首页> 外文会议>International Conference on Agro-Geoinformatics >Auto-extraction method of farmland irrigation and drainage system based on domestic high-resolution satellite images
【24h】

Auto-extraction method of farmland irrigation and drainage system based on domestic high-resolution satellite images

机译:基于国内高分辨率卫星影像的农田排灌系统自动提取方法

获取原文

摘要

Farmland irrigation and drainage system is one of the key hydraulic engineering facilities on farmland, large-scale, fast and accurate auto-extraction method for farmland irrigation and drainage system is a significant direction for remote sensing application. This paper proposes an intelligent extraction method for farmland irrigation and drainage system based on domestic high resolution satellite GF-2 images, which has considered both spectral features and geometry features of farmland irrigation and drainage system, and could be divided into four parts as: image scale converting, spectral model for canal identification, data extraction by spatial features and breakpoint connecting with morphology. The first step is to fuse the 1 meter panchromatic image and 4 meter multi-spectral image by Nearest Neighbor Diffusion pan sharpening algorithm that will output a high resolution multi-spectral image in 1meter. Then, construct a model of spectral relationship by red, green, blue and NIR (Near Infrared Reflection) bands, which will extract irrigation canals initially from images. Then, in order to distinguish roads with irrigation canals, we need to analyze the spatial features of them and design spatial rules to separate these two targets. The last problem is that there would be many disconnected irrigation channels as limited by the resolution of remote sensing images, the mathematical morphology method would be used for judging the topological relationships between breakpoints, which will be connected by dilation operators. This paper chose the Sanhulianjiang reservoir irrigation area to be experimental area, which is located in Hubei's Jiayu County. The main irrigation and drainage facilities in experimental area have been extracted through our method, and are contrasted to the water resource survey data. The comparison shows the accuracy of this method is credible; it could satisfy the needs as large-scale, fast extraction for irrigation and drainage system, which has huge potential in agriculture and water conservancy fields.
机译:农田排灌系统是农田的重要水利工程设施之一,大规模,快速,准确的农田排灌自动提取方法是遥感应用的重要方向。提出了一种基于国产高分辨率卫星GF-2图像的农田灌溉排水系统智能提取方法,该方法兼顾了农田灌溉排水系统的频谱特征和几何特征,可以分为四个部分:图像比例转换,用于运河识别的光谱模型,通过空间特征提取数据以及与形态相关的断点。第一步是通过最近邻扩散泛锐化算法将1米的全色图像和4米的多光谱图像融合在一起,该算法将在1米内输出高分辨率的多光谱图像。然后,通过红,绿,蓝和NIR(近红外反射)波段构建光谱关系模型,这将首先从图像中提取灌溉渠。然后,为了区分带有灌溉渠的道路,我们需要分析它们的空间特征并设计空间规则以将这两个目标分开。最后一个问题是,由于遥感图像的分辨率的限制,灌溉通道将断开,数学形态学方法将被用于判断断点之间的拓扑关系,这些断点将由膨胀算子连接起来。本文选择湖北省嘉reservoir县三湖连江水库灌区为试验区。通过本方法提取了实验区主要的排灌设施,并与水资源调查数据进行了对比。比较表明该方法的准确性是可信的。它可以满足灌溉排水系统大规模,快速提取的需求,在农业和水利领域具有巨大的潜力。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号