Skip to main content

Face Tracking

Face Tracking

Face Tracking Using the Kalman Filter

Face tracking is a very complex problem in computer vision. Attempts have been made to create a perfect object tracker. Many of the existing algorithms work under specific conditions, but all algorithms fail when it comes to wide-area surveillance videos where the area to be captured by video extends to square miles. The objects in these videos appear blurred and feature extraction becomes extremely difficult. The problem becomes more complex when the effects of external environmental factors like smog, fog, haze, etc. are also considered. The presence of shadows and occlusions make tracking the object even more difficult.

The main aim of this research will be to develop fast algorithms that are able to address all the aforementioned challenges and also to achieve real-time performance. The following are the objectives of this research:

  • To design an algorithm that can describe the object that has been selected by the user. This has to be implemented by taking into consideration the real-time constraints of the application as well as the accuracy of the tracking mechanism.
  • To implement a tracking mechanism that can handle occlusions.
  • To build an adaptive learning mechanism into the system, so the system learns about the object (its features as well as motion characteristics) in order to perfect the tracking mechanism.

Tracking Flow Chart

Video Demonstration

Face Recognition with Kalman Tracker

<< Back


Vision Lab, Dr. Vijayan Asari, Director

Kettering Laboratories
300 College Park
Dayton, Ohio 45469 - 0232