+ All Categories

Anomaly

Date post: 08-Nov-2014
Category:
Upload: indu-st
View: 21 times
Download: 0 times
Share this document with a friend
Description:
generalized background subtraction with outlier detection.
Popular Tags:
22
Anomaly Detection Based On Object Behavior By S.Indumathy II M.E CSE Guided by Mr.M.Varghese M.E (Ph.D) Infant Jesus College of Engg & Tech
Transcript
Page 1: Anomaly

Anomaly Detection Based On Object Behavior

By

S.Indumathy II M.E CSE Guided by

Mr.M.Varghese M.E (Ph.D) Infant Jesus College of Engg & Tech

Page 2: Anomaly

04/08/2023 Event based background subtraction 2

Background Subtraction

• Identify the foreground objects in an image or video frame

• Reduces the amount of data to be processed• Widely used in traffic monitoring, human

action recognition, etc.,

Page 3: Anomaly

04/08/2023 Event based background subtraction 3

Scope of the Project

Behavior subtraction – Generalized Background Subtraction, discover changes in scene dynamics

Temporal anomaly detection Track the path implicitly

Page 4: Anomaly

04/08/2023 Event based background subtraction 4

Objectives

To find Background behavior image. Detecting the anomalous behavior. To perform Vector Behavior Subtraction.

Page 5: Anomaly

04/08/2023 Event based background subtraction 5

Literature Survey

ViBe: A universal background subtraction algorithm for video sequences.

Olivier Barnich and Marc Van Droogenbroeck,June 2011.• Background Subtraction-Classification problem.• Classify a new pixel with respect to its immediate

neighborhood in the chosen color space • A new pixel is included in the background only if it has been

classified as a background sample.• Background model initialization and updating.

Page 6: Anomaly

04/08/2023 Event based background subtraction 6

Behavior Subtraction.

Pierre-Marc Jodoin, Venkatesh Saligrama, Janusz Konrad. Jan 2008.

• Anomaly detection.• Train the video sequence with the number of frames.• Two phase- Training phase and detection phase.

Page 7: Anomaly

04/08/2023 Event based background subtraction 7

Existing system

Classical approaches to BS are based on photometric scene properties.

Eg. GMM,KDE Temporal anomaly detection done by

constructing a dictionary of anomalies. Tracking algorithm applied to track the motion

path.

Page 8: Anomaly

04/08/2023 Event based background subtraction 8

Proposed system

• Behavior space Simple background subtraction followed

by background update.• Introduced the concept of event.• Applying event hypothesis test to detect the

anomaly.• Addressing deficiencies of motion path based

algorithm.

Page 9: Anomaly

04/08/2023 Event based background subtraction 9

Modules

The modules of the proposed system are• Background subtraction• Event Modeling• Behavior Image• Outlier detection• Background Subtraction with vector objects.

Page 10: Anomaly

04/08/2023 Event based background subtraction 10

Background Subtraction• Detects the presence of motion by simple

frame differencing and thresholding.

Each motion label captures the amount of activity occurring at a spatial location.

Page 11: Anomaly

04/08/2023 Event based background subtraction 11

Event Modeling Models the dynamic activity for each pixel at

the specific spatial location by using the motion labels and the object descriptor.

Page 12: Anomaly

04/08/2023 Event based background subtraction 12

Behavior Image• Identifies the background activity.• Nominal model can be obtained.• Implicitly tracks the path followed by moving

object.• Suppresses the event noise by averaging filter.

Page 13: Anomaly

04/08/2023 Event based background subtraction 13

Outlier detection• Hypothesis testing.

• accumulation of motion labels, object feature (size) and state transition

Page 14: Anomaly

04/08/2023 Event based background subtraction 14

Vector Behavior Subtraction• Vector object descriptor includes size and

direction in five components.• Components of vector descriptor are summed

up during busy period.• Enclose the nominal data for each pixel within

a cuboid.

Page 15: Anomaly

04/08/2023 Event based background subtraction 15

Screen shot-1

Page 16: Anomaly

04/08/2023 Event based background subtraction 16

contd…….

Behavior Image

Page 17: Anomaly

04/08/2023 Event based background subtraction 17

Screen shot - 2

Page 18: Anomaly

04/08/2023 Event based background subtraction 18

Screen shot-3

Page 19: Anomaly

04/08/2023 Event based background subtraction 19

Screen shot-4

Page 20: Anomaly

04/08/2023 Event based background subtraction 20

Conclusion

• Easy to implement ,uses little memory.• Content blind

Page 21: Anomaly

04/08/2023 Event based background subtraction 21

Reference• P.-M. Jodoin, V. Saligrama, and J. Konrad, “Behavior subtraction,”

Proc.SPIE Visual Commun. Image Process., vol. 6822, pp. 10.1–10.12, Jan.2008.

• A. Elgammal, R. Duraiswami, D. Harwood, and L. Davis, “Background and foreground modeling using nonparametric kernel density for visual surveillance,” Proc. IEEE, vol. 90, no. 7, pp. 1151–1163, Jul. 2002.

• O. Barnich and M. V. Droogenbroeck, “ViBe: A universal background subtraction algorithm for video sequences,” IEEE Trans. Image Process.,vol. 20, no. 6, pp. 1709–1724, Jun. 2011.

• P.-M. Jodoin, J. Konrad, V. Saligrama, and V. Veilleux-Gaboury, “Motion

detection with an unstable camera,” in Proc. 15th IEEE Int. Conf. Image

Process., Oct. 2008, pp. 229–232.

Page 22: Anomaly

04/08/2023 Event based background subtraction 22

THANK U


Recommended