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

Chakraborty, Debarati B.; Pal, Sankar K.

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