Object Tracking based on Active Contour Modeling

摘要

Object Tracking based on Active Contour Modeling is an image processing based technology that uses snapshots of the object under consideration to track it via robot in the real world. The objective has been to implement a unique methodology that employs the pursuing and adapting of contour to the current state of image, and hence track the object. The system can be implemented in drone planes wherein this algorithm can be used to guide the movement of the gun based on the movements of the object, or, in robot games with a slightly more advanced robot. Initially Image Processing is performed to reduce operation complexity and achieve swift real-time performance. A set of contour-based modeling algorithms is then implemented to `actively' track the subject. Also, relative transformation calculations are made to lock the target via robot, continuously. MATLAB is used to simulate and implement the system and it is tested on field with a ball placed on it and a robot tracking the ball. The experiments prove that the system successfully detects and tracks the object efficiently in the real world for all horizontal and vertical transitions.