Skip to main content

Research Repository

Advanced Search

Rough video conceptualization for real-time event precognition with motion entropy

Chakraborty, Debarati B.; Pal, Sankar K.

Authors

Sankar K. Pal



Abstract

This article defines a new methodology for pre-recognition of events with object motion analysis in a video without any prior knowledge. This unsupervised application is named as ‘conceptualization’. This conceptualization technique is also tested with real-time video data in an internet of things (IoT) architecture. The merits of rough sets in the framework of granular computing are explored to execute the task. The proposed method is designed for the video sequences that are acquired by simple static RGB sensors. Here the video sequences are granulated with our newly defined ‘motion granules’ and then those are modeled as rough sets over this granulation for moving object/ background estimation. Video conceptualization is performed afterwards by quantifying the approximation with a new measure, namely, motion entropy. The values obtained by this measure reflect the amount of uncertainty present in the motion of each individual moving object which enables precognition of events. The effectiveness of the proposed method is verified with extensive experiments in identifying the different motion patterns present in a video sequence. The frames with possibilities of events present therein are identified with this analysis. Both offline and real-time sequences are used for this verification. An IoT architecture is formed to test the proposed algorithm with physical devices in identifying the frames containing possible events.

Citation

Chakraborty, D. B., & Pal, S. K. (2021). Rough video conceptualization for real-time event precognition with motion entropy. Information Sciences, 543, 488-503. https://doi.org/10.1016/j.ins.2020.09.021

Journal Article Type Article
Acceptance Date Nov 2, 2020
Online Publication Date Nov 23, 2020
Publication Date Jan 8, 2021
Deposit Date Mar 13, 2024
Publicly Available Date Jul 31, 2025
Journal Information Sciences
Print ISSN 0020-0255
Publisher Elsevier
Peer Reviewed Peer Reviewed
Volume 543
Pages 488-503
DOI https://doi.org/10.1016/j.ins.2020.09.021
Public URL https://hull-repository.worktribe.com/output/4588900

Files





You might also like



Downloadable Citations