测绘通报 ›› 2021, Vol. 0 ›› Issue (2): 59-63.doi: 10.13474/j.cnki.11-2246.2021.0044

• 学术研究 • 上一篇    下一篇

像素级和对象级高分辨率遥感图像变化检测分析

赖正文1, 夏小云2   

  1. 1. 广州航海学院艺术设计学院, 广东 广州 510725;
    2. 嘉兴学院数理与信息工程学院, 浙江 嘉兴 314001
  • 收稿日期:2020-04-16 修回日期:2020-06-10 出版日期:2021-02-25 发布日期:2021-03-09
  • 作者简介:赖正文(1974-),男,硕士,讲师,主要研究方向为数字图像处理技术。E-mail:45731689@qq.com
  • 基金资助:
    国家自然科学基金(61703183)

Analysis on change detection of high-resolution remote sensing image combining pixel level and object level

LAI Zhengwen1, XIA Xiaoyun2   

  1. 1. School of Art and Design, Guangzhou Navigation University, Guangzhou 510725, China;
    2. School of Mathematics, Physics and Information Engineering, Jiaxing University, Jiaxing 314001, China
  • Received:2020-04-16 Revised:2020-06-10 Online:2021-02-25 Published:2021-03-09

摘要: 高分辨率遥感图像具有丰富的纹理信息,而像素级变化检测方法主要分析图像的光谱信息,导致将像素级变化检测方法用于高分辨率遥感图像具有一定的局限性。因此,本文提出了一种像素级与对象级相结合的高分辨率遥感图像变化检测方法,解决了像素级与对象级变化检测方法中存在的椒盐现象、误检等问题。首先,结合高分辨率遥感图像的多维特征,构建遥感图像变化检测模型;其次,利用随机森林分类器对图像进行分类,得到像素级变化检测结果;最后,将像素级变化检测结果与图像对象分割结果进行融合,得到图像变化区域和不变区域。试验结果表明,该算法具有较高的准确率和检测精度。

关键词: 遥感图像, 变化检测, 随机森林, 像素级, 对象级

Abstract: High-resolution remote sensing images have rich texture information, and pixel-level change detection methods mainly analyze the spectral information of the image. Therefore, the use of pixel-level change detection methods for high-resolution remote sensing images has certain limitations. In view of this, a high-resolution remote sensing image change detection method combining pixel level and object level is proposed to solve the problems of salt and pepper and misdetection in the pixel level and object level change detection methods. First, combining the multi-dimensional features of high-resolution remote sensing images, a remote sensing image change detection model is constructed. Secondly, the random forest classifier is used to classify the image, and the pixel-level change detection result is obtained. Finally, the pixel-level change detection result and the image object segmentation result are fused to obtain the image change area and the invariant area. Experimental results show that the algorithm has high accuracy and detection accuracy.

Key words: remote sensing image, change detection, random forest, pixel level, object level

中图分类号: