Presenter(s)

Kevin Krucki, Binu Nair

Download

Download Project (1.7 MB)

Description

This research involves live human re-identification on multi-camera systems. Each frame of multiple cameras needs to be captured and analyzed with image processing methods. First, a histogram of oriented gradients (HOG) is performed to identify a person in each frame. Next, Local Binary Pattern (LBP) descriptors are used on each person to determine certain set features about then. Lastly, a red, green, blue (RGB) color histogram is performed on a specific body mask. Each body is then given a label based on their LBP and color histogram information and that label will be sent to a database. This label should be the same across all the cameras. The process should also happen live. The research will include analysis of the difference between using a static body mask and using pose estimation for a more accurate color histogram. Also, regional descriptors will be used to better describe the human body. Lastly, the difference between YCrCb and RGB color histograms will be shown.

Publication Date

4-9-2014

Project Designation

Graduate Research

Primary Advisor

Vijayan Asari

Primary Advisor's Department

Vision Lab

Keywords

Stander Symposium project

Disciplines

Arts and Humanities | Business | Education | Engineering | Life Sciences | Medicine and Health Sciences | Physical Sciences and Mathematics | Social and Behavioral Sciences

Human Re-Identification in Multi-Camera Systems

Share

COinS