A Method of Tracking Active Object based on Monogenic Phase
Yang Cong · Computer Knowledge and Technology · 2012
Color,an available visual character,is widely used to track video object.But in the environment of variant lighting,the color is easy to change,and it makes the tracking method based on color not stable.Monogenic phase as a character including important informa tion of an image,has the property of brightness invariant.In order to improve the stability of tracking in the environment of variant light ing,a new method of tracking active video object is proposed in this paper based on monogenic phase.In the proposed method,the Gauss ian model is implied to update the background and then the active objects can be detected.After that,the phase histogram of every active object is calculated to make a comparison with the monogenic phase histogram of the object being tracked and thus the Bhattacharyya dis tance is computed to confirm which object is the best match to the video object being tracked.The experimental results demonstrate that the proposed method can achieve robust tracking performances in the environment of variant lighting.