Object Detection and Tracking
November 21, 2014
In this work, color based algorithm is selected for its simplicity and robustness. An object can be detected using color, shape, feature points or pre-learned templates etc. Feature based or learning algorithms could be tuned to detect an object with good accuracy however tend to have less flexibility, higher computational complexity and difficulty with partial occlusions. Multiple objects or noise of same color and changing lighting conditions pose challenges to this approach while it allows flexibility to select any colored object and is robust to scale variations or partial occlusions.