基于高分辨率遥感影像的耕地地块提取方法研究

Research on cultivated land parcel extraction based on high-resolution remote sensing image

针对耕地地块提取问题,提出了一种基于图像分割的耕地地块提取新方法。该方法以高分辨率遥感影像为基础,借助于边缘提取和数学形态学的方法,通过边缘检测、边缘闭合、区域标号和后处理四个步骤,提取耕地地块。该方法在IDL6.3平台下编程实现。将此方法应用于北京地区QuickBird多光谱遥感影像,结果表明此方法有较好的定位精度,又在一定程度上去除了噪声,具有较好的实用性。

Cultivated land parcel extraction is critical for national operational monitoring system of crop planting area on large scale, especially for the national scale like China. According to edge detection and mathematical morphology theories, an effective algorithm has been put forward to extract parcels based on high-resolution remote sensing image. The algorithm consists of edge detection, edge closing, region labeling and hysteresis. Experiment with Quickbird images indicates that this algorithm has better performance on not only location accuracy but also denoise. This algorithm is run on IDL 6. 3 software.