Virtual objects in augmented reality (AR) often appear to float atop real world surfaces, which makes it difficult to determine where they are positioned in space. This is problematic as many applications for AR require accurate spatial perception. In the current study, we examine how the way we render cast shadows-which act as an important monocular depth cue for creating a sense of contact between an object and the surface beneath it-impacts spatial perception. Over two experiments, we evaluate people's sense of surface contact given both traditional and non-traditional shadow shading methods in optical see-through augmented reality (OST AR), video see-through augmented reality (VST AR), and virtual reality (VR) head-mounted displays. Our results provide evidence that nontraditional shading techniques for rendering shadows in AR displays may enhance the accuracy of one's perception of surface contact. This finding implies a possible tradeoff between photorealism and accuracy of depth perception, especially in OST AR displays. However, it also supports the use of more stylized graphics like non-traditional cast shadows to improve perception and interaction in AR applications.

Augmented reality (AR) technologies have the potential to provide individuals with unique training and visualizations, but the effectiveness of these applications may be influenced by users' perceptions of the distance to AR objects. Perceived distances to AR objects may be biased if these objects do not appear to make contact with the ground plane. The current work compared distance judgments of AR targets presented on the ground versus off the ground when no additional AR depth cues, such as shadows, were available to denote ground contact. We predicted that without additional information for height off the ground, observers would perceive the off-ground objects as placed on the ground, but at farther distances. Furthermore, this bias should be exaggerated when targets were viewed with one eye rather than two. In our experiment, participants judged the absolute egocentric distance to various cubes presented on or off the ground with an action-based measure, blind walking. We found that observers walked farther for off-ground AR objects and that this effect was exaggerated when participants viewed off-ground objects with monocular vision compared to binocular vision. However, we also found that the restriction of binocular cues influenced participants' distance judgments for on-ground AR objects. Our results suggest that distances to off-ground AR objects are perceived differently than on-ground AR objects and that the elimination of binocular cues further influences how users perceive these distances.

The use of first-person self-avatars in immersive virtual environments (VEs) has grown over recent years. It is unknown, however, how visual feedback from a self-avatar influences a user's online actions and subsequent calibration of actions within an immersive VE. The current paper uses a prism throwing adaptation paradigm to test the role of a self-avatar arm or full body on action calibration in a VE. Participants' throwing accuracy to a target on the ground was measured first in a normal viewing environment, then with the visual field rotated clockwise about their vertical axis by 17° (prism simulation), and then again in the normal viewing environment with the prism distortion removed. Participants experienced either no-avatar, a first-person avatar arm and hand, or a first-person full body avatar during the entire experimental session, in a between-subjects manipulation. Results showed similar throwing error and adaptation during the prism exposure for all conditions, but a reduced aftereffect (displacement with respect to the target in the opposite direction of the prism-exposure) when the avatar arm or full body was present. The results are discussed in the context of how an avatar can provide a visual frame of reference to aid in action calibration.

Data ensembles are often used to infer statistics to be used for a summary display of an uncertain prediction. In a spatial context, these summary displays have the drawback that when uncertainty is encoded via a spatial spread, display glyph area increases in size with prediction uncertainty. This increase can be easily confounded with an increase in the size, strength or other attribute of the phenomenon being presented. We argue that by directly displaying a carefully chosen subset of a prediction ensemble, so that uncertainty is conveyed implicitly, such misinterpretations can be avoided. Since such a display does not require uncertainty annotation, an information channel remains available for encoding additional information about the prediction. We demonstrate these points in the context of hurricane prediction visualizations, showing how we avoid occlusion of selected ensemble elements while preserving the spatial statistics of the original ensemble, and how an explicit encoding of uncertainty can also be constructed from such a selection. We conclude with the results of a cognitive experiment demonstrating that the approach can be used to construct storm prediction displays that significantly reduce the confounding of uncertainty with storm size, and thus improve viewers' ability to estimate potential for storm damage.

Perceiving an accurate sense of absolute scale is important for the utility of virtual environments (VEs). Research shows that absolute egocentric distances are underestimated in VEs compared to the same judgments made in the real world, but there are inconsistencies in the amount of underestimation. We examined two possible factors in the variation in the magnitude of distance underestimation. We compared egocentric distance judgments in a high-cost (NVIS SX60) and low-cost (Oculus Rift DK2) HMD using both indoor and outdoor highly-realistic virtual models. Performance more accurately matched the intended distance in the Oculus compared to the NVIS, and regardless of the HMD, distances were underestimated more in the outdoor versus the indoor VE. These results suggest promise in future use of consumer-level wide field-of-view HMDs for space perception research and applications, and the importance of considering the context of the environment as a factor in the perception of absolute scale within VEs.

Many have experienced vastness, the feeling when the visual space seems to extend without limits away from you, making you feel like a small element within the space. For over 200 years, people have been writing about this experience, for example stating that vastness is important to the experience of awe [Mikulak 2015]. Yet vastness has received little attention in empirical research. Specifically, it is unknown which aspects of the visual stimulus contribute to perceived vastness. This may be due to the inherent difficulties in presenting a variety of vast stimuli while varying only specific visual cues. Using virtual reality addresses these difficulties, as this technology provides precise control over the presented visual stimuli. Here we investigate whether the feeling of vastness can be evoked using virtual reality and explore potential objective measures to assess vastness. We used three different measures during this experiment: 1) An avatar height adjustment task where participants had to adjust an avatar to be equivalent to their own height as viewed from a distance, 2) a distance estimation task and 3) a subjective vastness rating task. These tasks were performed in four environments: a plain (used in all subsequent environments for the ground and sky surfaces), a forest, a mountain and the mountain and forest environments combined. Our results indicate that the feeling of vastness can indeed be experienced to various degrees in virtual environments, demonstrating the potential of VR as a tool for exploring the perception of vastness. Yet the results combined suggest that the percept of vastness is a rather complex construct.

Spatial perception research in the real world and in virtual environments suggests that the body (e.g., hands) plays a role in the perception of the scale of the world. However, little research has closely examined how varying the size of virtual body parts may influence judgments of action capabilities and spatial layout. Here, we questioned whether changing the size of virtual feet would affect judgments of stepping over and estimates of the width of a gap. Participants viewed their disembodied virtual feet as small or large and judged both their ability to step over a gap and the size of gaps shown in the virtual world. Foot size affected both affordance judgments and size estimates such that those with enlarged virtual feet estimated they could step over larger gaps and that the extent of the gap was smaller. Shrunken feet led to the perception of a reduced ability to step over a gap and smaller estimates of width. The results suggest that people use their visually perceived foot size to scale virtual spaces. Regardless of foot size, participants felt that they owned the feet rendered in the virtual world. Seeing disembodied, but motion-tracked, virtual feet affected spatial judgments, suggesting that the presentation of a single tracked body part is sufficient to produce similar effects on perception, as has been observed with the presence of fully co-located virtual self-avatars or other body parts in the past.

The use of third-person self-avatars has become common in non-immersive virtual environments, due in part to the advantages of seeing a full body while moving around in space. However, little research has investigated how third-person self-avatars can benefit task performance in immersive virtual environments (IVEs). This paper explores a procedure that may enhance user ability to act in an IVE from the point of view of a self-avatar projected in front of the viewer. Such abilities could be crucial in applications such as remote operator control, when a third person view is necessary or is the only view available. We conducted an experiment to assess the role of dynamic visual-motor and haptic feedback with a third-person avatar on a user's ability to point to remembered targets from the location of the avatar. The dynamic condition was compared within-subjects to a static condition in which the user's movements were not tracked with the avatar but the same task was required. Overall, participants were relatively good at pointing to targets from the location of the avatar. However, participants in the dynamic condition showed pointing performance that corresponded more closely to the avatar's location than was the case in the static condition. Subjective reports of bodily self-perception were also included and showed increased self-identification, presence, and agency in the dynamic compared to the static condition.

Gesture-based controllers such as the Microsoft Kinect are low cost devices that allow a user to interact with complex, three-dimensional simulations using an interface argued to be more natural than game controllers, joy sticks, or a mouse and keyboard. This paper presents a controlled experimental evaluation of the use of Microsoft Kinect to support a 3D object manipulation task. Users were asked to match the orientation of objects with a manipulation interface that displayed either a self-avatar hand and arm or a sphere, both corresponding to users' arm gestures and wrist rotation. Our results show that while there was no overall difference in performance between the self-avatar and sphere visual display conditions, there were clear differences in the two visual display conditions as a function of gender and video-game experience.

Accurate perception of the size of 3D objects depicted on 2D desktop displays is important for many applications. Whether users perceive objects depicted on a display to be the same size as comparable real world objects is not well understood. We propose using affordances judgments as a way of measuring the perceived size of objects depicted in desktop virtual environments and the real world. The methodology involves indicating whether or not a particular action can be performed in a given environment, making it a flexible measure that can be used across different display technologies. In two studies, we test users' perceptions of size by asking them to make affordance judgments in both the real world and a geometrically matched desktop virtual environment. In the first study, users judge whether they can grasp an object and in the second study, they judge whether they can fit their hand through an opening. In both experiments we show that users perceive the size of objects in the desktop virtual environment to be smaller than in the real world.

Understanding how users perceive 3D geometric objects can provide a basis for creating more effective tools for visualization in applications such as CAD or medical imaging. This article examines how orientation indicators affect users' accuracy in perceiving the shape of a 3D object shown as multiple views. Multiple views force users to infer the orientation of an object and recognize corresponding features between distinct vantage points. These are difficult tasks, and not all users are able to carry them out accurately. We use a cognitive experimental paradigm to evaluate the effectiveness of two types of orientation indicators on a person's ability to compare views of objects presented in different orientations. The orientation indicators implemented were colocated, which shared a center-point with the 3D object, or noncolocated with (displaced from) the 3D object. The study accounts for additional factors including object complexity, axis of rotation, and users' individual differences in spatial abilities. Our results show that an orientation indicator helps users in comparing multiple views, and that the effect is influenced by the type of aid, a person's spatial ability, and the difficulty of the task. In addition to establishing an effect of an orientation indicator, this article helps demonstrate the application of a particular experimental paradigm and analysis, as well as the importance of considering individual differences when designing interface aids.

Most head-mounted displays (HMDs) suffer from substantial optical distortion, and vendor-supplied specifications for field-of-view often are at variance with reality. Unless corrected, such displays do not present perspective-related visual cues in a geometrically correct manner. Distorted geometry has the potential to affect applications of HMDs, which depend on precise spatial perception. This article provides empirical evidence for the degree to which common geometric distortions affect one type of spatial judgment in virtual environments. We show that minification or magnification in the HMD that would occur from misstated HMD field of view causes significant changes in distance judgments. Incorrectly calibrated pitch and pincushion distortion, however, do not cause statistically significant changes in distance judgments for the degree of distortions examined. While the means for determining the optical distortion of display systems are well known, they are often not used in non-see-through HMDs due to problems in measuring and correcting for distortion. As a result, we also provide practical guidelines for creating geometrically calibrated systems.

Research has shown that people are able to judge distances accurately in full-cue, real-world environments using visually directed actions. However, in virtual environments viewed with head-mounted display (HMD) systems, there is evidence that people act as though the virtual space is smaller than intended. This is a surprising result given how well people act in real environments. The behavior in the virtual setting may be linked to distortions in the available visual cues or to a person's ability to locomote without vision. Either could result from issues related to added mass, moments of inertia, and restricted field of view in HMDs. This article describes an experiment in which distance judgments based on normal real-world and HMD viewing are compared with judgments based on real-world viewing while wearing two specialized devices. One is a mock HMD, which replicated the mass, moments of inertia, and field of view of the HMD and the other an inertial headband designed to replicate the mass and moments of inertia of the HMD, but constructed to not restrict the field of view of the observer or otherwise feel like wearing a helmet. Distance judgments using the mock HMD showed a statistically significant underestimation relative to the no restriction condition but not of a magnitude sufficient to account for all the distance compression seen in the HMD. Indicated distances with the inertial headband were not significantly smaller than those made with no restrictions.

Panorama maps are aerial view paintings that depict complex, three-dimensional landscapes in a pleasing and understandable way. Painters and cartographers have developed techniques to create such artistic landscapes for centuries, but the process remains difficult and time-consuming. In this work, we derive principles and heuristics for panorama map creation of mountainous terrain from a perceptual and artistic analysis of two panorama maps of Yellowstone National Park. We then present methods to automatically produce landscape renderings in the visual style of the panorama map. Our algorithms rely on United States Geological Survey (USGS) terrain and classification data. Our surface textures are generated using perceptual metrics and artistic considerations, and use the structural information present in the terrain to guide the automatic placement of image space strokes for natural surfaces such as forests, cliffs, snow, and water.

This work uses an immersive virtual environment (IVE) to examine how people maintain a calibration between biomechanical and visual information for rotational self-motion. First, we show that no rotational recalibration occurs when visual and biomechanical rates of rotation are matched. Next, we demonstrate that mismatched physical and visual rotation rates cause rotational recalibration. Although previous work has shown that rotational locomotion can be recalibrated in real environments, this work extends the finding to virtual environments. We further show that people do not completely recalibrate left and right rotations independently when different visual--biomechanical discrepancies are used for left and right rotations during a recalibration phase. Finally, since the majority of participants did not notice mismatched physical and visual rotation rates, we discuss the implications of using such mismatches to enable IVE users to explore a virtual space larger than the physical space they are in.

This paper describes the use of a treadmill-based virtual environment (VE) to investigate the influence of visual motion on locomotion. First, we establish that a computer-controlled treadmill coupled with a wide field of view computer graphics display can be used to study interactions between perception and action. Previous work has demonstrated that humans recalibrate their visually directed actions to changing circumstances in their environment. Using a treadmill VE, we show that recalibration of action is reflected in the real world as a result of manipulating the relation between the visual indication of speed, presented using computer graphics, and the biomechanical speed of walking on a treadmill. We then extend this methodology to investigate whether the recalibration is based on perception of the speed of movement through the world or on the magnitude of optic flow itself. This was done by utilizing two different visual displays, which had essentially the same magnitude of optic flow, but which differed in the information present for the speed of forward motion. These results indicate that changes in optic flow are not necessary for recalibration to occur. The recalibration effect is dependent, at least in part, on visual perception of the speed of self-movement.

No abstract available.

For humans to effectively interact with their environment, it is important for the visual system to determine the absolute size and distance of objects. Previous experiments performed in full-cue, real-world environments have demonstrated that blind walking to targets serves as an accurate indication of distance perception, up to about 25 m. In contrast, the same task performed in virtual environments (VEs) using head-mounted displays shows significant underestimation in walking. To date, blind walking is the only visually directed action task that has been used to evaluate distance perception in VEs beyond reaching distances. The possible influence of the response measure itself on absolute distance perception in virtual environments is currently an open question. Blind walking involves locomotion and the egocentric updating of the environment with one's own movement. We compared this measure to blind throwing, a task that involves the initiation of a movement directed by vision, but no further interaction within the environment. Both throwing and walking were compressed in the VE but accurate in the real world. We suggest that distance compression found in VEs may be a result of a general perceptual origin rather than specific to the response measure.

Distance judgments are difficult in current virtual environments, limiting their effectiveness in conveying spatial information. This problem is apparent when contact occurs while a user is manipulating objects. In particular, the computer graphics used to support current-generation immersive interfaces does a poor job of providing the visual cues necessary to perceive when contact between objects is about to occur. This perception of imminent contact is important in human motor control. Its absence prevents a sense of naturalness in interactive displays which allow for object manipulation. This paper reports results from an experiment evaluating the effectiveness of binocular disparity, cast shadows and diffuse inter-reflections in signaling imminent contact in a manipulation task.

Reverse engineering of mechanical parts requires extraction of information about an instance of a particular part sufficient to replicate the part using appropriate manufacturing techniques. This is important in a wide variety of situations, since functional CAD models are often unavailable or unusable for parts which must be duplicated or modified. Computer vision techniques applied to three-dimensional (3-D) data acquired using noncontact, 3-D position digitizers have the potential for significantly aiding the process. Serious challenges must be overcome, however, if sufficient accuracy is to be obtained and if models produced from sensed data are to be truly useful for manufacturing operations. The paper describes a prototype of a reverse engineering system which uses manufacturing features as geometric primitives. This approach has two advantages over current practice. The resulting models can be directly imported into feature-based CAD systems without loss of the semantics and topological information inherent in feature-based representations. In addition, the feature-based approach facilitates methods capable of producing highly accurate models, even when the original 3-D sensor data has substantial errors.

We propose to deploy and exploit a large number of inexpensive sensors to obtain information or trigger actions over a wide geographic area. Sensors may be of diverse physical natures: acoustic, IR, seismic, chemical, magnetic, thermal, etc. We describe three major issues: (1) sensor distribution patterns, (2) local sensor frames, and (3) autonomous robot sensor snow exploitation techniques.

For an autonomous robot navigating in an unstructured outdoor environment, staying close to a path is crucial to successfully reaching its goal. Although the degree of accuracy with which it estimates its own location affects its ability to stay on the path, accuracy in estimate of lateral distance from the path is far more important for successful navigation than accuracy in estimate of position along the path. Utilizing methods based only on relative angular measurements between landmarks in the environment, we draw from techniques used in statistical pattern recognition to show how landmarks can be chosen for localization which will not only give good estimate of location in spite of the measurement error, but will also keep the robot on the path. We demonstrate how identical landmark configurations can produce very different results in localizing to a path and show how simple heuristics can be used to choose the best configuration for path localization.>

A robot navigating in an unstructured outdoor environment must determine its own location in spite of problems due to environmental conditions, sensor limitations and map inaccuracies, exact measurements are seldom known, and the combination of approximate measures can lead to large errors in self-localization. The conventional approach to this problem has been to deal with the errors either during processing or after they occur. The authors maintain that it is possible to limit the errors before they occur. The authors analyze how measurement errors affect errors in localization and propose that a simple algorithm can be used to exploit the geometric properties of landmarks in the environment in order to decrease errors in localization. The authors' goal is to choose landmarks that will provide the best localization regardless of measurement error, determine the best areas in which to identify new landmarks to be used for further localization and choose paths that will provide the least chance of "straying". The authors show the result of implementing this concept in experiments run in simulation with USGS 30 m DEM data for a robot statically locating, following a path and identifying new landmarks.>

A method for visually detecting moving objects from a moving camera using point correspondences in two orthographic views is described. The method applies a simple structure-from-motion analysis and then identifies those points inconsistent with the interpretation of the scene as a single rigid object. It is effective even when the actual motion parameters cannot be recovered. Demonstrations are presented using point correspondences automatically determined from real image sequences.>

A method is presented for recovering the rotation of a moving object using both velocity and acceleration. Acceleration is a useful form of information to incorporate into a structure-from motion system. It allows the use of clustering algorithms for recovering the motion parameters of a rotating object, which in turn allows the weakening of the spatial continuity constraint to a spatial coherence constraint, together with a temporal continuity constraint. This allows the integration of information over both time and space without limiting assumptions about surface structure.>

Multiple views of a scene can provide important information about the structure and dynamic behavior of three-dimensional objects. Many of the methods that recover this information require the determination of optical flow-the velocity, on the image, of visible points on object surfaces. An important class of techniques for estimating optical flow depend on the relationship between the gradients of image brightness. While gradient-based methods have been widely studied, little attention has been paid to accuracy and reliability of the approach. Gradient-based methods are sensitive to conditions commonly encountered in real imagery. Highly textured surfaces, large areas of constant brightness, motion boundaries, and depth discontinuities can all be troublesome for gradient-based methods. Fortunately, these problematic areas are usually localized can be identified in the image. In this paper we examine the sources of errors for gradient-based techniques that locally solve for optical flow. These methods assume that optical flow is constant in a small neighborhood. The consequence of violating in this assumption is examined. The causes of measurement errors and the determinants of the conditioning of the solution system are also considered. By understanding how errors arise, we are able to define the inherent limitations of the technique, obtain estimates of the accuracy of computed values, enhance the performance of the technique, and demonstrate the informative value of some types of error.

An important application of machine vision systems is the recognition of known three-dimensional objects. A major difficulty arises when two or more objects project the same or similar two-dimensional image, often resulting in misclassification and degradation of system performance. The changes in images which result from the motion of objects provide a source of three-dimensional information which can greatly aid the classification process, but this three-dimensional analysis is computationally complex and subject to many sources of error. This work develops a methodology which utilizes the information derived from the apparent changes in object features over time to facilitate the recognition task, without the need to actually recover the three-dimensional structure of the objects under view. The basic approach is to generate a ``feature signature'' by combining the feature measurements of the individual regions in a long sequence of images. The static information in the individual frames is analyzed along with the temporal information from the entire sequence. These techniques are particularly applicable in situations where static image processing methods cannot discriminate between ambiguous objects. Two example implementations are presented to illustrate the application of the techniques of object recognition using motion information.

In dynamic scenes, the presence of object boundaries is often signaled by the appearance or disappearance of occluded surfaces over time. Such regions of surface accretion or deletion can be found using matching techniques similar to those used to determine optical flow in an image sequence. Regions in one frame that are not adequately matched by any region in previous frames correspond to accretion. Regions that have no matches in subsequent frames correspond to deletion. In either case, an occlusion boundary is present. Furthermore, by associating accretion or deletion regions with a surface on one side of a boundary, it is possible to determine which side of the boundary is being occluded. This association can be based purely on visual motion-the accretion or deletion region moves with the same image velocity as the remaining visible surface to which it is attached.

Optical flow can be used to locate dynamic occlusion boundaries in an image sequence. We derive an edge detection algorithm sensitive to changes in flow fields likely to be associated with occlusion. The algorithm is patterned after the Marr-Hildreth zero-crossing detectors currently used to locate boundaries in scalar fields. Zero-crossing detectors are extended to identify changes in direction and/or magnitude in a vector-valued flow field. As a result, the detector works for flow boundaries generated due to the relative motion of two overlapping surfaces, as well as the simpler case of motion parallax due to a sensor moving through an otherwise stationary environment. We then show how the approach can be extended to identify which side of a dynamic occlusion boundary corresponds to the occluding surface. The fundamental principal involved is that at an occlusion boundary, the image of the surface boundary moves with the image of the occluding surface. Such information is important in interpreting dynamic scenes. Results are demonstrated on optical flow fields automatically computed from real image sequences.

An algorithm for matching images of real world scenes is presented. The matching is a specification of the geometrical disparity between the images and may be used to partially reconstruct the three-dimensional structure of the scene. Sets of candidate matching points are selected independently in each image. These points are the locations of small, distinct features which are likely to be detectable in both images. An initial network of possible matches between the two sets of candidates is constructed. Each possible match specifies a possible disparity of a candidate point in a selected reference image. An initial estimate of the probability of each possible disparity is made, based on the similarity of subimages surrounding the points. These estimates are iteratively improved by a relaxation labeling technique making use of the local continuity property of disparity that is a consequence of the continuity of real world surfaces. The algorithm is effective for binocular parallax, motion parallax, and object motion. It quickly converges to good estimates of disparity, which reflect the spatial organization of the scene.

A procedure is demonstrated for locating textural boundaries in the digital image representation of a natural scene. The technique involves development of an edge operator capable of integrating multiple textural features into a single boundary determination. The process is designed to simulate actual perception of textural discontinuities. Success of the system is demonstrated on pictures with prominent perceived boundaries not detectable by methods based only on differences in average brightness.

Computer image processing techniques have been applied to particle counting and sizing in electron microscope images. Distributions of particle sizes were computed for several images and compared to manually computed distributions. The results of these experiments indicate that automatic particle counting within a reasonable error and computer processing time is feasible. The significance of the results is that the tedious task of manually counting a large number of particles can be eliminated while still providing the scientist with accurate results.

A central problem in the area of scene analysis is that of segmenting a scene into its natural objects. Current work emphasizes the semantic approach in which a priori knowledge of the shape of an object is used. Yet there is much to learn about more primitive cues for segmentation such as texture, color, and brightness. In the case of human perception, segmentation appears to be due to a multiplicity of cues which operate in a redundant fashion.

The advent of increased government involvement in occupational health maintenance with compensation for affected individuals is requiring new approaches to medical decision making. One aspect of this involvement will perhaps include the automatic mass diagnostic screening of medical films for the detection of a specffic abnormality with an occupational etiology. The results of two complementary approaches for performing diagnostic screening for presence and profusion of coal worker's pneumoconiosis from the routine posterior-anterior chest radiograph are presented. The first is a digital approach utilizing the measurement of image texture, while the second uses hybrid optical-digital methods involving the optical Fourier tramsform. Both approaches yielded classification results comparable to experienced radiologists.