Pedestrian re-recognition method based on multi-feature fusion

The invention discloses a pedestrian re-recognition method based on multi-feature fusion. Pedestrian extraction is conducted firstly, specifically, a pedestrian is extracted from the background; bodypart recognition is conducted, specifically, three prominent parts of the body are identified; multip...

Full description

Saved in:
Bibliographic Details
Main Authors LIAO KAIYANG, WU JIMEI, LI CONG, ZHAO FAN
Format Patent
LanguageChinese
English
Published 30.11.2018
Subjects
CALCULATING
COMPUTING
COUNTING
HANDLING RECORD CARRIERS
PHYSICS
PRESENTATION OF DATA
RECOGNITION OF DATA
RECORD CARRIERS
Online AccessGet full text

Cover

More Information
Summary:The invention discloses a pedestrian re-recognition method based on multi-feature fusion. Pedestrian extraction is conducted firstly, specifically, a pedestrian is extracted from the background; bodypart recognition is conducted, specifically, three prominent parts of the body are identified; multiple feature extraction is conducted to establish a pedestrian feature representation model; the pedestrian feature representation model is subjected to feature variation to eliminate the influence of feature spatial distribution inconsistency of different cameras; a distance measuring method suitable for pedestrian re-recognition based on the varied multiple features is designed; and the proposed distance measure method is conducted to carry out distance measurement to the pedestrian feature representation model of a training set and a test set, and matching results are obtained to achieve pedestrian re-recognition in a wide area camera network. A robust feature representation model and a distance measure method sui
Bibliography:Application Number: CN201810642499