|M.Sc Student||Arie Shraiber|
|Subject||Parametric Active Contours With Their Applications For|
Object Boundary Detection And Tracking
|Department||Department of Electrical Engineering||Supervisor||Mr. Tannenbaum Allen Robert|
|Full Thesis text|
Snakes, or active contour models, are used extensively in computer vision and image processing applications, particularly for locating object boundaries. However, problems associated with initialization and poor convergence to boundary concavities, have limited their use. In this work, Gradient Vector Flow (GVF algorithm) was implemented; it replaces the traditional external force for active contours, largely solving both of the above problems. This external force is computed as a diffusion of the gradient vectors of a gray-level or binary edge map derived from the image. It differs fundamentally from traditional snake external forces, in that it cannot be written as the negative gradient of a potential function, and the corresponding snake is formulated directly from a force balance condition rather than a variational formulation. In addition, the numerical problem caused by the discretization process is solved by a new resampling sub-algorithm.
The KGVF algorithm, which uses the Gradient Vector Flow (GVF) model with the Kalman filter for tracking moving objects, is also implemented. As stated, the GVF algorithm ensures a larger capture range and a greater ability to contract the object boundaries. The Kalman filter is estimation algorithms that can predict the next moving object position and velocity.
In a simulation section, traditional snake algorithm performances were compared with GVF performances, with and without a resampling mechanism. The last two examples demonstrate the capability of the KGVF to track global object motion and local deformations.