摘要
针对高分遥感图像区域合并分割过程中错误检测和修复自动化程度较低的问题,基于逐级合并且兼容多种区域合并准则的特性,设计了一种面向高分遥感图像区域合并分割过程中的影像分割错误无监督检测与修复方法。该方法利用不同区域合并准则的优势,通过对交叉评估局部不同区域合并模式,选择最佳合并模式并锁定优化结果。通过基于熵的无监督评估方法,在3幅土地覆盖类型各异的影像验证了本方法的有效性,结果表明提出的方法能够有效检测并修复分割错误。研究实现了区域合并分割过程中自动错误无监督检测与修复,为高分遥感图像分割后处理阶段的分割结果优化提供了方法参考。
To address the issue of low automation in error detection and repair during the process of merging and segmenting high-resolution remote sensing images.Based on the characteristics of step-by-step merging and compatibility with multiple region merging criteria,this paper proposes an unsupervised detection and repair method for image segmentation errors in the process of region merging and segmentation of high-resolution remote sensing images.This method utilizes the advantages of different region merging criteria,selects the best merging mode through cross evaluation of local different region merging modes,and locks in the optimization results.The effectiveness of this method was validated through an entropy based unsupervised evaluation method on three images with different land cover types,and the results showed that the proposed method can effectively detect and repair segmentation errors.The research has achieved automatic unsupervised detection and repair of errors in the process of region merging and segmentation,providing a method reference for optimizing segmentation results in the post-processing stage of high-resolution remote sensing image segmentation.
作者
车森
赵云鹏
张曦
CHE Sen;ZHAO Yunpeng;ZHANG Xi(Information Engineering University,Zhengzhou 450001,China)
出处
《测绘科学》
CSCD
北大核心
2023年第11期82-90,共9页
Science of Surveying and Mapping
关键词
面向对象影像分析
区域合并分割
分割错误
分割尺度
object-based image analysis
region-merging image segmentation
segmentation errors
segmentation scale
作者简介
车森(1978—),副教授,博士,主要研究方向为空间数据融合与可视化。E-mail:chxycs@126.com
