基于图割与GVF Snake的凹型目标快速提取算法

Fast concave object extraction algorithm based on graph cuts and GVF Snake

将图割理论与GVF Snake模型有机结合,提出了一种凹型目标的快速提取算法。首先用图割算法对初始轮廓线迭代变形,使其在快速提取非凹型段目标边界的同时将轮廓线有效地置于梯度矢量流力场的"有效逼近域"内,然后用GVF Snake算法继续对轮廓线迭代变形,提取凹型段目标边界。实验表明,该算法能快速、准确提取凹型目标。

Combining Gradient Vector Flow （GVF） Snake model with graph cuts theory, a fast concave object extraction algorithm was proposed. First the initial contour was deformed with the graph cuts algorithm, which could put the initial contour in the valid approach region of the GVF field to extract non-concave part of the object boundary. Then the active contour was input to Snake model and began its evolvement to concave part of the object boundary. Experimental results show that the algorithm can converge to concave object boundary fast and accurately.