Multi-people posture recognition method based on optical flow positioning and sliding window detection

The invention discloses a multi-people posture recognition method based on optical flow positioning and sliding window detection. The method comprises the following steps of: obtaining an image optical flow vector in a video, obtaining the area of a moving object in an image, and utilizing a method...

Full description

Saved in:
Bibliographic Details
Main Authors ZHU PENGHAI, GONG FAMING, GONG WENJUAN, LI XIAORAN
Format Patent
LanguageChinese
English
Published 03.05.2017
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 multi-people posture recognition method based on optical flow positioning and sliding window detection. The method comprises the following steps of: obtaining an image optical flow vector in a video, obtaining the area of a moving object in an image, and utilizing a method based on body parts to carry out human body posture recognition on the area of the moving object; carrying out the sliding window detection on an extracted movement area, and comparing similarity between the image and the contour template of the body part; utilizing a tree structure to carry out modeling on a human body; using a histogram intersection operation to finish characteristic matching recognition work; and calculating the corresponding relationship of the movement area, and carrying out multi-people posture recognition, people counting and people tracking. By use of the method, the governance system of big data is utilized, and video data is utilized to make a contribution to aspects including problem fin
Bibliography:Application Number: CN201611013187