between the tools and joining members. Therefore, this paper aims to investigate the in-situ oscillation behavior of the USMW process by using a high-speed image capturing. The recorded image-sequences are analyzed by a two-frame motion Gunnar Farneback algorithm. A two-dimensional vector field is obtained from which the displacement and relative movement between the tools and the joining members can be computed. Different welding parameters and metal alloys (AW-1050A, CW-008A) are set to study their effects on welding sequence. Microstructural investigations and T-peel tensile testing are used to validate the high-speed observations. Based on the measured amplitude profiles, USMW can be divided into six characteristic welding stages. Additionally, the measurements illustrate the dependence of the bond formation on the welding parameters and in particular on the welded materials. The results provide a more profound understanding of the thermo-mechanical mechanism during USMW and will introduce a prospective improvement in quality and process stability.

Motion-Induced Blindness Using Increments and Decrements of Luminance

Abstract Motion-induced blindness describes the disappearance of stationary elements of a scene when other, perhaps non-overlapping, elements of the scene are in motion. We measured the effects of increment (200.0 cd/m2) and decrement targets (15.0 cd/m2) and masks presented on a grey background (108.0 cd/m2), tapping into putative ON- and OFF-channels, on the rate of target disappearance psychophysically. We presented two-frame motion, which has coherent motion energy, and dynamic Glass patterns and dynamic anti-Glass patterns, which do not have coherent motion energy. Using the method of constant stimuli, participants viewed stimuli of varying durations (3.1 s, 4.6 s, 7.0 s, 11 s, or 16 s) in a given trial and then indicated whether or not the targets vanished during that trial. Psychometric function midpoints were used to define absolute threshold mask duration for the disappearance of the target. 95% confidence intervals for threshold disappearance times were estimated using a bootstrap technique for each of the participants across two experiments. Decrement masks were more effective than increment masks with increment targets. Increment targets were easier to mask than decrement targets. Distinct mask pattern types had no effect, suggesting that perceived coherence contributes to the effectiveness of the mask. The ON/OFF dichotomy clearly carries its influence to the level of perceived motion coherence. Further, the asymmetry in the effects of increment and decrement masks on increment and decrement targets might lead one to speculate that they reflect the ‘importance’ of detecting decrements in the environment.

Contrast interactions in two-frame motion discrimination imply a binocular site of contrast gain control for motion

Contrast interactions in two-frame motion discrimination imply a binocular site of contrast gain control for motion

What Visual Information is Used for Stereoscopic Depth Displacement Discrimination?

There are two ways to detect a displacement in stereoscopic depth, namely by monitoring the change in disparity over time (CDOT) or by monitoring the interocular velocity difference (IOVD). Though previous studies have attempted to understand which cue is most significant for the visual system, none has designed stimuli that provide a comparison in terms of relative efficiency between them. Here we used two-frame motion and random-dot noise to deliver equivalent strengths of CDOT and IOVD information to the visual system. Using three kinds of random-dot stimuli, we were able to isolate CDOT or IOVD or deliver both simultaneously. The proportion of dots delivering CDOT or IOVD signals could be varied, and we defined the discrimination threshold as the proportion needed to detect the direction of displacement (towards or away). Thresholds were similar for stimuli containing CDOT only, and containing both CDOT and IOVD, but only one participant was able to consistently perceive the displacement for stimuli containing only IOVD. We also investigated the effect of disparity pedestals on discrimination. Performance was best when the displacement crossed the reference plane, but was not significantly different for stimuli containing CDOT only and those containing both CDOT and IOVD. When stimuli are specifically designed to provide equivalent two-frame motion or disparity-change, few participants can reliably detect displacement when IOVD is the only cue. This challenges the notion that IOVD is involved in the discrimination of direction of displacement in two-frame motion displays.

Psychophysical measurements of referenced and unreferenced motion processing using high-resolution retinal imaging

Motion detection thresholds with a stationary frame of reference are significantly lower than unreferenced motion thresholds. To account for this, previous studies have postulated the existence of compensatory mechanisms, driven by the presence of a surround, that cancel the effects of eye movements. In the present study we used an adaptive optics scanning laser ophthalmoscope (AOSLO) to investigate the effects of retinal jitter due to fixation eye movements on referenced and unreferenced motion thresholds. The stimuli were produced by modulation of the AOSLO imaging beam, so that the absolute retinal position of targets was recorded. In Experiment 1 subjects made up/down motion judgments of a dark horizontal bar presented against a stationary 1-degree bright background. In Experiment 2 unreferenced motion thresholds were measured with isolated bright horizontal bars in otherwise complete darkness. In both experiments, AOSLO images for each trial were analyzed offline to extract retinal jitter and the retinal position of targets. For referenced motion, the results were consistent with complete compensation for eye movements by the visual system. In the unreferenced motion case eye movements adversely affected motion judgments, although there was evidence of partial compensation for such eye movements. Compensatory processes completely cancel the effect of fixation jitter for referenced motion but such compensation is partial for unreferenced motion.

Binocular contrast interactions in two-frame motion discrimination

Previous studies have shown that two-frame motion detection thresholds are elevated if one frame's contrast is raised, despite the increase in average contrast--the "contrast paradox". In this study, we investigated if such contrast interactions occurred at a monocular or binocular site of visual processing. Two-frame motion direction discrimination thresholds were measured for motion frames that were presented binocularly, dichoptically or interocularly. Thresholds for each presentation condition were measured for motion frames that comprised either matched or unmatched contrasts. The results showed that contrast mechanisms producing the contrast paradox combine contrast signals from both eyes prior to motion computation. Furthermore, the results are consistent with the existence of monocular and binocular contrast gain control mechanisms that coexist either as combined or independent systems.

Robust online appearance models for visual tracking

We propose a framework for learning robust, adaptive, appearance models to be used for motion-based tracking of natural objects. The model adapts to slowly changing appearance, and it maintains a natural measure of the stability of the observed image structure during tracking. By identifying stable properties of appearance, we can weight them more heavily for motion estimation, while less stable properties can be proportionately downweighted. The appearance model involves a mixture of stable image structure, learned over long time courses, along with two-frame motion information and an outlier process. An online EM-algorithm is used to adapt the appearance model parameters over time. An implementation of this approach is developed for an appearance model based on the filter responses from a steerable pyramid. This model is used in a motion-based tracking algorithm to provide robustness in the face of image outliers, such as those caused by occlusions, while adapting to natural changes in appearance such as those due to facial expressions or variations in 3D pose.

Isotropic integration of binocular disparity and relative motion in the perception of three-dimensional shape.

Richards (1985) showed that veridical three-dimensional shape may be recovered from the integration of binocular disparity and retinal motion information, but proposed that this integration may only occur for horizontal retinal motion. Psychophysical evidence supporting the combination of stereo and motion information is limited to the case of horizontal motion (Johnston et al., 1994), and has been criticised on the grounds of potential object boundary cues to shape present in the stimuli. We investigated whether veridical shape can be recovered under more general conditions. Observers viewed cylinders that were defined by binocular disparity, two-frame motion or a combination of disparity and motion, presented at simulated distances of 30 cm, 90 cm or 150 cm. Horizontally and vertically oriented cylinders were rotated about vertical and horizontal axes. When rotation was about the cylinder's own axis, no boundary cues to shape were introduced. Settings were biased for the disparity and two-frame motion stimuli, while more veridical shape judgements were made under all conditions for combined cue stimuli. These results demonstrate that the improved perception of three-dimensional shape in these stimuli is not a consequence of the presence of object boundary cues, and that the combination of disparity and motion is not restricted to horizontal image motion.

Minimum displacement thresholds for binocular three-dimensional motion

Thresholds for the detection of motion in depth in the median plane of the head are substantially poorer than those for motion in the frontoparallel plane. This suggests the existence of two independent mechanisms for their detection. Any three-dimensional (3-D) motion can be decomposed into components of motion in the frontoparallel plane and in the median plane of the head. Can human performance for the detection and discrimination of other 3-D motions be predicted by a combination of responses from the two independent mechanisms? Minimum displacement thresholds ( d min) for the detection of 3-D motion and the direction discrimination of 3-D motion were measured for a wide range of 3-D directions. d min data were modelled in terms of the probability summation of a pair of independent motion mechanisms, one responding to motion in the median plane of the head, the second to motion in the frontoparallel plane. Detection of 3-D motion was well predicted by probability summation across a range of 3-D directions. Direction discrimination of 3-D motion was similarly well fit by the probability summation model for multiframe motion displays for some observers. However for two-frame motion displays, direction discrimination for 3-D motion was best fit by a model using only a motion mechanism in the frontoparallel plane. Detection and direction discrimination thresholds for 3-D motion can therefore be explained on the basis of one or two mechanisms, sensitive to motion in the frontoparallel plane and in the median plane of the head.

Orientation sensitivity in human visual motion processing

Orientation tuning of receptive fields is well documented in the spatial domain, but considerable variability exists amongst published estimates of orientation sensitivity of motion receptive fields. We used a two-frame motion sequence, in which one frame was binary noise and the other was a horizontally displaced and filtered version of the same noise field, to examine the orientation sensitivity of human motion mechanisms. Initially, orientations orthogonal to the direction of motion were removed from each filtered frame. Observers indicated perceived direction of motion in a single interval, binary choice task. D max was determined for different amounts of removed orientations, and found to remain constant across the removal of energy up to approximately ±60 deg from vertical. In a second experiment, the orientations removed were now parallel to the direction of motion of the stimulus. D max fell as a cosine function with increasing removal of orientation information, in agreement with off-orientation looking or matched filtering predictions. The two experiments show the presence of mechanisms both broadly tuned and more narrowly tuned for orientation. A control experiment introduced an interstimulus interval between the two frames of our motion sequence. Performance on the direction discrimination task was severely degraded, indicating that the original results are not explicable in terms of a feature-tracking or long-range motion process. The presence of both broadly and narrowly tuned mechanisms implies multiple possible solutions to the processing of coherent plaid motion.

A contrast paradox in stereopsis, motion detection, and vernier acuity

Stereoacuity improves with increasing contrast, unless the increase is monocular. In this case performance paradoxically suffers. This study examined whether this contrast paradox occurs for two other classes of visual judgment: two-frame motion and vernier acuity. We constructed three homologous tasks in which the two components of a gabor stimulus (stereo half-images, motion frames, vernier components) were either both high contrast, both low contrast, or mismatched. The contrast paradox was evident in all three tasks and showed a similar spatial frequency dependence. We suggest the contrast paradox results from the combination of mismatched signals by a single filter.

Defaults in stereoscopic and kinetic depth perception.

This study presents three findings concerning the mechanisms of depth perception. First, the shape of the three-dimensional percept evoked by two-frame motion is defined solely by the rotation component around an axis in the frontoparallel plane; the visual system assigns a default value to this rotation component to arrive at a unique solution. Second, when the visual axes of two eyes are almost parallel, the visual system uses a default vergence value to reconstruct stereoscopic depth. Third, the default vergence and default rotation angles are highly correlated across subjects. This correlation implies that the two modalities share a common scaling default at an internal level.

Independent processing across spatial frequency in moving broadband patterns.

The aim of the experiments was to discover whether the visual system has independent access to motion information at different spatial scales when presented with a broadband stimulus. Subjects were required to discriminate between a pair of two-frame motion sequences, one containing a coherently displacing pattern and the other containing a pattern with high-frequency noise. The stimuli were either narrowband (1 octave) or broadband (6 octaves spanning 0.23-15.0 cycles deg-1) and their power spectra were either flat or followed a 1/f2 function. For the broadband stimuli, noise was introduced cumulatively into increasingly lower frequencies. For the narrowband stimuli, noise was introduced into the same frequency band as the signal. All stimuli could be defined by the lowest noise frequency (nl) they contained. For each stimulus, the largest spatial displacement across the two frames at which the task could be performed was measured (dmax). For the narrowband stimuli, dmax increased as nl was lowered. This was true over the entire frequency range for the 1/f2 stimuli, though the task became impossible for the flat-spectrum stimuli at the lowest frequencies. This is attributed to the very low contrast of these latter stimuli. The dmax values for the broadband stimuli tended to shadow those of the narrowband stimuli with the equivalent values of nl being around 25% lower. The data were modelled by spatiotemporally filtering the stimuli and considering the amount of directional power in the signal and noise sequences. The results suggest that there must be multiple spatial-frequency channels in operation, and that for broadband patterns the visual system has perceptual access to these individual channel outputs, utilising different filters depending on the task requirements.

Anisotropy for direction discrimination in a two-frame apparent motion display.

Direction discrimination (upward/downward or left/right) for a Gabor patch in a two-frame motion display was measured as a function of the inter-frame displacement size of the component grating with the stimulus position (center, left, right, upper and lower visual fields) as a parameter. The results showed that, for vertical motion in the center, left, right and lower visual fields, the observers saw downward motion more frequently than upward motion, whereas for vertical motion in the upper field and for horizontal motion, no preference for one of the two opposite directions was obtained. Human motion vision is anisotropic in the lower half of the visual field.

What determines the maximum displacement limit for spatially broadband kinematograms?

Two experiments are described that are designed to investigate what determines the maximum spatial displacement detectable (dmax) for spatially broadband patterns exposed in a two-frame motion sequence. In experiment 1, dmax was found to be 1.63 times greater for a two-dimensional (2-D) broadband random pattern with a 1/f Fourier amplitude spectrum (equal contrast in each octave) than for a 2-D binary-valued random-dot pattern with a flat spectrum (higher contrast in higher-frequency octaves). In experiment 2, dmax was shown to vary in inverse proportion to the lowest stimulus frequency for random patterns with a one-octave bandwidth and normalized contrast. Furthermore, when these five one-octave patterns were summed together, dmax for this new five-octave pattern was found to be only 1.46 times lower than dmax for the lowest-frequency one-octave pattern presented alone. A model is described in which direction discrimination is based on the nearest-neighbor matching of zero crossings in the output of a single-spatial-filter bandpass in both spatial frequency and orientation. Data from the model show that the difference between dmax for the five-octave and the lowest one-octave patterns can be accounted for by the same filter passing some of the additional higher frequencies in the former pattern. Furthermore, it is argued that all the data can be accounted for by assuming that dmax is determined by the coarsest spatial filter activated by each stimulus. Modeling the results of both experiments suggests that the bandwidth of this filter is ~2.6 octaves and reaches peak sensitivity at ~0.47 c/deg. The model is shown to be capable of accounting for a wide range of other two-frame dmax data.

Cross-modal slant and curvature matching of stereo- and motion-induced surfaces

In many laboratory setups and in many day-to-day situations, a unique solution of the structure-from-two-views problem is unobtainable. Yet, when the visual system is presented with two projections in a sequence, it nevertheless appears to generate a reasonably stable percept of structure. In the research reported here, we examined whether the same surface would be perceived when subjects were presented with a pair of views that alternated in time monocularly (two-frame motion) or were shown simultaneously to both eyes (stereo). In Experiment 1, we studied slant perception: human observers were asked to match the slant of a motion-induced planar surface with its stereo-induced counterpart. In Experiment 2, the perceived curvature of parabolic surfaces was matched in a similar way. The results show that motion-induced slant is matched with a higher value of the stereo-induced slant. However, the curvature experiment showed that motion-induced curvature is matched with a lower stereo-induced curvature. One possible explanation may be that the slant and curvature are internally inconsistent in at least one of the modalities.

Passive ranging using a moving camera

AbstractThis article addresses the problem of determining the 3‐dimensional locations of salient points in the environment of a moving camera based on a monocular image sequence obtained by the camera. The camera's translational and rotational velocities are assumed to be known approximately via inertial sensors. The motion of the camera is represented by a constant velocity model. Salient points in the image sequence are extracted using Gabor wavelets and tracked using labeled graph matching. The 3‐D positions of the selected environmental points relative to the camera are then estimated recursively using an extended Kalman filter (EKF), after initialization by two‐frame motion stereo. The motion parameters of the camera are also refined simultaneously. Experimental results on real data are given. © 1992 John Wiley & Sons, Inc.

Effects of pattern element density upon displacement limits for motion detection in random binary luminance patterns.

The upper motion displacement threshold (Dmax) was determined with two-frame motion sequences of random binary luminance patterns, over a range of pattern element sizes and densities. Dmax was little affected by density at small element sizes (less than 5 arcmin), in agreement with previous reports. However, at larger element sizes (greater than 9 arcmin) Dmax increased as element density was reduced in the range 50-5%. We explain our findings by a model which takes into account spatial-frequency filtering prior to motion detection, and the effects of pattern density upon the statistics of random binary patterns. We also implicate the dependence of Dmax upon the contrast energy of the elements in broadband patterns, and provide a direct demonstration that Dmax is contrast limited over a wide range of pattern contrasts (72-2.5%). Previous reports that Dmax is independent of density should be modified to take into account the complex effects of density upon the statistics of random patterns, and the existence of physiological filtering prior to motion detection.