Anomaly Detection for Video Surveillance Applications

Abstract

We

Automated Surveillance in Distributed, Visual Networks: An Empirical Comparison of Recent Algorithms

“…It must be remembered here that the thresholds ν 1 and ν 2 must be manually set in KOAD, while KEAD incorporates autonomous setting for all important algorithm parameters [5]. The strikingly low performance of the benchmark NCD-based algorithm may be attributed to the fact that this algorithm requires a significantly longer training period, and needs to maintain a significantly larger database of images to compare new arrivals against [12].…”

“…A high-resolution visual sensor captures a lot of details which may qualify as noise given our objective of detecting physical intruders. An elegant way of eliminating such noise is by using the Canny edge detector [10] to obtain an edge image where the step edges are enhanced. Figure 2 shows the Canny edge images corresponding to the example raw images from Fig.…”

“…Au et al have presented a celebrated algorithm in [10] where a set of novel images are stored, and arriving images are compared to this set. A scene is considered anomalous when the maximum similarity between the given scene and all previously viewed scenes is below a given threshold.…”

Real-Time Intruder Detection in Surveillance Networks Using Adaptive Kernel Methods

“…They compared their results to those obtained using a range of other machine learning-based algorithms such as Neural Network, Decision Trees, Naïve Bayes and k-Nearest Neighbor methods. Au et al presented an algorithm where novel images are stored, and future images are compared against it using a similarity metric based on mutual information [9]. They achieve sparsity by only storing the novel images.…”

Anomaly detection for internet surveillance

“…This can focus on identifying anomalous sequences with respect to a database of normal sequences, identifying an anomalous subsequence within a long sequence, or identifying a pattern in a sequence whose frequency of occurrence is anomalous [7]. Applications of anomaly detection are widespread and vary from intrusion detection in computer networks [31], credit card fraud detection [30], medical applications such as EEG analysis [29], to forensic applications such as the detection of abnormal behavior in surveillance videos [2]. Major challenges consist of sieving out the usually infrequently occurring anomalies from the total (often massive) data, and handling the problem of adaptation: certain (but not all) events that primarily occur as anomalies become accepted as 'normal' events, and should not be detected over and over again.…”