209,348 research outputs found
Lightweight Agent Framework for Camera Array Applications
No full textThis paper describes a lightweight middleware agent framework (LAF) for coordinating a large array of computers with attached cameras to construct high resolution video-rate image sequences. Compared to existing camera middleware, LAF provides more than a remote sensor access API. The use of an agent framework allows reconfigurable and transparent access to cameras, as well as software agents capable of intelligent processing. It also eases maintenance by encouraging code reuse. Other features include an automatic discovery mechanism at startup, and multiple language bindings. Performance tests showed the lightweight nature of the framework while validating its correctness and scalability. Two different camera agents were implemented to provide access to a large array of distributed cameras. Correct operation of these camera agents was confirmed via several image processing agents
Optimizing Compton camera performance
No full textAmore realistic simulation approach is used to study the behavior of the Compton camera in
this thesis than previous studies to date. The Compton camera differs from gamma cameras
in that the collimator is replaced by a detector known as the ‘scatterer’. Gamma rays may be
Compton scattered in the scatterer and subsequently detected by an ‘absorber’ which is the
equivalent of the detector in a gamma camera. By measuring the energies and the positions
of the points on the scatterer and the absorber where the incident and scattered gamma rays
interacted with the detectors, an image of the source can be reconstructed. Because there is
no collimator present, the potential sensitivity of the Compton camera is much higher than
the gamma camera, resulting in reduced acquisition times.
Most of the work described in this thesis was done with the GEANT4 Monte Carlo
simulation software. GEANT4 has been proven to be very robust and efficient in modelling
physics problems of radiation transport and interactions with matter in complex geometries.
Four major studies are carried out to estimate and optimize the performance of this
novel equipment. The first study takes a look at the scatterer’s imaging parameters with
the aim of prescribing an optimal scatterer material and geometry. In the second study,
the contribution of the absorber to the overall Compton camera performance is evaluated,
considering detector material, interaction type and geometry. The third study explores the
limitations imposed by the detector energy threshold and dead time on the Compton camera
performance, using a simplified model of the general electronic architecture. An evaluation
of Compton camera for scintimammography was performed in the fourth study. For
this study, three dual-head Compton camera models (Si/CZT, Si/LaBr₃:Ce and Si/NaI(Tl)
Compton cameras) were simulated, and the effect of scintillation photons’ interactions with
the photomultipliers was implemented.
The results show that silicon of about 1 cm thickness would be adequate as the Compton
camera scatterer. Analyses suggest however, that the choice of silicon is not completely
flawless. Doppler broadening for this detector material contributes as much as 7.3 mm
and 2.4 mm to full-width-at-half-maximum (FWHM) image resolution at 140.5 keV and
511 keV respectively. On the other hand, detector spatial resolution which accounts for
the least image degradation at 140.5 keV is found to be the dominant degrading factor
at 511 keV, suggesting that the absorber parameters play major roles in image resolution
at higher diagnostic energies. Findings further suggest that cadmium zinc telluride (CZT)
would be themost suitable detector as the absorber since thematerial demonstrated the highest efficiency and least positioning error due to multiple interactions as well as good spatial
resolution. The inclusion of the energy threshold and detector dead time at 140.5 keV, reduced
the Compton camera detection efficiency by 48% and 17% respectively, but improved
the image resolution from 10.7 mm to 9.5 mm at the source-to-scatterer distance of 5 cm.
At 511 keV, the inclusion of these parameters reduced the efficiency by 6% and 13% respectively,
but made no significant difference on the camera resolution. For a challenging
detection case in scintimammography, 5 mm breast tumours of tumour/background uptakes
of 10:1 and 6:1 at 511 keV were used. The best signal-to-noise ratio (SNR) was attained
for the Si/CZT Compton camera model, with the SNR values of 12.2 and 5.3.
It is therefore envisioned that with an optimal camera geometry, improved reconstruction
technique and adequate filter algorithm, the combination of Si and CZT as the scatterer
and the absorber of the Compton camera would make a very promising imaging system for
nuclear medicine studies at higher gamma ray energies where the collimated SPECT systems
perform very poorly due to increased septal penetration. It is equally evident from the
studies that with improved technology, new detectors such as LaBr₃:Ce could replace the
traditional NaI(Tl) detector as imaging detectors
Automatic camera selection for activity monitoring in a multi-camera system for tennis
No full textIn professional tennis training matches, the coach needs to be able to view play from the most appropriate angle in order to monitor players' activities. In this paper, we describe and evaluate a system for automatic camera selection from a network of synchronised cameras within a tennis sporting arena. This work combines synchronised video streams from multiple cameras into a single summary video suitable for critical review by both tennis players and coaches. Using an overhead camera view, our system automatically determines the 2D tennis-court calibration resulting in a mapping that relates a player's position in the overhead camera to their position and size in another camera view in the network. This allows the system to determine the appearance of a player in each of the other cameras and thereby choose the best view for each player via a novel technique. The video summaries are evaluated in end-user studies and shown to provide an efficient means of multi-stream visualisation for tennis player activity monitoring
Trajectory based video analysis in multi-camera setups
No full textPhDThis thesis presents an automated framework for activity analysis in multi-camera
setups. We start with the calibration of cameras particularly without overlapping
views. An algorithm is presented that exploits trajectory observations in each view
and works iteratively on camera pairs. First outliers are identified and removed
from observations of each camera. Next, spatio-temporal information derived from
the available trajectory is used to estimate unobserved trajectory segments in areas
uncovered by the cameras. The unobserved trajectory estimates are used to estimate
the relative position of each camera pair, whereas the exit-entrance direction of
each object is used to estimate their relative orientation. The process continues and
iteratively approximates the configuration of all cameras with respect to each other.
Finally, we refi ne the initial configuration estimates with bundle adjustment, based
on the observed and estimated trajectory segments. For cameras with overlapping
views, state-of-the-art homography based approaches are used for calibration.
Next we establish object correspondence across multiple views. Our algorithm
consists of three steps, namely association, fusion and linkage. For association,
local trajectory pairs corresponding to the same physical object are estimated using
multiple spatio-temporal features on a common ground plane. To disambiguate
spurious associations, we employ a hybrid approach that utilises the matching results
on the image plane and ground plane. The trajectory segments after association
are fused by adaptive averaging. Trajectory linkage then integrates segments and generates a single trajectory of an object across the entire observed area.
Finally, for activities analysis clustering is applied on complete trajectories. Our
clustering algorithm is based on four main steps, namely the extraction of a set of
representative trajectory features, non-parametric clustering, cluster merging and
information fusion for the identification of normal and rare object motion patterns.
First we transform the trajectories into a set of feature spaces on which Meanshift
identi es the modes and the corresponding clusters. Furthermore, a merging
procedure is devised to re fine these results by combining similar adjacent clusters.
The fi nal common patterns are estimated by fusing the clustering results across all
feature spaces. Clusters corresponding to reoccurring trajectories are considered as
normal, whereas sparse trajectories are associated to abnormal and rare events.
The performance of the proposed framework is evaluated on standard data-sets
and compared with state-of-the-art techniques. Experimental results show that
the proposed framework outperforms state-of-the-art algorithms both in terms of
accuracy and robustness
Performance Analysis for Gait in Camera Networks
No full textThis paper deploys gait analysis for subject identification in multi-camera surveillance scenarios. We present a new method for viewpoint independent markerless gait analysis that does not require camera calibration and works with a wide range of directions of walking. These properties make the proposed method particularly suitable for gait identification in real surveillance scenarios where people and their behaviour need to be tracked across a set of cameras. Tests on 300 synthetic and real video sequences, with subjects walking freely along different walking directions, have been performed. Since the choice of the cameras' characteristics is a key-point for the development of a smart surveillance system, the performance of the proposed approach is measured with respect to different video properties: spatial resolution, frame-rate, data compression and image quality. The obtained results show that markerless gait analysis can be achieved without any knowledge of camera's position and subject's pose. The extracted gait parameters allow recognition of people walking from different views with a mean recognition rate of 92.2% and confirm that gait can be effectively used for subjects' identification in a multi-camera surveillance scenario
A new camera for high-resolution infrared imaging of works of art
No full textA new camera – SIRIS (scanning infrared imaging system) – developed at the National Gallery in London allows high-resolution images to be made in the near infrared region (900–1700 nm). The camera is based on a commercially available 320 × 256 pixel indium gallium arsenide area array sensor. This relatively small sensor is moved across the focal plane of the camera using two orthogonal translation stages to give images of c. 5000 × 5000 pixels. The main advantages of the SIRIS camera over scanning infrared devices or sequential image capture and mosaic assembly are its comparative portability and rapid image acquisition – making a 5000 × 5000 pixel image takes less than 20 minutes. The SIRIS camera can operate at a range of resolutions; from around 2.5 pixels per millimetre over an area of up to 2 × 2 m to 10 pixels per millimetre when examining an area measuring 0.5 × 0.5 m. The development of the mechanical, optical and electronic components of the camera, including the design of a new lens, is described. The software used to control image capture and to assemble the individual frames into a seamless mosaic image is mentioned. The camera was designed primarily to examine underdrawings in paintings; preliminary results from test targets and paintings imaged in situ are presented and the quality of the images compared with those from other cameras currently used for this application
Automated camera ranking and selection using video content and scene context
No full textPhDWhen observing a scene with multiple cameras, an important problem to solve is to automatically
identify “what camera feed should be shown and when?” The answer to this question is of interest
for a number of applications and scenarios ranging from sports to surveillance. In this thesis we
present a framework for the ranking of each video frame and camera across time and the camera
network, respectively. This ranking is then used for automated video production. In the first stage
information from each camera view and from the objects in it is extracted and represented in a way
that allows for object- and frame-ranking. First objects are detected and ranked within and across
camera views. This ranking takes into account both visible and contextual information related to
the object. Then content ranking is performed based on the objects in the view and camera-network
level information. We propose two novel techniques for content ranking namely: Routing Based
Ranking (RBR) and Multivariate Gaussian based Ranking (MVG). In RBR we use a rule based
framework where weighted fusion of object and frame level information takes place while in MVG
the rank is estimated as a multivariate Gaussian distribution. Through experimental and subjective
validation we demonstrate that the proposed content ranking strategies allows the identification of
the best-camera at each time.
The second part of the thesis focuses on the automatic generation of N-to-1 videos based on the
ranked content. We demonstrate that in such production settings it is undesirable to have frequent
inter-camera switching. Thus motivating the need for a compromise, between selecting the best
camera most of the time and minimising the frequent inter-camera switching, we demonstrate that
state-of-the-art techniques for this task are inadequate and fail in dynamic scenes. We propose three
novel methods for automated camera selection. The first method (¡go f ) performs a joint optimization
of a cost function that depends on both the view quality and inter-camera switching so that a
i
Abstract ii
pleasing best-view video sequence can be composed. The other two methods (¡dbn and ¡util) include
the selection decision into the ranking-strategy. In ¡dbn we model the best-camera selection
as a state sequence via Directed Acyclic Graphs (DAG) designed as a Dynamic Bayesian Network
(DBN), which encodes the contextual knowledge about the camera network and employs the past
information to minimize the inter camera switches. In comparison ¡util utilizes the past as well
as the future information in a Partially Observable Markov Decision Process (POMDP) where the
camera-selection at a certain time is influenced by the past information and its repercussions in
the future. The performance of the proposed approach is demonstrated on multiple real and synthetic
multi-camera setups. We compare the proposed architectures with various baseline methods
with encouraging results. The performance of the proposed approaches is also validated through
extensive subjective testing
Camera-trap station setup.
No full textThe a) plan view and b) profile view of the camera trap station setup. Components are: A, drift fence (~30 x ~300 cm); B, cork floor tile (30 x 30 x 0.6 cm); C, passive infrared (PIR) detection bands to demonstrate orientation; D, Reconyx HC600 camera trap; E, PVC bait-holder; and F, camera mounting pole. The camera trap is mounted ~70 cm above the tile with the PIR sensor oriented furthest from the mounting pole.</p
Reliable camera motion estimation from compressed MPEG videos using machine learning approach
No full textAs an important feature in characterizing video content, camera motion has been widely applied in various multimedia and computer vision applications. A novel method for fast and reliable estimation of camera motion from MPEG videos is proposed, using support vector machine for estimation in a regression model trained on a synthesized sequence. Experiments conducted on real sequences show that the proposed method yields much improved results in estimating camera motions while the difficulty in selecting valid macroblocks and motion vectors is skipped
Sensor node localisation using a stereo camera rig
No full textIn this paper, we use stereo vision processing techniques to
detect and localise sensors used for monitoring simulated
environmental events within an experimental sensor network testbed. Our sensor nodes communicate to the camera through patterns emitted by light emitting diodes (LEDs). Ultimately, we envisage the use of very low-cost, low-power,
compact microcontroller-based sensing nodes that employ
LED communication rather than power hungry RF to transmit data that is gathered via existing CCTV infrastructure.
To facilitate our research, we have constructed a controlled
environment where nodes and cameras can be deployed and
potentially hazardous chemical or physical plumes can be
introduced to simulate environmental pollution events in a
controlled manner. In this paper we show how 3D spatial
localisation of sensors becomes a straightforward task when
a stereo camera rig is used rather than a more usual 2D
CCTV camera
- …
