在這畢業論文中,是以 Active Contour 為基礎,希望減少傳統的 Snake 在輪廓收尋判斷上的錯誤。特別是在凹型的輪廓上,傳統的 Snake 所運用的能量最小化的概念,或是近幾年所提出的 GVF,用散佈方向向量的概念等。似乎在對於凹陷型的輪廓上,都無法達到太好的效果。
本篇論文提出的方法,主要是希望能讓 Snake 順著邊界去做移動,並且在移動的過程中,不停的去做判斷,而不是在一開始就在整張圖像上佈滿方向。要讓 Snake 能依照在不同的情況,能有不同的移動方向,藉此來因應各種不同的圖形與情況。 In this paper, we propose a model for active contours to detect object in a given image. This model based on Active Contour technique, and reduce traditional snake can''t fit the concave contour. In recent years, many peple presnet the other method to improve the traditional snake such as GVF and GGVF. Their method usually use the direction distrubute, but the method can''t fit with special concave contours. Our approach can let snake move along the boundaries and on-the-fly check where has the edge point that is not be snake point cover yet. Our method can putty fit the special concave contours.