Skip navigation

Detecting Change using HTM for a Security Surveillance Application

Detecting Change using HTM for a Security Surveillance Application

Elsayed, Mennatallah and Melis, Wim J.C. ORCID: 0000-0003-3779-8629 (2019) Detecting Change using HTM for a Security Surveillance Application. In: Robotics and Artificial Intelligence Summit 2019, 10-12/06/2019, Osaka, Japan. (Submitted)

[img] PDF (Author Accepted Abstract)
23944 MELIS_Detecting_Change_using_HTM_for_a_Security_Surveillance_Application_2019.pdf - Accepted Version
Restricted to Registered users only

Download (110kB) | Request a copy
[img] PDF (Acceptance Email)
23944 MELIS_Acceptance_Email_2019.pdf - Additional Metadata
Restricted to Repository staff only

Download (114kB) | Request a copy

Abstract

Detecting change is part of human nature. Humans build up virtual images of their surroundings over time and notice any change instantly which all happens thanks to their brain. While all of this is functions very well in the human brain, there are substantial benefits to an efficient change detection mechanism, but then created virtually. To replicate the structural and algorithmic properties of the human brain, this paper presents a system based on Hierarchal Temporal Memory (HTM) to detect change in a series of images. Moreover, it integrates the change-detecting HTM machine into a robotic security guard, responsible for monitoring and guarding valuable items. The system was designed to serve a security surveillance application to process and analyse several image inputs during non-working hours and identifying changes e.g. related to unauthorised access. The input to the system is a series of multiple images that were initially processed with MATLAB in order to resizing them from a 256*256 matrix to a smaller size, to then convert them into a single line vector in CSV format to suit HTM’s input requirements. In the various tests being performed, it is shown that HTM detects spatial as well as temporal anomalies, when they occur once or twice, however, if the pattern occurs more regularly, then it gets learned and becomes a familiar pattern, meaning it is no longer detected as an anomaly.

Item Type: Conference or Conference Paper (Paper)
Uncontrolled Keywords: Hierarchical Temporal Memory, Detecting Change,
Subjects: T Technology > TA Engineering (General). Civil engineering (General)
Faculty / Department / Research Group: Faculty of Engineering & Science
Faculty of Engineering & Science > Department of Engineering Science
Faculty of Engineering & Science > Future Technology and the Internet of Things
Last Modified: 05 Jul 2019 23:35
Selected for GREAT 2016: None
Selected for GREAT 2017: None
Selected for GREAT 2018: None
Selected for GREAT 2019: None
URI: http://gala.gre.ac.uk/id/eprint/23944

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year

View more statistics