Integrating OpenCV with Tekkotsu

Ilya Matiach

15-494: Cognitive Robotics

• Home
• Project
• Problem
• Porting OpenCV
• Face Detection
• Following Object Motion
• Centering Camera
• Face Recognition Using Sift
• Code
• Videos
• Future Work
• Sources

 

 

 

 

Face Tracking Using OpenCV

Copied from http://www.cognotics.com/opencv/servo_2007_series:

Camshift uses three main steps for tracking a face:

1. It creates a color histogram to represent the face (see image below)
2. It calculates a "face probability" for every pixel in the next frame
3. Based on the probability it shifts the location of the rectangle

Example Color Histogram for a Face:

Please note I did not use a tracker for the eyes.  I just used the function CvHaarDetect, much like for detecting a face.  It worked much faster however since I could already predefine the area where the eyes are expected to be found.