China
Africa
Explore chapters and articles related to this topic
Global Geoid Determination
Published in Petr Vaníček, Nikolaos T. Christou, GEOID and Its GEOPHYSICAL INTERPRETATIONS, 2020
In this scenario, the global anomaly set is used to imply, through Equation 12 or the improved ellipsoidal surface version, a set of potential coefficients [C¯nms]T that can be compared to the a priori coefficient set that must be independent of the terrestrial anomaly data set. We now write: () F=[C¯nms]s−[C¯nms]T=0
Smart Visual Surveillance Technique for Better Crime Management
Published in Siddhartha Bhattacharyya, Anirban Mukherjee, Indrajit Pan, Paramartha Dutta, Arup Kumar Bhaumik, Hybrid Intelligent Techniques for Pattern Analysis and Understanding, 2017
T. J. Narendra Rao, Jeny Rajan
A method for detecting both local and global anomalies using the hierarchical spatio-temporal interest point (STIP) feature representation and Gaussian process regression was proposed by Cheng et al. [31] in 2015. The authors used two level hierarchical representations for the events and their interactions. Under low-level representation, the STIP features of an event are extracted and are defined using a suitable descriptor. These descriptors of the normal events are quantized into a visual vocabulary using the K-means algorithm based on Euclidean distance. The local anomaly detection is achieved by measuring the k-nearest neighbors (k-NN) distance of a test cuboid against the visual vocabulary. The possible interactions in the video are acquired by extracting the ensembles of the nearby STIP features. In order to find the frequent geometric relations of the nearby STIP features from training videos, the ensembles are clustered using a bottom-up greedy clustering approach, into a high level codebook of interaction templates. Each template in the codebook is formulated into a k-NN regression problem and a model is constructed using Gaussian process regression (GPR) for learning and inferring. The likelihood based on semantic and structural similarities of a test ensemble with respect to the GPR models of normal events is calculated using the global negative log likelihood (GNLL) for global anomaly detection. The GPR method can detect both local and global anomalies through hierarchical event representation. It is adaptive and can learn new interactions while individually locating anomalous events. Also, the method is robust to noise.
Anomaly Testing
Published in Jeffrey P. Simmons, Lawrence F. Drummy, Charles A. Bouman, Marc De Graef, Statistical Methods for Materials Science, 2019
A pixel is anomalous in the context of a background. In global anomaly detection that background is the full image, but in local anomaly detection that background is restricted to the immediate neighborhood, often defined in terms of an annulus that surrounds the pixel. Local anomaly detection is one of the most straightforward (and, in practice, more effective) ways to exploit the spatial structure of imagery.
Tucker visual search-based hybrid tracking model and Fractional Kohonen Self-Organizing Map for anomaly localization and detection in surveillance videos
Published in The Imaging Science Journal, 2018
Avinash Ratre, Vinod Pankajakshan
Hu et al. [7] had constructed a network, Deep Incremental Slow Feature Analysis (D-IncSFA) with the aim of learning data-driven representation for video AD without depending on hand-crafted representation. The D-IncSFA network could perform both feature extraction and AD, completing the task in a single step. This technique could detect global anomaly like crowd panic and various types of video anomalies such that the memory and the computation requirement was less. However, it was computationally complex with limited memory for dynamic scenes.
Hyperspectral anomaly detection: a performance comparison of existing techniques
Published in International Journal of Digital Earth, 2022
Noman Raza Shah, Abdur Rahman M. Maud, Farrukh Aziz Bhatti, Muhammad Khizer Ali, Khurram Khurshid, Moazam Maqsood, Muhammad Amin
Depending on the method used for background estimation, anomaly detection algorithms can be broadly classified as either global or local algorithms. In global anomaly detection algorithms, the background is estimated from the complete hyperspectral data cube available before testing individual pixels. On the contrary, in local algorithms, the background is estimated using neighbors of the pixel under test. As a result, local anomaly detection algorithms are typically more computationally intensive compared to their global counterpart.
Multiple profiles sensor-based monitoring and anomaly detection
Published in Journal of Quality Technology, 2018
Chen Zhang, Hao Yan, Seungho Lee, Jianjun Shi
In general, the choice of R depends on the specific OC scenario of most interest. As mentioned in Mei (2011) and Liu, Mei, and Shi (2015), a larger R leads to a better detection performance for global anomaly patterns that may occur in a lot of sensors but with small magnitudes, while a smaller R results in a better detection performance for extreme anomaly patterns that occur in only few sensors but with large magnitudes.