Visual surveillance for moving vehicles

An overview is given of a vision system for locating, recognising and tracking multiple vehicles, using an image sequence taken by a single camera mounted on a moving vehicle. The camera motion is estimated by matching features on the ground plane from one image to the next. Vehicle detection and hypothesis generation are performed using template correlation and a 3D wire frame model of the vehicle is fitted to the image. Once detected and identified, vehicles are tracked using dynamic filtering. A separate batch mode filter obtains the 3D trajectories of nearby vehicles over an extended time. Results are shown for a motorway image sequence.

Visual surveillance using deformable models of vehicles

This paper presents recent developments to a vision-based traffic surveillance system which relies extensively on the use of geometrical and scene context. Firstly, a highly parametrised 3-D model is reported, able to adopt the shape of a wide variety of different classes of vehicle (e.g. cars, vans, buses etc.), and its subsequent specialisation to a generic car class which accounts for commonly encountered types of car (including saloon, batchback and estate cars). Sample data collected from video images, by means of an interactive tool, have been subjected to principal component analysis (PCA) to define a deformable model having 6 degrees of freedom. Secondly, a new pose refinement technique using “active” models is described, able to recover both the pose of a rigid object, and the structure of a deformable model; an assessment of its performance is examined in comparison with previously reported “passive” model-based techniques in the context of traffic surveillance. The new method is more stable, and requires fewer iterations, especially when the number of free parameters increases, but shows somewhat poorer convergence. Typical applications for this work include robot surveillance and navigation tasks.

Driving saccade to pursuit using image motion

Within the context of active vision, scant attention has been paid to the execution of motion saccades—rapid re-adjustments of the direction of gaze to attend to moving objects. In this paper we first develop a methodology for, and give real-time demonstrations of, the use of motion detection and segmentation processes to initiate capture saccades towards a moving object. The saccade is driven by both position and velocity of the moving target under the assumption of constant target velocity, using prediction to overcome the delay introduced by visual processing. We next demonstrate the use of a first order approximation to the segmented motion field to compute bounds on the time-to-contact in the presence of looming motion. If the bound falls below a safe limit, a panic saccade is fired, moving the camera away from the approaching object. We then describe the use of image motion to realize smooth pursuit, tracking using velocity information alone, where the camera is moved so as to null a single constant image motion fitted within a central image region. Finally, we glue together capture saccades with smooth pursuit, thus effecting changes in both what is being attended to and how it is being attended to. To couple the different visual activities of waiting, saccading, pursuing and panicking, we use a finite state machine which provides inherent robustness outside of visual processing and provides a means of making repeated exploration. We demonstrate in repeated trials that the transition from saccadic motion to tracking is more likely to succeed using position and velocity control, than when using position alone.

Near real-time flood detection in urban and rural areas using high resolution Synthetic Aperture Radar images

A near real-time flood detection algorithm giving a synoptic overview of the extent of flooding in both urban and rural areas, and capable of working during night-time and day-time even if cloud was present, could be a useful tool for operational flood relief management. The paper describes an automatic algorithm using high resolution Synthetic Aperture Radar (SAR) satellite data that builds on existing approaches, including the use of image segmentation techniques prior to object classification to cope with the very large number of pixels in these scenes. Flood detection in urban areas is guided by the flood extent derived in adjacent rural areas. The algorithm assumes that high resolution topographic height data are available for at least the urban areas of the scene, in order that a SAR simulator may be used to estimate areas of radar shadow and layover. The algorithm proved capable of detecting flooding in rural areas using TerraSAR-X with good accuracy, and in urban areas with reasonable accuracy. The accuracy was reduced in urban areas partly because of TerraSAR-X’s restricted visibility of the ground surface due to radar shadow and layover.

Benchmark datasets for detection and tracking

There is a rising demand for the quantitative performance evaluation of automated video surveillance. To advance research in this area, it is essential that comparisons in detection and tracking approaches may be drawn and improvements in existing methods can be measured. There are a number of challenges related to the proper evaluation of motion segmentation, tracking, event recognition, and other components of a video surveillance system that are unique to the video surveillance community. These include the volume of data that must be evaluated, the difficulty in obtaining ground truth data, the definition of appropriate metrics, and achieving meaningful comparison of diverse systems. This chapter provides descriptions of useful benchmark datasets and their availability to the computer vision community. It outlines some ground truth and evaluation techniques, and provides links to useful resources. It concludes by discussing the future direction for benchmark datasets and their associated processes.

Near real-time flood detection in urban and rural areas using high resolution Synthetic Aperture Radar images

A near real-time flood detection algorithm giving a synoptic overview of the extent of flooding in both urban and rural areas, and capable of working during night-time and day-time even if cloud was present, could be a useful tool for operational flood relief management. The paper describes an automatic algorithm using high resolution Synthetic Aperture Radar (SAR) satellite data that builds on existing approaches, including the use of image segmentation techniques prior to object classification to cope with the very large number of pixels in these scenes. Flood detection in urban areas is guided by the flood extent derived in adjacent rural areas. The algorithm assumes that high resolution topographic height data are available for at least the urban areas of the scene, in order that a SAR simulator may be used to estimate areas of radar shadow and layover. The algorithm proved capable of detecting flooding in rural areas using TerraSAR-X with good accuracy, classifying 89% of flooded pixels correctly, with an associated false positive rate of 6%. Of the urban water pixels visible to TerraSAR-X, 75% were correctly detected, with a false positive rate of 24%. If all urban water pixels were considered, including those in shadow and layover regions, these figures fell to 57% and 18% respectively.

Visual ambiguity in the oeuvre of the Gela Painter: a new lekythos from Thessaly

This paper presents a previously unpublished Attic lekythos and discusses visual ambiguity as an intentional drawing style used by a vase painter who conceptualised the many possible relationships between pot and user, object and subject. The Gela Painter endowed this hastily manufactured and decorated lekythos with visual effects that drew the viewer into an inherently ambivalent motif: a mounting Dionysos. This motif, like other Dionysian themes, had a vogue in late Archaic times but did not necessarily invoke chthonic associations. It had the potential to be consumed in diverse contexts, including religious festivals, by a wide range of audiences. Such images were not given to the viewer fully through visual perception but through interpretation.

Keeping the world at hand: rapid visuomotor processing for hand-object interactions

The existence of hand-centred visual processing has long been established in the macaque premotor cortex. These hand-centred mechanisms have been thought to play some general role in the sensory guidance of movements towards objects, or, more recently, in the sensory guidance of object avoidance movements. We suggest that these hand-centred mechanisms play a specific and prominent role in the rapid selection and control of manual actions following sudden changes in the properties of the objects relevant for hand-object interactions. We discuss recent anatomical and physiological evidence from human and non-human primates, which indicates the existence of rapid processing of visual information for hand-object interactions. This new evidence demonstrates how several stages of the hierarchical visual processing system may be bypassed, feeding the motor system with hand-related visual inputs within just 70 ms following a sudden event. This time window is early enough, and this processing rapid enough, to allow the generation and control of rapid hand-centred avoidance and acquisitive actions, for aversive and desired objects, respectively

Microsaccades distinguish between global and local visual processing

Much is known about the functional mechanisms involved in visual search. Yet, the fundamental question of whether the visual system can perform different types of visual analysis at different spatial resolutions still remains unsettled. In the visual-attention literature, the distinction between different spatial scales of visual processing corresponds to the distinction between distributed and focused attention. Some authors have argued that singleton detection can be performed in distributed attention, whereas others suggest that even such a simple visual operation involves focused attention. Here we showed that microsaccades were spatially biased during singleton discrimination but not during singleton detection. The results provide support to the hypothesis that some coarse visual analysis can be performed in a distributed attention mode.

Dissociating between object affordances and spatial compatibility effects using early response components

Perception and action are tightly linked: objects may be perceived not only in terms of visual features, but also in terms of possibilities for action. Previous studies showed that when a centrally located object has a salient graspable feature (e.g., a handle), it facilitates motor responses corresponding with the feature's position. However, such so-called affordance effects have been criticized as resulting from spatial compatibility effects, due to the visual asymmetry created by the graspable feature, irrespective of any affordances. In order to dissociate between affordance and spatial compatibility effects, we asked participants to perform a simple reaction-time task to typically graspable and non-graspable objects with similar visual features (e.g., lollipop and stop sign). Responses were measured using either electromyography (EMG) on proximal arm muscles during reaching-like movements, or with finger key-presses. In both EMG and button press measurements, participants responded faster when the object was either presented in the same location as the responding hand, or was affordable, resulting in significant and independent spatial compatibility and affordance effects, but no interaction. Furthermore, while the spatial compatibility effect was present from the earliest stages of movement preparation and throughout the different stages of movement execution, the affordance effect was restricted to the early stages of movement execution. Finally, we tested a small group of unilateral arm amputees using EMG, and found residual spatial compatibility but no affordance, suggesting that spatial compatibility effects do not necessarily rely on individuals’ available affordances. Our results show dissociation between affordance and spatial compatibility effects, and suggest that rather than evoking the specific motor action most suitable for interaction with the viewed object, graspable objects prompt the motor system in a general, body-part independent fashion

Note: Active optical detection of cloud from a balloon platform

A disposable backscatter instrument is described for optical detection of cloud in the atmosphere from a balloon-carried platform. It uses an ultra-bright light emitting diode (LED) illumination source with a photodiode detector. Scattering of the LED light by cloud droplets generates a small optical signal which is separated from background light fluctuations using a lock-in technique. The signal to noise obtained permits cloud detection using the scattered LED light, even in daytime. The response is interpreted in terms of the equivalent visual range within the cloud. The device is lightweight (150 g) and low power (∼30 mA), for use alongside a conventional meteorological radiosonde.

Instantaneous threat detection based on a semantic representation of activities, zones and trajectories

Threat detection is a challenging problem, because threats appear in many variations and differences to normal behaviour can be very subtle. In this paper, we consider threats on a parking lot, where theft of a truck’s cargo occurs. The threats range from explicit, e.g. a person attacking the truck driver, to implicit, e.g. somebody loitering and then fiddling with the exterior of the truck in order to open it. Our goal is a system that is able to recognize a threat instantaneously as they develop. Typical observables of the threats are a person’s activity, presence in a particular zone and the trajectory. The novelty of this paper is an encoding of these threat observables in a semantic, intermediate-level representation, based on low-level visual features that have no intrinsic semantic meaning themselves. The aim of this representation was to bridge the semantic gap between the low-level tracks and motion and the higher-level notion of threats. In our experiments, we demonstrate that our semantic representation is more descriptive for threat detection than directly using low-level features. We find that a person’s activities are the most important elements of this semantic representation, followed by the person’s trajectory. The proposed threat detection system is very accurate: 96.6 % of the tracks are correctly interpreted, when considering the temporal context.

Reverse correlation reveals how observers sample visual information when estimating three-dimensional shape

Human observers exhibit large systematic distance-dependent biases when estimating the three-dimensional (3D) shape of objects defined by binocular image disparities. This has led some to question the utility of disparity as a cue to 3D shape and whether accurate estimation of 3D shape is at all possible. Others have argued that accurate perception is possible, but only with large continuous perspective transformations of an object. Using a stimulus that is known to elicit large distance-dependent perceptual bias (random dot stereograms of elliptical cylinders) we show that contrary to these findings the simple adoption of a more naturalistic viewing angle completely eliminates this bias. Using behavioural psychophysics, coupled with a novel surface-based reverse correlation methodology, we show that it is binocular edge and contour information that allows for accurate and precise perception and that observers actively exploit and sample this information when it is available.

Environmental factors and features that influence visual search in a 3D WIMP interface

The challenge of moving past the classic Window Icons Menus Pointer (WIMP) interface, i.e. by turning it ‘3D’, has resulted in much research and development. To evaluate the impact of 3D on the ‘finding a target picture in a folder’ task, we built a 3D WIMP interface that allowed the systematic manipulation of visual depth, visual aides, semantic category distribution of targets versus non-targets; and the detailed measurement of lower-level stimuli features. Across two separate experiments, one large sample web-based experiment, to understand associations, and one controlled lab environment, using eye tracking to understand user focus, we investigated how visual depth, use of visual aides, use of semantic categories, and lower-level stimuli features (i.e. contrast, colour and luminance) impact how successfully participants are able to search for, and detect, the target image. Moreover in the lab-based experiment, we captured pupillometry measurements to allow consideration of the influence of increasing cognitive load as a result of either an increasing number of items on the screen, or due to the inclusion of visual depth. Our findings showed that increasing the visible layers of depth, and inclusion of converging lines, did not impact target detection times, errors, or failure rates. Low-level features, including colour, luminance, and number of edges, did correlate with differences in target detection times, errors, and failure rates. Our results also revealed that semantic sorting algorithms significantly decreased target detection times. Increased semantic contrasts between a target and its neighbours correlated with an increase in detection errors. Finally, pupillometric data did not provide evidence of any correlation between the number of visible layers of depth and pupil size, however, using structural equation modelling, we demonstrated that cognitive load does influence detection failure rates when there is luminance contrasts between the target and its surrounding neighbours. Results suggest that WIMP interaction designers should consider stimulus-driven factors, which were shown to influence the efficiency with which a target icon can be found in a 3D WIMP interface.

Automatic attention lateral asymmetry in visual discrimination tasks

There is evidence that automatic visual attention favors the right side. This study investigated whether this lateral asymmetry interacts with the right hemisphere dominance for visual location processing and left hemisphere dominance for visual shape processing. Volunteers were tested in a location discrimination task and a shape discrimination task. The target stimuli (S2) could occur in the left or right hemifield. They were preceded by an ipsilateral, contralateral or bilateral prime stimulus (S1). The attentional effect produced by the right S1 was larger than that produced by the left S1. This lateral asymmetry was similar between the two tasks suggesting that the hemispheric asymmetries of visual mechanisms do not contribute to it. The finding that it was basically due to a longer reaction time to the left S2 than to the right S2 for the contralateral S1 condition suggests that the inhibitory component of attention is laterally asymmetric.