中文摘要：

在无线传感器网络中，利用骨架信息可以提高路由性能，也可以用于网络定位、导航以及分割等。以往的骨架算法往往假定边界节点被全部识别，但在绝大多数网络中，尤其是低密度网络，边界节点往往很难被全部正确识别。提出了一种基于距离变换的低复杂度、分布式骨架提取算法，该算法不要求所有边界节点被识别。实验结果表明，该算法对于边界不准确时能够得到较好骨架，同时对于边界点缺失具有鲁棒性。

英文摘要：

In wireless sensor networks, skeleton information has been successfully used to improve routing performance; and also can be used in localization, navigation and segmentation, etc. Existing solutions often assume that all boundary nodes have been correctly recognized, in most cases, especially for networks with low node density, the boundary nodes are hardly fully identified. This paper proposed a distributed skeleton extraction algorithm of low complexity based on distance transform. The proposed algorithm did not require that all boundary nodes were correctly identified. Experiment results show that the proposed algorithm can achieve a good approximation of skeleton even under incomplete or inaccurate boundaries, and also is robust to the boundary incompleteness.