A DCT based gradient vector flow snake for object boundary detection

Abstract

The paper proposes a new gradient vector flow (GVF) snake based on the discrete cosine transform (DCT) for object boundary detection. In order to develop the DCT GVF snake, for each pixel in the spatial domain, we compute the two dimensional DCT in its local neighborhood. A contrast measure derived from the local DCT coefficients is used to compute the gradient vector flow field, which is employed as the external force to drive the snake to the object edges. The advantage of the DCT GVF snake lies in its robustness to multiplicative noise. The DCT GVF snake is applied to object boundary detection in images with multiplicative noise. Experiments show the boundary detection result obtained by the DCT GVF snake is more accurate than that obtained by the original GVF snake.