Open AccessParticle Filtering Tracking Study of Automatic Extraction Tracking Range
Author(s) -
Jian Cao,
Da Chun Wu
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1748/3/032050
Subject(s) - particle filter , computer vision , tracking (education) , artificial intelligence , computer science , context (archaeology) , noise (video) , filter (signal processing) , image (mathematics) , psychology , paleontology , pedagogy , biology
What needs to be solved is the problem of automatic tracking of pedestrians in a complex monitoring environment. In the actual monitoring environment, there are usually chaotic scenes, noise, light changes, and constant changes in human motion, in this context, the post-test probability and observation probability are non-Gaussic, nonlinear, so the framework of particle filtering is chosen to solve the pedestrian tracking problem. In target modeling, human motion is non-rigid body deformation, and color features for the target plane rotation, non-rigid deformation, partial masking and other situations are more robust, so in tracking pedestrians to choose color features. This paper proposes an overlay algorithm that automatically selects the maximum attribute area to determine the trace area. Finally, this paper uses color features to realize the automatic tracking of pedestrians under the theoretical framework of particle filtering.