A neural system for automated CCTV surveillance

Hunter, Andrew; Owens, Jonathan D.; Carpenter, Mark

doi:10.1049/ic:20030040

Cited by 5 publications

(4 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Neural network-based approaches for learning of typical paths and trajectory modeling are proposed in [11], [12], [13]. Other than the computational complexity and lack of adaptability of neural networks, a major disadvantage of these methods is their inability to handle incomplete and partial trajectories.…”

Section: Related Workmentioning

confidence: 99%

Probabilistic Modeling of Scene Dynamics for Applications in Visual Surveillance

Saleemi

Shafique

Shah

2009

IEEE Trans. Pattern Anal. Mach. Intell.

137

View full text Add to dashboard Cite

Abstract-We propose a novel method to model and learn the scene activity, observed by a static camera. The proposed model is very general and can be applied for solution of a variety of problems. The motion patterns of objects in the scene are modeled in the form of a multivariate nonparametric probability density function of spatiotemporal variables (object locations and transition times between them). Kernel Density Estimation is used to learn this model in a completely unsupervised fashion. Learning is accomplished by observing the trajectories of objects by a static camera over extended periods of time. It encodes the probabilistic nature of the behavior of moving objects in the scene and is useful for activity analysis applications, such as persistent tracking and anomalous motion detection. In addition, the model also captures salient scene features, such as the areas of occlusion and most likely paths. Once the model is learned, we use a unified Markov Chain Monte Carlo (MCMC)-based framework for generating the most likely paths in the scene, improving foreground detection, persistent labeling of objects during tracking, and deciding whether a given trajectory represents an anomaly to the observed motion patterns. Experiments with real-world videos are reported which validate the proposed approach.

show abstract

Section: Related Workmentioning

confidence: 99%

Probabilistic Modeling of Scene Dynamics for Applications in Visual Surveillance

Saleemi

Shafique

Shah

2009

IEEE Trans. Pattern Anal. Mach. Intell.

137

View full text Add to dashboard Cite

show abstract

“…Automated video surveillance systems monitor CCTV systems and detect anomalous or suspicious behavior [6,17,11,10,8], with a view to alerting human operators who can take appropriate actions. The most popular approaches rely on motion detection algorithms to identify objects of interest.…”

Section: Introductionmentioning

confidence: 99%

“…We are concerned with the approach where the algorithm assigns an object identity to newly detected silhouettes, and attempts to maintain this identity through successive frames, deleting objects when they ultimately disappear [1,4,13,6].…”

Section: Introductionmentioning

confidence: 99%

Multiple object tracking using a neural cost function

Humphreys¹,

Hunter

2009

Image and Vision Computing

View full text Add to dashboard Cite

This paper presents a new approach to the tracking of multiple objects in CCTV surveillance using a combination of simple neural cost functions based on Self-Organizing Maps, and a greedy assignment algorithm. Using a reference standard data set and an exhaustive search algorithm for benchmarking, we show that the cost function plays the most significant role in realizing high levels of performance. The neural cost function’s context-sensitive treatment of appearance, change of appearance and trajectory yield better tracking than a simple, explicitly designed cost function. The algorithm matches 98.8% of objects to within 15 pixels

show abstract

“…In recent years, two trends have arisen that motivate the concept of image fusion performed in a distributed manner. Firstly, advancements in wireless sensor network (WSN) technology [37,72,81] have meant that small and low power sensor units equipped with microprocessors, wireless communications systems, and imaging sensors are becoming affordable for distributed surveillance [70,63] or virtual reality mapping of real world environments [51]. These may be used to form adhoc WSNs with random topologies, such as aerial surveillance drone 'flocks' [74] that fuse images of the terrain below their flight paths at each time sample.…”

Section: Application To Distributed Image Fusionmentioning

confidence: 99%

On the Primal-Dual Method of Multipliers and its Applications

O’Connor¹

View full text Add to dashboard Cite

<p>With ever growing sources of digital data and the reductions in cost of small-scale wireless processing nodes, equipped with various sensors, microprocessors, and communication systems, we are seeing an increasing need for efficient distributed processing algorithms and techniques. This thesis focuses on the Primal-Dual Method of Multipliers (PDMM) as it applies to wireless sensor networks, and develops new algorithms based on PDMM more appropriate for the limitations on processing power, battery life, and memory that these devices suffer from. We develop FS-PDMM and QA-PDMM that greatly improve the efficiency of local node computations when dealing with regularized optimization problems and smooth cost function optimization problems, respectively. We combine these approaches to form the FSQA-PDMM algorithm that may be applied to problems with smooth cost functions and non-smooth regularization functions. Additionally, these three methods often eliminate the need for numerical optimization packages, reducing the memory cost on our nodes. We present the FT-PDMM algorithm for finite-time convergence of quadratic consensus problems, reducing the number of in-network iterations required for network convergence. Finally, we present two signal processing applications that benefit from our theoretical work: a distributed sparse near-field acoustic beamformer; and a distributed image fusion algorithm for use in imaging arrays. Simulated experiments confirm the benefit of our approaches, and demonstrate the computational gains to be made by tailoring our techniques towards sensor networks.</p>

show abstract

A neural system for automated CCTV surveillance

Cited by 5 publications

References 12 publications

Probabilistic Modeling of Scene Dynamics for Applications in Visual Surveillance

Probabilistic Modeling of Scene Dynamics for Applications in Visual Surveillance

Multiple object tracking using a neural cost function

On the Primal-Dual Method of Multipliers and its Applications

Contact Info

Product

Resources

About