ANOMALY DETECTION OF EVENTS IN CROWDED ENVIRONMENT AND STUDY OF VARIOUS BACKGROUND SUBTRACTION METHODS

2017 ◽  
Vol 5 (1) ◽  
pp. 32
Author(s):  
Meenal Suryakant Vatsaraj ◽  
Rajan Vishnu Parab ◽  
Prof.D.S Bade
Keyword(s):  
Anomaly Detection ◽  
Random Field ◽  
Markov Random Field ◽  
Background Subtraction ◽  
Computational Cost ◽  
Gaussian Mixture ◽  
Histogram Of Oriented Gradients ◽  
Markov Random ◽  
Wide Range ◽  
Crowded Environment

Anomalous behavior detection and localization in videos of the crowded area that is specific from a dominant pattern are obtained. Appearance and motion information are taken into account to robustly identify different kinds of an anomaly considering a wide range of scenes. Our concept based on histogram of oriented gradients and markov random field easily captures varying dynamic of the crowded environment. Histogram of oriented gradients along with well known markov random field will effectively recognize and characterizes each frame of each scene. Anomaly detection using artificial neural network consist both appearance and motion features which extract within spatio temporal domain of moving pixels that ensures robustness to local noise and thus increases accuracy in detection of a local anomaly with low computational cost. To extract a region of interest we have to subtract background. Background subtraction is done by various methods like Weighted moving mean, Gaussian mixture model, Kernel density estimation.

ANOMALY DETECTION OF EVENTS IN CROWDED ENVIRONMENT AND STUDY OF VARIOUS BACKGROUND SUBTRACTION METHODS

2017 ◽  
Vol 5 (1) ◽  
pp. 32
Author(s):  
Meenal Suryakant Vatsaraj ◽  
Rajan Vishnu Parab ◽  
D S Bade
Keyword(s):  
Anomaly Detection ◽  
Random Field ◽  
Markov Random Field ◽  
Background Subtraction ◽  
Region Of Interest ◽  
Gaussian Mixture ◽  
Anomalous Behavior ◽  
Markov Random ◽  
Wide Range ◽  
Spatio Temporal

Anomalous behavior detection and localization in videos of the crowded area that is specific from a dominant pattern are obtained. Appearance and motion information are taken into account to robustly identify different kinds of an anomaly considering a wide range of scenes. Our concept based on a histogram of oriented gradients and Markov random field easily captures varying dynamic of the crowded environment.Histogram of oriented gradients along with well-known Markov random field will effectively recognize and characterizes each frame of each scene. Anomaly detection using artificial neural network consist both appearance and motion features which extract within spatio temporal domain of moving pixels that ensures robustness to local noise and thus increases accuracy in detection of a local anomaly with low computational cost.To extract a region of interest we have to subtract background. Background subtraction is done by various methods like Weighted moving mean, Gaussian mixture model, Kernel density estimation. 

A one-step Bayesian inversion framework for three-dimensional reservoir characterization based on a Gaussian mixture model – A Norwegian Sea demonstration

Geophysics ◽  
2020 ◽  
pp. 1-68
Author(s):  
Torstein Fjeldstad ◽  
Per Avseth ◽  
Henning Omre
Keyword(s):  
Random Field ◽  
Markov Random Field ◽  
Gaussian Mixture ◽  
Three Dimensions ◽  
Monte Carlo Algorithm ◽  
Petrophysical Properties ◽  
Norwegian Sea ◽  
Spatial Coupling ◽  
Markov Random ◽  
One Step

A one-step approach for Bayesian prediction and uncertainty quantification of lithology/fluid classes, petrophysical properties and elastic attributes conditional on prestack 3D seismic amplitude-versus-offset data is presented. A 3D Markov random field prior model is assumed for the lithology/fluid classes to ensure spatially coupled lithology/fluid class predictions in both the lateral and vertical directions. Conditional on the lithology/fluid classes, we consider Gauss-linear petrophysical and rock physics models including depth trends. Then, the marginal prior models for the petrophysical properties and elastic attributes are multivariate Gaussian mixture models. The likelihood model is assumed to be Gauss-linear to allow for analytic computation. A recursive algorithm that translates the Gibbs formulation of the Markov random field into a set of vertical Markov chains is proposed. This algorithm provides a proposal density in a Markov chain Monte Carlo algorithm such that efficient simulation from the posterior model of interest in three dimensions is feasible. The model is demonstrated on real data from a Norwegian Sea gas reservoir. We evaluate the model at the location of a blind well, and we compare results from the proposed model with results from a set of 1D models where each vertical trace is inverted independently. At the blind well location, we obtain at most a 60 % reduction in the root mean square error for the proposed 3D model compared to the model without lateral spatial coupling.

The Research on Fast and Efficient Algorithm in Multi - Cameras-Relay Target Tracking

2012 ◽  
Vol 190-191 ◽  
pp. 1198-1204
Author(s):  
Xiao Ping Li ◽  
Chun Hui Liang ◽  
Si Chen ◽  
Hui Jie Cui ◽  
Jian Qiang Xu ◽  
...  
Keyword(s):  
Random Field ◽  
Target Tracking ◽  
Markov Random Field ◽  
High Efficiency ◽  
Random Field Theory ◽  
Markov Random ◽  
Camshift Algorithm ◽  
Wide Range ◽  
Real Time Tracking ◽  
Video Target Tracking

In computer vision field, video target tracking has important significance of research. This article proposed a fast and efficient multi-cameras-relay target tracking method directed at crucial technology questions in video target tracking. First, it separated the foreground and background of monitoring scene using of Markov random field theory, and then established the Markov random field model of foreground and background, so as to accomplish the moving object recognition; The next, it located the accurate positioning and tracking utilizing the combination of Kalman filtering and improved Camshift algorithm; finally, it made more-cameras-relay target tracking on the basis of coordination and synchronous between cameras, to achieve high efficiency of tracking and strong robustness and monitoring of the wide range of target real-time tracking.

Utilization of Markov Random Field for Large Images: Multiframe Work and Bayesian Approach

2005 ◽  
Vol 277-279 ◽  
pp. 183-188
Author(s):  
Myung Hee Jung ◽  
Eui Jung Yun ◽  
Sy Woo Byun
Keyword(s):  
Random Field ◽  
Markov Random Field ◽  
Bayesian Approach ◽  
Computational Cost ◽  
Remotely Sensed Data ◽  
Global Features ◽  
Suggested Approach ◽  
Texture Modeling ◽  
Markov Random ◽  
Region Labeling

Markov Random Field (MRF) models have been successfully utilized in many digital image processing problems such as texture modeling and region labeling. Although MRF provides a well-defined statistical approach for the analysis of images, one disadvantage is the expensive computational cost for the processing and sampling of large images, since global features are assumed to be specified through local descriptions. In this study, a methodology is explored that reduces the computational burden and increases the speed of image analysis for large images, especially airborne and space-based remotely sensed data. The Bayesian approach is suggested as a reasonable alternative method in parameter estimation of MRF models; the utilization of a multiresolution framework is also investigated, which provides convenient and efficient structures for the transition between local and global features. The suggested approach is applied to the simulation of spatial pattern using MRF.

scholarly journals Mapping Mouse Brain Slice Sequence to a Reference Brain Without 3D Reconstruction

2018 ◽  
Author(s):  
Jing Xiong ◽  
Jing Ren ◽  
Liqun Luo ◽  
Mark Horowitz
Keyword(s):  
Random Field ◽  
Markov Random Field ◽  
Brain Slice ◽  
Neural Circuits ◽  
Brain Slices ◽  
Nonrigid Registration ◽  
Histogram Of Oriented Gradients ◽  
Markov Random ◽  
L2 Norm ◽  
Better Than

AbstractHistological brain slices are widely used in neuroscience to study anatomical organization of neural circuits. Since data from many brains are collected, mapping the slices to a reference atlas is often the first step in interpreting results. Most existing methods rely on an initial reconstruction of the volume before registering it to a reference atlas. Because these slices are prone to distortion during sectioning process and often sectioned with nonstandard angles, reconstruction is challenging and often inaccurate. We propose a framework that maps each slice to its corresponding plane in the atlas to build a plane-wise mapping and then perform 2D nonrigid registration to build pixel-wise mapping. We use the L2 norm of the Histogram of Oriented Gradients (HOG) of two patches as the similarity metric for both steps, and a Markov Random Field formulation that incorporates tissue coherency to compute the nonrigid registration. To fix significantly distorted regions that are misshaped or much smaller than the control grids, we trained a context-aggregation network to segment and warp them to their corresponding regions with thin plate spline. We have shown that our method generates results comparable to an expert neuroscientist and is significantly better than reconstruction-first approaches.

Sign in / Sign up

Export Citation Format

Share Document