基于高分辨率遥感影像的灌排沟渠提取方法

Method of Extracting Irrigation and Drainage Ditches Based on High Remote Sensing Image

科学有效的监管保障了土地整治工程顺利开展。农田水利工程是土地整治项目的重要内容,根据高空间分辨率遥感影像中农田水利灌排沟渠特征,提出一种农田灌排沟渠提取方法。通过对影像中灌排沟渠的光谱值进行统计分析,确定其取值范围,通过二值化和中值滤波去除噪点后得到灌排沟渠的初步结果,最后利用Canny算子对部分有行道树遮挡区域的沟渠进行边缘补充提取,最终得到研究区的灌排沟渠。以内蒙古达拉特旗土地整治某项目区为例,以World View-2为数据源,对提出的方法体系进行了验证,结果表明精度达到91.46%,满足土地整治工程监管的需要。提出的方法很好地解决了微小弯曲和被遮蔽沟渠问题,可为土地整治工程实施监管提供支持。

Farmland water conservancy is an important content of land consolidation projects which scientific and effective supervision can ensure projects carried out smoothly. According to distribution and geometric characteristics of irrigation and drainage ditches in high resolution remote sensing image,an approach to extract irrigation and drainage ditches from high resolution remote sensing image was presented. Spectral range of irrigation,drainage ditches and restrain the noise by binaryzation and median filter were analyzed. Rest of ditches which covered by the border trees were extracted by Canny algorithm. Taking Dalate' s land remediation project area of Inner Mongolia as an example,taking World View-2 as the data source,system was carried out to verify the proposed method. Result showed that the extraction accuracy was 91. 46%,which met the demand of land consolidation projects. This method solved the problem that current main algorithms could not extract micro bending and tree-coved ditches accurately. It also provided an effective technical supporting farmland grading and farmland linear-object facilities monitoring.