Premium
Automated object tracking, event detection, and recognition for high‐speed video of drop formation phenomena
Author(s) -
Radcliffe Andrew J.,
Reklaitis Gintaras V.
Publication year - 2021
Publication title -
aiche journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.958
H-Index - 167
eISSN - 1547-5905
pISSN - 0001-1541
DOI - 10.1002/aic.17245
Subject(s) - computer science , video tracking , artificial intelligence , computer vision , event (particle physics) , software , automation , real time computing , process (computing) , data mining , object (grammar) , engineering , programming language , operating system , physics , quantum mechanics , mechanical engineering
Optical imaging technologies have the potential to provide detailed information which can inform process design decisions via modeling of critical phenomena or provide innovative process sensors for use in online monitoring and control strategies. In this work, novel algorithms are developed for automated object tracking, event detection, and classification in high‐speed imaging sequences of drop breakup and coalescence. Using generalization of the physical patterns, a combined strategy extracts quantitative information about the spatiotemporal trajectory and events of every fluid‐object, mapping its lineage and relations to other objects without the need for a training dataset. This high‐level reference frame makes the extracted information easily accessible for evaluation of quantities over sets of objects or events of interest. Included results demonstrate the capability of the software for handling complex multievent scenarios at acceptable execution times, processing 2.7 GB image data (22,000 images/30 videos) in 110–115 s in MATLAB on a standard desktop computer.