改进Balloon Snake算法提取遥感影像水体边界

发布时间：2018-06-20 19:17

本文选题：遥感影像 + 河中岛　；参考：《遥感学报》2017年03期

【摘要】：相比于正交T-Snake算法,传统Balloon Snake算法不受格网约束,能够更加精确提取高分辨率遥感影像上水体,然而尚不能解决河中岛引起的拓扑冲突问题。由此,本文提出了一种改进Balloon Snake算法(T-Balloon Snake),设计了目标内部由于空洞引起的拓扑冲突检测与处理机制。算法首先在目标水体内部手动设置一个初始轮廓,在曲线每次膨胀前对曲线进行自相交检测,若存在拓扑冲突则对曲线进行分裂与合并处理,最终直至迭代停止。实验选取了三类湖泊与河流高分影像,实现了复杂遥感影像水体矢量边界的一次性精确提取。实验结果表明,该方法可直接有效提取水体中深凹以及岛状边界,在提取结果正确性和完整性上都优于传统Balloon Snake算法。
[Abstract]:Compared with orthogonal T-Snake algorithm, the traditional Balloon snake algorithm is not constrained by grid and can extract water body from high resolution remote sensing image more accurately. However, it can not solve the topological conflict caused by the island in river. Therefore, an improved Balloon snake algorithm, T-Balloon Snakeon, is proposed in this paper, and a mechanism for detecting and handling topological conflicts caused by holes in the target is designed. The algorithm first manually sets an initial contour in the target water body, detects the curve self-intersection before each expansion of the curve, splits and merges the curve if there is a topological conflict, and finally stops the iteration. Three kinds of high score images of lakes and rivers are selected in the experiment, and the complex remote sensing images of water body vector boundary are extracted accurately at one time. The experimental results show that this method can directly and effectively extract the deep concave and island boundary in water, and is superior to the traditional Balloon snake algorithm in the correctness and integrity of the extraction results.
【作者单位】：武汉大学测绘遥感信息工程国家重点实验室;武汉大学地球空间信息技术协同创新中心;
【基金】：装备预研教育部联合基金(编号:6141A02022318) 湖北省自然科学基金创新群体(编号:2016CFA003)~~
【分类号】：TP751

【相似文献】