Speed tuning in elementary motion detectors of the correlation type

A prominent model of visual motion detection is the so-called correlation or Reichardt detector. Whereas this model can account for many properties of motion vision, from humans to insects (review, Borst and Egelhaaf 1989), it has been commonly assumed that this scheme of motion detection is not well suited to the measurement of image velocity. This is because the commonly used version of the model, which incorporates two unidirectional motion detectors with opposite preferred directions, produces a response which varies not only with the velocity of the image, but also with its spatial structure and contrast. On the other hand, information on image velocity can be crucial in various contexts, and a number of recent behavioural experiments suggest that insects do extract velocity for navigational purposes (review, Srinivasan et al. 1996). Here we show that other versions of the correlation model, which consists of a single unidirectional motion detector or incorporates two oppositely directed detectors with unequal sensitivities, produce responses which vary with image speed and display tuning curves that are substantially independent of the spatial structure of the image. This surprising feature suggests simple strategies of reducing ambiguities in the estimation of speed by using components of neural hardware that are already known to exist in the visual system. Received: 30 April 1998 / Accepted in revised form: 18 September 1998     

Active vision in insects: an analysis of object-directed zig-zag flights in wasps (Odynerus spinipes , Eumenidae)

Ground-nesting wasps (Odynerus spinipes, Eumenidae) perform characteristic zig-zag flight manoeuvres when they encounter a novel object in the vicinity of their nests. We analysed flight parameters and flight control mechanisms and reconstructed the optical flow fields which the wasps generate by these flight manoeuvres. During zig-zag flights, the wasps move sideways and turn to keep the object in their frontal visual field. Their turning speed is controlled by the relative motion between object and background. We find that the wasps adjust their rotational and translational speed in such a way as to produce a specific vortex field of image motion that is centred on the novel object. As a result, differential image motion and changes in the direction of motion vectors are maximal in the vicinity and at the edges of the object. Zig-zag flights thus seem to be a `depth from motion' procedure for the extraction of object-related depth information. Accepted: 31 August 1997     

Image motion environments: background noise for movement-based animal signals

Understanding the evolution of animal signals has to include consideration of the structure of signal and noise, and the sensory mechanisms that detect the signals. Considerable progress has been made in understanding sounds and colour signals, however, the degree to which movement-based signals are constrained by the particular patterns of environmental image motion is poorly understood. Here we have quantified the image motion generated by wind-blown plants at 12 sites in the coastal habitat of the Australian lizard Amphibolurus muricatus. Sampling across different plant communities and meteorological conditions revealed distinct image motion environments. At all locations, image motion became more directional and apparent speed increased as wind speeds increased. The magnitude of these changes and the spatial distribution of image motion, however, varied between locations probably as a function of plant structure and the topographic location. In addition, we show that the background motion noise depends strongly on the particular depth-structure of the environment and argue that such microhabitat differences suggest specific strategies to preserve signal efficacy. Movement-based signals and motion processing mechanisms, therefore, may reveal the same type of habitat specific structural variation that we see for signals from other modalities.     

Motion perception and motion estimation by total-least squares

A computational model of motion perception is proposed. The model, which is gradient-based, adheres to the neural constraint that transmitted signals are positive-valued functions by posing the estimation of image motion as a quadratic programming problem combined with total-least squares: a model that assumes that image signals are contaminated by noise in both the spatial and temporal dimensions. By shrinking motion estimates with a regularizer whose subtractive effect introduces a contrast dependent speed threshold into motion computations, it is shown that the total-least squares model when posed as a quadratic programming problem, is capable of explaining both increases and decreases in perceived speed as these effects were reported by Thompson (1982) to vary as a function of image contrast and temporal frequency. The correlation that exists between the model's contrast speed response and results reported from visual psychophysics is consistent with the view that the visual system assumes that image signals may be contaminated by noise in both the spatial and the temporal domain, and that visual motion is influenced by the consequence of these assumptions.     

A Bio-Inspired,Motion-Based Analysis of Crowd Behavior Attributes Relevance to Motion Transparency,Velocity Gradients,and Motion Patterns

The analysis of motion crowds is concerned with the detection of potential hazards for individuals of the crowd. Existing methods analyze the statistics of pixel motion to classify non-dangerous or dangerous behavior, to detect outlier motions, or to estimate the mean throughput of people for an image region. We suggest a biologically inspired model for the analysis of motion crowds that extracts motion features indicative for potential dangers in crowd behavior. Our model consists of stages for motion detection, integration, and pattern detection that model functions of the primate primary visual cortex area (V1), the middle temporal area (MT), and the medial superior temporal area (MST), respectively. This model allows for the processing of motion transparency, the appearance of multiple motions in the same visual region, in addition to processing opaque motion. We suggest that motion transparency helps to identify “danger zones” in motion crowds. For instance, motion transparency occurs in small exit passages during evacuation. However, motion transparency occurs also for non-dangerous crowd behavior when people move in opposite directions organized into separate lanes. Our analysis suggests: The combination of motion transparency and a slow motion speed can be used for labeling of candidate regions that contain dangerous behavior. In addition, locally detected decelerations or negative speed gradients of motions are a precursor of danger in crowd behavior as are globally detected motion patterns that show a contraction toward a single point. In sum, motion transparency, image speeds, motion patterns, and speed gradients extracted from visual motion in videos are important features to describe the behavioral state of a motion crowd.     

The spatial origin of a perceptual transition in binocular rivalry

When the left and the right eye are simultaneously presented with incompatible images at overlapping retinal locations, an observer typically reports perceiving only one of the two images at a time. This phenomenon is called binocular rivalry. Perception during binocular rivalry is not stable; one of the images is perceptually dominant for a certain duration (typically in the order of a few seconds) after which perception switches towards the other image. This alternation between perceptual dominance and suppression will continue for as long the images are presented. A characteristic of binocular rivalry is that a perceptual transition from one image to the other generally occurs in a gradual manner: the image that was temporarily suppressed will regain perceptual dominance at isolated locations within the perceived image, after which its visibility spreads throughout the whole image. These gradual transitions from perceptual suppression to perceptual dominance have been labeled as traveling waves of perceptual dominance. In this study we investigate whether stimulus parameters affect the location at which a traveling wave starts. We varied the contrast, spatial frequency or motion speed in one of the rivaling images, while keeping the same parameter constant in the other image. We used a flash-suppression paradigm to force one of the rival images into perceptual suppression. Observers waited until the suppressed image became perceptually dominant again, and indicated the position at which this breakthrough from suppression occurred. Our results show that the starting point of a traveling wave during binocular rivalry is highly dependent on local stimulus parameters. More specifically, a traveling wave most likely started at the location where the contrast of the suppressed image was higher than that of the dominant one, the spatial frequency of the suppressed image was lower than that of the dominant one, and the motion speed of the suppressed image was higher than that of the dominant one. We suggest that a breakthrough from suppression to dominance occurs at the location where salience (the degree to which a stimulus element stands out relative to neighboring elements) of the suppressed image is higher than that of the dominant one. Our results further show that stimulus parameters affecting the temporal dynamics during continuous viewing of rival images described in other studies, also affect the spatial origin of traveling waves during binocular rivalry.     

Tetragonisca guard bees interpret expanding and contracting patterns as unintended displacement in space

Guard bees of the stingless bee Tetragonisca angustula (Apidae: Meliponinae) hover in stable positions in front of the nest to protect the flight corridor leading to the nest entrance against insect intruders. To unravel the visual control of station keeping, we exposed these hovering guards to expanding and contracting patterns at the nest front. The bees fly away from an expanding pattern and towards the centre of a contracting pattern along a line connecting their initial position and the centre of expansion regardless of where in the visual field they view the pattern. The response of bees to a spinning radial pattern is different: they fly parallel to the pattern, up and down or forward and backward depending on whether they initially hover to the side, above or below the centre of rotation. The bees respond to horizontal and to vertical expansion and contraction. They also adjust their distance relative to a rotating spiral which produces a realistic flow field and thus allowed us to test to what extent the bees minimize image motion speed. We find that guard bees indeed move in the appropriate direction to minimize the image motion speed they experience. A comparison of bees hovering at different distances from the nestfront at the onset of pattern motion and experiencing very different image velocities shows that the dynamics of the reaction is quite uniform. At the pattern velocities tested, we did not find evidence that guard bees use image motion to control their flight speed. The bees' response rather suggests that the underlying mechanism might be insensitive to the size of motion vectors. Accepted: 2 April 1997     

Quantification of phase synchronization phenomena and their importance for verbal memory processes

The tangential neurons in the lobula plate region of the flies are known to respond to visual motion across broad receptive fields in visual space.When intracellular recordings are made from tangential neurons while the intact animal is stimulated visually with moving natural imagery,we find that neural response depends upon speed of motion but is nearly invariant with respect to variations in natural scenery. We refer to this invariance as velocity constancy. It is remarkable because natural scenes, in spite of similarities in spatial structure, vary considerably in contrast, and contrast dependence is a feature of neurons in the early visual pathway as well as of most models for the elementary operations of visual motion detection. Thus, we expect that operations must be present in the processing pathway that reduce contrast dependence in order to approximate velocity constancy.We consider models for such operations, including spatial filtering, motion adaptation, saturating nonlinearities, and nonlinear spatial integration by the tangential neurons themselves, and evaluate their effects in simulations of a tangential neuron and precursor processing in response to animated natural imagery. We conclude that all such features reduce interscene variance in response, but that the model system does not approach velocity constancy as closely as the biological tangential cell.     

Evidence for velocity-tuned motion-sensitive descending neurons in the honeybee.

Behavioural experiments suggest the existence of two functionally distinct movement-sensitive pathways in honeybees: one mediates optomotor behaviour, consisting of reflexive turning responses preventing deviations from course, and the other controls flight speed. The first consists of direction-selective neurons responding optimally to a particular temporal frequency of motion, regardless of the pattern's spatial structure. The temporal frequency dependence matches the temporal tuning of the optomotor output. Behavioural experiments suggest the second pathway contains velocity-tuned cells, which generate equal-sized responses for any given image velocity, for patterns with a range of spatial structures. Here, recordings were made from direction-selective neurons in the honeybee's ventral nerve cord. Neurons were tested for responses to motion at velocities of 40-1000 deg s(-1) using four gratings with spatial periods of 11-76 degrees. In addition to temporal frequency-dependent optomotor neurons, direction-selective cells were found that had the same shaped velocity-response functions for all four patterns. The velocity-tuning properties of these cells suggest a possible role in monitoring flight speed because their velocity tuning matches the image velocities encountered during free flight and landing behaviour.     

Factors affecting feeding selectivity of visual predators on the copepod Acartia tonsa: locomotion,visibility and escape responses

Visual predation by fish on copepods involves prey encounter, attack and capture; during any of these processes prey selection can occur. Developmental changes in copepods, including increases in swimming speed, size and image contrast increase the encounter rate and distance at which they can be detected by predators. Copepods compensate for this increase vulnerability with age through diel vertical migration and improved escape capabilities. This study quantifies the changes in swimming speed and movement pattern with developmental stage of the copepod Acartia tonsa, using a video-computer system for motion analysis. Changes in visible size and image contrast with developmental stage were quantified under simulated natural illumination conditions using a video based image analysis system. The escape responses of the naupliar stages of the copepod Acartia tonsa were quantified in response to a stationary pipette sucking in water at a constant speed. Accurate quantification of the parameters that affect feeding selectivity of planktivorous fish will provide the basis for evaluation of their relative importance in future studies.     

A model of visual detection of angular speed for bees

A fly or bee's responses to widefield image motion depend on two basic parameters: temporal frequency and angular speed. Rotational optic flow is monitored using temporal frequency analysers, whereas translational optic flow seems to be monitored in terms of angular speed. Here we present a possible model of an angular speed detector which processes input signals through two parallel channels. The output of the detector is taken as the ratio of the two channels’ outputs. This operation amplifies angular speed sensitivity and depresses temporal frequency tuning. We analyse the behaviour of two versions of this model with different filtering properties in response to a variety of input signals. We then embody the detector in a simulated agent's visual system and explore its behaviour in experiments on speed control and odometry. The latter leads us to suggest a new algorithm for optic flow driven odometry.     

Diverse speed response properties of motion sensitive neurons in the fly’s optic lobe

Speed and acceleration are fundamental components of visual motion that animals can use to interpret the world. Behavioral studies have established that insects discriminate speed largely independently of contrast and spatial frequency, and physiological recordings suggest that a subset of premotor descending neurons is in this sense speed-selective. Neural substrates and mechanisms of speed selectivity in insects, however, are unknown. Using blow flies Phaenicia sericata, intracellular recordings and dye-fills were obtained from medulla and lobula complex neurons which, though not necessarily speed-selective themselves, are positioned to participate in circuits that produce speed-selectivity in descending neurons. Stimulation with sinusoidally varied grating motion (0–200°/s) provided a range of instantaneous velocities and accelerations. The resulting speed response profiles are indicative of four distinct speed ranges, supporting the hypothesis that the spatiotemporal tuning of mid-level neurons contains sufficient diversity to account for the emergence of speed selectivity at the descending neuron level. This type of mechanism has been proposed to explain speed discrimination in both insects and mammals, but has seemed less likely for insects due to possible constraints on small brains. Two additional recordings are suggestive of acceleration-selectivity, a potentially useful visual capability that is of uncertain functional significance for arthropods.     

Computational aspects of motion perception in natural and artificial vision systems.

In this paper a computational scheme for motion perception in artificial and natural vision systems is described. The scheme is motivated by a mathematical analysis in which first-order spatial properties of optical flow, such as singular points and elementary components of optical flow, are shown to be salient features for the computation and analysis of visual motion. The fact that different methods for the computation of optical flow produce similar results is explained in terms of the simple spatial structure of the image motion of rigid bodies. Singular points and elementary flow components are used to compute motion parameters, such as time-to-collision and angular velocity, and also to segment the visual field into areas which correspond to different motions. Then a number of biological implications are discussed. Electrophysiological findings suggest that the brain perceives visual motion by detecting and analysing optical flow components. However, the cortical neurons, which seem to detect elementary flow components, are not able to extract these components from more complex flows. A simple model for the organization of the receptive field of these cells, which is consistent with anatomical and electrophysiological data, is described at the end of the paper.     

Perceived motion in complementary afterimages: verification of a neural network theory

Steady fixation of a regular pattern like a bar grating or concentric circles leads to a complementary afterimage at pattern offset. The afterimage has the appearance of shimmering lines that are locally orthogonal to the orientations of the inducing image. Additionally, the afterimage includes motion running parallel to the orientation of the afterimage lines. We argue that this afterimage motion supports the existence of a cue to motion that is based on the spatial organization of oriented responses. This cue was previously proposed after analysis of a neural network model of visual perception. We test predictions of the model on various types of complementary afterimage inducing stimuli. When a contrast or size gradient is included in the inducing image, the afterimage motion moves toward the higher part of the gradient, in agreement with the model. Implications of this cue for computational and neurophysiological theories of motion perception are discussed.     

Non-directional motion detectors can be used to mimic optic flow dependent behaviors

Insect navigational behaviors including obstacle avoidance, grazing landings, and visual odometry are dependent on the ability to estimate flight speed based only on visual cues. In honeybees, this visual estimate of speed is largely independent of both the direction of motion and the spatial frequency content of the image. Electrophysiological recordings from the motion-sensitive cells believed to underlie these behaviors have long supported spatio-temporally tuned correlation-type models of visual motion detection whose speed tuning changes as the spatial frequency of a stimulus is varied. The result is an apparent conflict between behavioral experiments and the electrophysiological and modeling data. In this article, we demonstrate that conventional correlation-type models are sufficient to reproduce some of the speed-dependent behaviors observed in honeybees when square wave gratings are used, contrary to the theoretical predictions. However, these models fail to match the behavioral observations for sinusoidal stimuli. Instead, we show that non-directional motion detectors, which underlie the correlation-based computation of directional motion, can be used to mimic these same behaviors even when narrowband gratings are used. The existence of such non-directional motion detectors is supported both anatomically and electrophysiologically, and they have been hypothesized to be critical in the Dipteran elementary motion detector (EMD) circuit.     

Estimation of the potential speed of range expansion of an introduced species: characteristics and applicability of the gamma model

The potential speed at which the range of an introduced species expands in its optimal environment can be predicted by using the gamma model proposed by Yamamura (Popul Ecol 46:87–101, 2004). In this paper we first clarify the difference between the gamma model and Einstein’s Brownian motion model. We then apply the model to the ragweed beetle, Ophraella communa LeSage that rapidly expanded its distribution in Japan after it was first found in 1996. The parameters of the model are estimated by conducting a field experiment. The species’ net reproductive rate is examined in the laboratory. By combining these estimates, we estimate the potential speed of range expansion to be 82 km per generation and 329 km per year, while the observed speed is estimated to be 77 km per year, the observed speed being considerably slower than the potential speed. This discrepancy may be due to the low reproductive rate caused by mortality in the actual field. The applicability of the gamma model to the econometric data is also briefly discussed.     

Unconscious Local Motion Alters Global Image Speed

Accurate motion perception of self and object speed is crucial for successful interaction in the world. The context in which we make such speed judgments has a profound effect on their accuracy. Misperceptions of motion speed caused by the context can have drastic consequences in real world situations, but they also reveal much about the underlying mechanisms of motion perception. Here we show that motion signals suppressed from awareness can warp simultaneous conscious speed perception. In Experiment 1, we measured global speed discrimination thresholds using an annulus of 8 local Gabor elements. We show that physically removing local elements from the array attenuated global speed discrimination. However, removing awareness of the local elements only had a small effect on speed discrimination. That is, unconscious local motion elements contributed to global conscious speed perception. In Experiment 2 we measured the global speed of the moving Gabor patterns, when half the elements moved at different speeds. We show that global speed averaging occurred regardless of whether local elements were removed from awareness, such that the speed of invisible elements continued to be averaged together with the visible elements to determine the global speed. These data suggest that contextual motion signals outside of awareness can both boost and affect our experience of motion speed, and suggest that such pooling of motion signals occurs before the conscious extraction of the surround motion speed.     

Bayesian modeling of perceived surface slant from actively-generated and passively-observed optic flow

We measured perceived depth from the optic flow (a) when showing a stationary physical or virtual object to observers who moved their head at a normal or slower speed, and (b) when simulating the same optic flow on a computer and presenting it to stationary observers. Our results show that perceived surface slant is systematically distorted, for both the active and the passive viewing of physical or virtual surfaces. These distortions are modulated by head translation speed, with perceived slant increasing directly with the local velocity gradient of the optic flow. This empirical result allows us to determine the relative merits of two alternative approaches aimed at explaining perceived surface slant in active vision: an "inverse optics" model that takes head motion information into account, and a probabilistic model that ignores extra-retinal signals. We compare these two approaches within the framework of the bayesian theory. The "inverse optics" bayesian model produces veridical slant estimates if the optic flow and the head translation velocity are measured with no error; because of the influence of a "prior" for flatness, the slant estimates become systematically biased as the measurement errors increase. The bayesian model, which ignores the observer's motion, always produces distorted estimates of surface slant. Interestingly, the predictions of this second model, not those of the first one, are consistent with our empirical findings. The present results suggest that (a) in active vision perceived surface slant may be the product of probabilistic processes which do not guarantee the correct solution, and (b) extra-retinal signals may be mainly used for a better measurement of retinal information.     

A model of motion adaptation and motion after-effects based upon principal component regression

A computational model to help explain effects of adaptation to moving signals is compared with established energy (linear regression) models of motion detection. The proposed model assumes that processed image signals are subject to error in both dimensions of space and time. This assumption constrains models of motion perception to be based upon principal component regression rather than linear regression. It is shown that response suppression of model complex cell neurons that input into the model may account for (1) increases in perceived speed after adaptation to static patterns and testing with slowly moving patterns, (2) significant increases in perceived speed after adaptation to patterns moving at a medium speed and testing at high speed, and (3) decreases in perceived speed in the opponent direction to a quickly moving adapting signal. Neither of predictions (2) or (3) are general features of established accounts of motion detection by visual processes based upon linear regression. Comparisons of the proposed model's speed transfer function with existing psychophysical data suggests that the visual system processes motion signals with the tacit assumption that image measurements are subject to error in both space and time. Received: 24 January 2000 / Accepted in revised form: 8 May 2000     

Deep-learning-based motion correction in optical coherence tomography angiography

Optical coherence tomography angiography (OCTA) is a widely applied tool to image microvascular networks with high spatial resolution and sensitivity. Due to limited imaging speed, the artifacts caused by tissue motion can severely compromise visualization of the microvascular networks and quantification of OCTA images. In this article, we propose a deep-learning-based framework to effectively correct motion artifacts and retrieve microvascular architectures. This method comprised two deep neural networks in which the first subnet was applied to distinguish motion corrupted B-scan images from a volumetric dataset. Based on the classification results, the artifacts could be removed from the en face maximum-intensity-projection (MIP) OCTA image. To restore the disturbed vasculature induced by artifact removal, the second subnet, an inpainting neural network, was utilized to reconnect the broken vascular networks. We applied the method to postprocess OCTA images of the microvascular networks in mouse cortex in vivo. Both image comparison and quantitative analysis show that the proposed method can significantly improve OCTA image by efficiently recovering microvasculature from the overwhelming motion artifacts.