Real-Time Object Detection and Tracking System

Jared Holsopple
John Ruppert
Justin Hnatow

Project Overview

The operational objective of this project is to create a system to detect and track objects in real - time. In order to achieve this objective, a system of two cameras and two PC workstations will be developed to capture and process a visual scene. Each camera will perform a specialized function - one will act as a Scene View Camera (SVC) and one as an Object View Camera (OVC). The SVC will have a relatively wide view and monitor a scene, while the OVC will zoom in to a particular object of interest in the viewing range of the SVC. The object of interest for this project will be a human face. To track a human face, color segmentation, facial detection, and object tracking algorithms will be developed. The color segmentation will be done using a simple look-up table. The face detection algorithm will use a Support Vector Machine. The tracking will be done using the CAMSHIFT algorithm. To coordinate the cameras, an Intel OpenCV implementation of projective geometry will be used.

Made by JJJ February 1, 2005