Abstract
Active contours as variational techniques have been widely applied in image processing and computer vision. In this paper, a shape based active contour is proposed for segmenting circular object from the cluttered background. Being different from the general shape based active contours that estimate the shape parameters by gradient descent method, the proposed shape based active contour directly estimates the parameters of the circular shape without any iteration, which overcomes the difficulty in the parameter initialisation. Therefore, the estimation of the circular shape parameter is robust. The experimental results show the good performance of the proposed shape based active contour on circle detection.