Active contours or Snakes are extensively used in computer vision and image processing applications, to locate the object boundaries. Low convergence speed and high complexity in computing have significantly limited their utilities. By taking these problems into consideration, the present research focuses on a novel way in rapid image segmentation methodology. This method utilizes subdivision curves in combination with the Gradient Vector Flow (GVF) snakes to overcome these problems. GVF snakes use region energy minimization which is superior to the mass-spring model of the Tamed snake in whole contour. Furthermore, subdivision curves provide a hierarchical and smooth representation of a shape which is significantly in fine scales. After every step of subdivision, reversely compute the region energy of the subdivision polygon and the local adaptive compensation is carried out. A discrete curvature estimator is used to avoid additional computing in the flat regions of a curve. Therefore, only the segments with high curvature or with fine details require more reverse subdivision computing. Reverse subdivision scheme gives the required flexibility while dealing with a local adaptive compensation. The above-mentioned scheme is similar to dynamic programming. This leads the convergence computing to the appropriate subdivision direction. Rapid reverse computing and absolute reversible and lossless are significant advantages of this scheme. It determines the speed and briefness of the Subdivision Snakes Model. Active Subdivision Snake Model (ASSM) will be very efficient in to detect objects when they are at motion and image registration.
A Novel Numerical Algorithm to Estimate the Subdivision Depth of Binary Subdivision Schemes