Abstract: A skeleton extraction method is presented based on external pressures around the border points of a binary image. Firstly, the boundaries of binary images are marked to determine forces on border points. By analysis of three conditions of pressure, that whether the point is the skeleton unit or not is determined. The iteration process is carried out until no new skeleton unit appears, then the preliminary skeleton is achieved. Redundant points are searched and erased based on the shape feature analysis by 3×3 area in the skeleton. Then the ultimate skeleton structure map is obtained. The skeleton structure diagram created by EPM is compared with that by the skeleton extraction function（Bwmorph） in commercial software Matlab 6.5. It shows that skeleton lines by EPM are smoother and there are less tiny branches than those in Matlab 6.5. Finally, the method is applied in the sea surface temperature fronts extraction. The results show the position and structure of extracted fronts is accurate and the EPM is feasible in skeleton extraction.

Key words: External Pressure Method（EPM）, binary image, skeleton extraction

摘要： 提出了基于边界外部压力的二值图像结构化骨架提取方法。先标记出二值图像的边界，判断各个边界点的受力情况，通过三个条件判断各个边界点是否为骨架单元。迭代这一过程，直到没有新的骨架单元产生，即得到初步的骨架结构。通过对骨架中3×3区域内骨架形状分析，对骨架中的冗赘单元进行查找并删除。将该方法得到的骨架结构图与软件Matlab 6.5骨架提取函数（Bwmorph）得到的骨架结构进行了比较，表明用该方法得到的结果比Matlab 6.5中得到的骨架线条更平滑，而且细小分枝更少。将该方法应用到海表温度锋面的提取上，结果显示锋面位置与结构准确，说明利用该方法进行骨架提取是可行的。

关键词: 外部压力法, 二值图像, 骨架提取