一种基于改进种子区域生长的农村房屋提取

摘要

利用遥感影像对房屋进行精确提取,可以为农村土地利用规划、居民点用地适宜性评价以及进一步开展居民点整治、快速提供基础地理空间数据等工作提供便利.以无人机影像为实验数据,本文利用一种基于改进的种子区域生长算法对房屋进行提取,该方法解决了传统算法的抗噪能力差以及需要事先确定种子数量和位置等问题.随后,运用阈值处理与数学形态学对分割结果进行进一步处理.实验结果表明,采用本文方法提取农村房屋精度较高.%Using remote sensing imagery to conduct accurate house extraction can quickly provide basic geo-spatial data for rural land use planning,residential land suitability evaluation and further residential renovation.Taking UAV imagery as test data,this paper conducts house extraction based on an improved seeded region growing algorithm,which overcomes the shortcomings of poor anti noise and the need to determine the seed number & location in advance of traditional algorithm.The initial segmentation result is subsequently further processed by the method of thresholding combined with mathematical morphology.The experimental result shows that this method could extract rural house with a relatively high accuracy.