Reversals of visual depth caused by motion parallax

The role of the velocity and direction of retinal movement in the determination of apparent depth from motion parallax was examined. Motion parallax was produced either by linking the movement of random-dots to head movement or by making this motion independent of the head movement. The results show that apparent depth was largely estimated from the velocity difference between the stimuli. The direction of retinal movement in the absence of head movement did not determine whether the pattern appeared to protrude or recede. Information about direction linked to head movement was able to stabilize protrusion/recession by providing a cue for the location of the fixation point. Depth reversal occurred less frequently in the presence than in the absence of head movement. When the fixation point shifted from the apparently protruding pattern to the apparently receding pattern, in both the presence and absence of head movement, depth reversal was readily observed.     

Automatic recognition of eye blinking in spontaneously occurring behavior

Previous researchin automatic facial expression recognition has been limited to recognition of gross expression categories (e.g., joy or anger) in posed facial behavior under well-controlled conditions (e.g., frontal pose and minimal out-of-plane head motion). We have developed a system that detects a discrete and important facial action (e.g., eye blinking) in spontaneously occurring facial behavior that has been measured with a nonfrontal pose, moderate out-of-plane head motion, and occlusion. The system recovers three-dimensional motion parameters, stabilizes facial regions, extracts motion and appearance information, and recognizes discrete facial actions in spontaneous facial behavior. We tested the system in video data from a two-person interview. The 10 subjects were ethnically diverse, action units occurred during speech, and out-of-plane motion and occlusion from head motion and glasses were common. The video data were originally collected to answer substantive questions in psychology and represent a substantial challenge to automated action unit recognition. In analysis of blinks, the system achieved 98% accuracy.     

The visual representation of three-dimensional, rotating objects.

Depth rotations can reveal new object parts and result in poor recognition of "static" objects (Biederman & Gerhardstein, 1993). Recent studies have suggested that multiple object views can be associated through temporal contiguity and similarity (Edelman & Weinshall, 1991; Lawson, Humphreys & Watson, 1994; Wallis, 1996). Motion may also play an important role in object recognition since observers recognize novel views of objects rotating in the picture plane more readily than novel views of statically re-oriented objects (Kourtzi & Shiffrar, 1997). The series of experiments presented here investigated how different views of a depth-rotated object might be linked together even when these views do not share the same parts. The results suggest that depth rotated object views can be linked more readily with motion than with temporal sequence alone to yield priming of novel views of 3D objects that fall in between "known" views. Motion can also enhance path specific view linkage when visible object parts differ across views. Such results suggest that object representations depend on motion processes.     

HIPS: Image processing under UNIX. Software and applications

HIPS (Human Information Processing Laboratory’s Image processing System) is a software system for image processing that runs under the UNIX operating system. HIPS is modular and flexible: it provides automatic documentation of its actions, and is relatively independent of special equipment. It has proved its usefulness in the study of the perception of American Sign Language (ASL). Here, we demonstrate some of its applications in the study of vision, and as a tool in general signal processing. Ten examples of HIPS-generated stimuli and—in some cases—analyses are provided, including the spatial filtering analysis of two types of visual illusions; the study of frequency channels with sine-wave gratings and band-limited noise; 3-dimensional perceptual reconstruction from 2-dimensional images in the kinetic depth effect; the perception of depth in random dot stereograms and cinematograms; and the perceptual segregation of objects induced by differential dot motion. Finally, examples of noise-masked, cartoon coded, and hierarchically encoded ASL images are provided.     

The Poggendorf illusion in stereoscopic space.

The stereoscopic depth separation between the bisecting rectangle and the oblique line of a Poggendorf configuration was manipulated by varying the direction and magnitude of disparity carried by the rectangle. Based upon data of 6 subjects, the magnitude of the illusion decreased with increasing depth separation regardless of the direction of disparity. Depth separation varied directly with disparity. These findings make plain that depth adjacency can operate symmetrically in stereoscopic space.     

Characterizing Human Expertise Using Computational Metrics of Feature Diagnosticity in a Pattern Matching Task

Forensic evidence often involves an evaluation of whether two impressions were made by the same source, such as whether a fingerprint from a crime scene has detail in agreement with an impression taken from a suspect. Human experts currently outperform computer‐based comparison systems, but the strength of the evidence exemplified by the observed detail in agreement must be evaluated against the possibility that some other individual may have created the crime scene impression. Therefore, the strongest evidence comes from features in agreement that are also not shared with other impressions from other individuals. We characterize the nature of human expertise by applying two extant metrics to the images used in a fingerprint recognition task and use eye gaze data from experts to both tune and validate the models. The Attention via Information Maximization (AIM) model (Bruce & Tsotsos, 2009) quantifies the rarity of regions in the fingerprints to determine diagnosticity for purposes of excluding alternative sources. The CoVar model (Karklin & Lewicki, 2009) captures relationships between low‐level features, mimicking properties of the early visual system. Both models produced classification and generalization performance in the 75%–80% range when classifying where experts tend to look. A validation study using regions identified by the AIM model as diagnostic demonstrates that human experts perform better when given regions of high diagnosticity. The computational nature of the metrics may help guard against wrongful convictions, as well as provide a quantitative measure of the strength of evidence in casework.     

Three-dimensional structure from long-range apparent motion

Experiments are reported which show that three-dimensional structure can be perceived from two-dimensional image motions carried by objects defined solely by the differences in binocular and/or temporal correlation (ie disparity or motion discontinuities). This demonstrates that the kinetic depth effect is independent of motion detection in the luminance domain and that its relevant input comes from detectors based on some form of identity preservation of objects or features over time, ie the long-range processes of apparent motion.     

Identifying the mechanisms underpinning recognition of structured sequences of action

We present three experiments to identify the specific information sources that skilled participants use to make recognition judgements when presented with dynamic, structured stimuli. A group of less skilled participants acted as controls. In all experiments, participants were presented with filmed stimuli containing structured action sequences. In a subsequent recognition phase, participants were presented with new and previously seen stimuli and were required to make judgements as to whether or not each sequence had been presented earlier (or were edited versions of earlier sequences). In Experiment 1, skilled participants demonstrated superior sensitivity in recognition when viewing dynamic clips compared with static images and clips where the frames were presented in a nonsequential, randomized manner, implicating the importance of motion information when identifying familiar or unfamiliar sequences. In Experiment 2, we presented normal and mirror-reversed sequences in order to distort access to absolute motion information. Skilled participants demonstrated superior recognition sensitivity, but no significant differences were observed across viewing conditions, leading to the suggestion that skilled participants are more likely to extract relative rather than absolute motion when making such judgements. In Experiment 3, we manipulated relative motion information by occluding several display features for the duration of each film sequence. A significant decrement in performance was reported when centrally located features were occluded compared to those located in more peripheral positions. Findings indicate that skilled participants are particularly sensitive to relative motion information when attempting to identify familiarity in dynamic, visual displays involving interaction between numerous features.     

Identifying the mechanisms underpinning recognition of structured sequences of action

We present three experiments to identify the specific information sources that skilled participants use to make recognition judgements when presented with dynamic, structured stimuli. A group of less skilled participants acted as controls. In all experiments, participants were presented with filmed stimuli containing structured action sequences. In a subsequent recognition phase, participants were presented with new and previously seen stimuli and were required to make judgements as to whether or not each sequence had been presented earlier (or were edited versions of earlier sequences). In Experiment 1, skilled participants demonstrated superior sensitivity in recognition when viewing dynamic clips compared with static images and clips where the frames were presented in a nonsequential, randomized manner, implicating the importance of motion information when identifying familiar or unfamiliar sequences. In Experiment 2, we presented normal and mirror-reversed sequences in order to distort access to absolute motion information. Skilled participants demonstrated superior recognition sensitivity, but no significant differences were observed across viewing conditions, leading to the suggestion that skilled participants are more likely to extract relative rather than absolute motion when making such judgements. In Experiment 3, we manipulated relative motion information by occluding several display features for the duration of each film sequence. A significant decrement in performance was reported when centrally located features were occluded compared to those located in more peripheral positions. Findings indicate that skilled participants are particularly sensitive to relative motion information when attempting to identify familiarity in dynamic, visual displays involving interaction between numerous features.     

Stereopsis from reversible and irreversible patterns

Eight subjects reported stereoscopic depth as a function of the magnitude and direction of disparity carried by a reversible grid or an irreversible solid surface. The former alternated between a diamond seen against a grating or a uniform rectangular grid pattern with stereoscopic depth reported only when the diamond was perceived. Depth was attenuated for the grid compared to the solid surface patterns. Disparity can be defined by perceptually extracted forms with depth present only when those forms are bounded by visible contours.     

Accelerating self-motion displays produce more compelling vection in depth

We examined the vection in depth induced when simulated random self-accelerations (jitter) and periodic self-accelerations (oscillation) were added to radial expanding optic flow (simulating constant-velocity forward self-motion). Contrary to the predictions of sensory-conflict theory frontal-plane jitter and oscillation were both found to significantly decrease the onsets and increase the speeds of vection in depth. Depth jitter and oscillation had lesser, but still significant, effects on the speed of vection in depth. A control experiment demonstrated that adding global perspective motion which simulated a constant-velocity frontal-plane self-motion had no significant effect on vection in depth induced by the radial component of the optic flow. These results are incompatible with the notion that constant-velocity displays produce optimal vection. Rather, they indicate that displays simulating self-acceleration can often produce more compelling experiences of self-motion in depth.     

Priming and Recognition of Human Motion Patterns

Left-right orientation and size incongruence is known to affect recognition memory for objects but not object priming. In the present study, the effects of study-test changes in left-right orientation and size on old-new recognition decisions and long-term priming of human motion patterns were examined. Experiment 1 showed effects of orientation incongruence on both recognition and priming. Experiment 2 showed an effectof size incongruence on recognition memory but not on priming. It is suggested that the representations of human actions that underlie human motion priming are on a level that preserve orientation, possibly because of the importance of dynamic information for perceiving motion patterns or because encoding of human motion is governed by a body schema (e.g. Reed & Farah, 1995). In contrast, low-level metric information such as size is inconsequential to priming because priming involves identification of shape, which is not affected by size transformations. The effect of size on recognition memory, on the other hand, shows thatexplicitrecognition decisions may draw on any available episodic information, including metric attributes, to make an old new discrimination.     

TO WHAT EXTENT DO UNIQUE PARTS INFLUENCE RECOGNITION ACROSS CHANGES IN VIEWPOINT?

Abstract— We investigated how varying the number of unique pans within an object influences recognition across changes in viewpoint The stimuli were shaded objects composed of five three-dimensional volumes tinted end to end with varying connection angles Of the five volumes, zero, one, three, or jive were qualitatively distinct (e g, brick vs cone), the rest being tubes Sequential-matching and naming tasks depth Three major results stand out First, regardless of the number of distinct parts, there was increasingly poorer recognition performance viewpoint change for objects with one unique part was less than that for the other objects Third, additional pans beyond a single unique part produced strong viewpoint dependency comparable to that ob representations encode both quantitative and qualitative features     

On Time and Deepening in Psychoanalysis

The metaphor of depth has most often been used in analytic thinking to denote extension in space or time. Depth, and its companion term deepening, may also be used to describe quality rather than distance; something is deep when it is serious or important. The deepening of analysis is linked to temporality but not in a linear way. Analysis deepens as it sets in motion a series of changes in the experience of time: The past comes alive and is worked through. Past and present come into a new relation with one another. The analytic frame and the rhythm of the exchange between patient and analyst also bring to life past experiences and fantasies to do with time. Both patient and analyst must enter a fluid time-state in order for deepening to occur. Clinical examples illustrate these dimensions of temporal experience and the way they emerge in the deepening analytic process.     

Salient stimuli in advertising: the effect of contrast interval length and type on recall

Salient auditory stimuli (e.g., music or sound effects) are commonly used in advertising to elicit attention. However, issues related to the effectiveness of such stimuli are not well understood. This research examines the ability of a salient auditory stimulus, in the form of a contrast interval (CI), to enhance recall of message-related information. Researchers have argued that the effectiveness of the CI is a function of the temporal duration between the onset and offset of the change in the background stimulus and the nature of this stimulus. Three experiments investigate these propositions and indicate that recall is enhanced, providing the CI is 3 s or less. Information highlighted with silence is recalled better than information highlighted with music.     

Action Interrupted: Movement and Breakpoints in the Processing of Motion Violations in Toddlers and Adults

From early in development, segmenting events unfolding in the world in meaningful ways renders input more manageable and facilitates interpretation and prediction. Yet, little is known about how children process action structure in events composed of multiple coarse-grained actions. More importantly, little is known about the time course of action processing in young children or about the specific features that recruit attention. This is particularly true when we consider action that pauses unexpectedly—as actions sometimes do—thereby violating the expectation of a continuous unfolding of motion. We assessed visual preference to intact and disrupted actions embedded within a multiaction event in toddlers and adults. In 1 condition, pauses were inserted at intact action boundaries, whereas in the other condition, they disrupted action. Attention in both groups was recruited to the disrupted relative to intact events. Time-course analyses, however, revealed developmental differences in sensitivity to the movement features (e.g., motion, pauses, and transitions) of disrupted events.     

Early motion processes and the kinetic depth effect

It has been known for over 30 years that motion information alone is sufficient to yield a vivid impression of three-dimensional object structure. For example, a computer simulation of a transparent sphere, the surface of which is randomly speckled with dots, gives no impression of depth when presented as a stationary pattern on a visual display. As soon as the sphere is made to rotate in a series of discrete steps or frames, its 3-D structure becomes apparent. Three experiments are described which use this stimulus, and find that depth perception in these conditions depends crucially on the spatial and temporal properties of the display:

1. Depth is seen reliably only for between-frame rotations of less than 15°, using two-frame and four-frame sequences.

2. Parametric observations using a wide range of frame durations and inter-frame intervals reveal that depth is seen only for inter-frame intervals below 80 msec and is optimal when the stimulus can be sampled at intervals of about 40-60 msec.

3. Monoptic presentation of two frames of the stimulus is sufficient to yield depth, but the impression is destroyed by dichoptic presentation.

These data are in close agreement with the observed limits of direction perception in experiments using “short-range” stimuli. It is concluded that depth perception in the motion display used in these experiments depends on the outputs of low-level or “short-range” motion detectors.     

Three-channel convolutional neural networks for vegetable leaf disease recognition

The color information of diseased leaf is the main basis for leaf based plant disease recognition. To make use of color information, a novel three-channel convolutional neural networks (TCCNN) model is constructed by combining three color components for vegetable leaf disease recognition. In the model, each channel of TCCNN is fed by one of three color components of RGB diseased leaf image, the convolutional feature in each CNN is learned and transmitted to the next convolutional layer and pooling layer in turn, then the features are fused through a fully connected fusion layer to get a deep-level disease recognition feature vector. Finally, a softmax layer makes use of the feature vector to classify the input images into the predefined classes. The proposed method can automatically learn the representative features from the complex diseased leaf images, and effectively recognize vegetable diseases. The experimental results validate that the proposed method outperforms the state-of-the-art methods of the vegetable leaf disease recognition.     

Memory for positive,negative and neutral events in younger and older adults: Does emotion influence binding in event memory?

When remembering an event, it is important to remember both the features of the event (e.g., a person and an action) and the connections among features (e.g., who performed which action). Emotion often enhances memory for stimulus features, but the relationship between emotion and the binding of features in memory is unclear. Younger and older adults attempted to remember events in which a person performed a negative, positive or neutral action. Memory for the action was enhanced by emotion, but emotion did not enhance the ability of participants to remember which person performed which action. Older adults were more likely than younger adults to make binding errors in which they incorrectly remembered a familiar actor performing a familiar action that had actually been performed by someone else, and this age-related associative deficit was found for both neutral and emotional actions. Emotion not only increased correct recognition of old events for older and younger adults but also increased false recognition of events in which a familiar actor performed a familiar action that had been performed by someone else. Thus, although emotion may enhance memory for the features of an event, it does not increase the accuracy of remembering who performed which action.