This paper introduces a new external force model for active contours and deformable surfaces, called gradient vector flow (GVF). The GVF field is computed as a diffusion of the gradient vectors of a gray-level or binary edge map derived from the image. Unlike traditional snake external forces, GVF 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. The GVF snake has several advantages over traditional snakes, including insensitivity to initialization and the ability to move into boundary concavities. The GVF snake can be initialized inside, outside, or across the object's boundary, and does not require prior knowledge about whether to shrink or expand toward the boundary. It also has a large capture range, allowing it to be initialized far from the boundary. The GVF snake is able to move into boundary concavities, which is a key advantage over traditional snakes. The GVF field is derived from an energy functional that minimizes the energy of the vector field. The GVF field is computed by solving a pair of decoupled linear partial differential equations that diffuse the gradient vectors of a gray-level or binary edge map. The GVF snake is distinguished from nearly all previous snake formulations in that its external forces cannot be written as the negative gradient of a potential function. The GVF snake is able to move into boundary concavities, which is a key advantage over traditional snakes. The GVF snake is also able to handle noisy images and has a large capture range. The GVF snake is able to be initialized far from the boundary, which is a key advantage over traditional snakes. The GVF snake is able to move into boundary concavities, which is a key advantage over traditional snakes. The GVF snake is able to handle noisy images and has a large capture range. The GVF snake is able to be initialized far from the boundary, which is a key advantage over traditional snakes. The GVF snake is able to move into boundary concavities, which is a key advantage over traditional snakes. The GVF snake is able to handle noisy images and has a large capture range. The GVF snake is able to be initialized far from the boundary, which is a key advantage over traditional snakes. The GVF snake is able to move into boundary concavities, which is a key advantage over traditional snakes.This paper introduces a new external force model for active contours and deformable surfaces, called gradient vector flow (GVF). The GVF field is computed as a diffusion of the gradient vectors of a gray-level or binary edge map derived from the image. Unlike traditional snake external forces, GVF 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. The GVF snake has several advantages over traditional snakes, including insensitivity to initialization and the ability to move into boundary concavities. The GVF snake can be initialized inside, outside, or across the object's boundary, and does not require prior knowledge about whether to shrink or expand toward the boundary. It also has a large capture range, allowing it to be initialized far from the boundary. The GVF snake is able to move into boundary concavities, which is a key advantage over traditional snakes. The GVF field is derived from an energy functional that minimizes the energy of the vector field. The GVF field is computed by solving a pair of decoupled linear partial differential equations that diffuse the gradient vectors of a gray-level or binary edge map. The GVF snake is distinguished from nearly all previous snake formulations in that its external forces cannot be written as the negative gradient of a potential function. The GVF snake is able to move into boundary concavities, which is a key advantage over traditional snakes. The GVF snake is also able to handle noisy images and has a large capture range. The GVF snake is able to be initialized far from the boundary, which is a key advantage over traditional snakes. The GVF snake is able to move into boundary concavities, which is a key advantage over traditional snakes. The GVF snake is able to handle noisy images and has a large capture range. The GVF snake is able to be initialized far from the boundary, which is a key advantage over traditional snakes. The GVF snake is able to move into boundary concavities, which is a key advantage over traditional snakes. The GVF snake is able to handle noisy images and has a large capture range. The GVF snake is able to be initialized far from the boundary, which is a key advantage over traditional snakes. The GVF snake is able to move into boundary concavities, which is a key advantage over traditional snakes.