基于嵌入式的机器人视觉伺服寻迹系统研究-Dynamic Active Contours for Visual Tracking.pdf

562 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 51, NO. 4, APRIL 2006 Dynamic Active Contours for Visual Tracking Marc Niethammer, Allen Tannenbaum, and Sigurd Angenent Abstract—Visual tracking using active contours is usually set in or our environment (e.g., controlling the movement of a plane, a static framework. The active contour tracks the object of interest based on visual input). The latter will most likely encompass the in a given frame of an image sequence. A subsequent prediction first (possibly resulting in nested control loops). Both tasks can step ensures good initial placement for the next frame. This ap- proach is unnatural; the curve evolution gets decoupled from the be accomplished by means of feedback mechanisms. In either actual dynamics of the objects to be tracked. True dynamical ap- case we need a good estimate of object position. Once we have proaches exist, all being marker particle based and thus prone to this estimation, we either have fulfilled our task (e.g., for surveil- the shortcomings of such particle-based implementations. In par- lance applications), or use this information to close a control ticular, topological changes are not handled naturally in this frame- loop. This brings to mind a broad range of applications. Indeed, work. The now classical level set approach is tailored for evolutions of manifolds of codimension one. However, dynamic curve evolu- be it for medical or military use, the need for visual tracking is tion is at least a codimension two problem. We propose an effi- ubiquitous. cient, level set based approach for dynamic curve evolution, which Visual feedback control differs significantly from classical addresses