基于eCognition的遥感图像面向对象分类方法对比分析

打开文本图片集

关键词：面向对象;eCognition；多尺度分割；影像分类

中图分类号：TP751 文献标志码：A 文章编号:1003-5168(2025)22-0117-04

DOI: 10.19968/j.cnki.hnkj.1003-5168.2025.22.020

A Comparative Analysis of Object-Oriented Classification Methods for Remote Sensing Images Based on eCognition

XIAO Lie TANGYulin ZENGXiaowei (Hunan Provincial Communications Planning,Survey & Design Institute Co.,Ltd., Changsha 41oooo, China)

Abstract: [Purposes] This study systematically evaluates the effectiveness of different image segmentation and classification algorithms in remote sensing image processing tasks,aiming to provide references for applications such as plant detection,and pest and disease identification.[Methods] Guided by the fundamental principle of object-based classification,experiments were conducted on the AISD dataset using eCognition software to perform a comparative analysis of various segmentation and classification algorithms.[Findings] Experimental results indicate that multiresolution segmentation is more conducive to delineating ground objects.Compared to other algorithms,the Bayesian and Random Tree algorithms yielded beter performance.[Conclusions] The object-oriented remote sensing image clasification method based on eCognition can effectively improve the classification accuracy and demonstrates reliability.

Keywords: object-oriented; eCognition; multiresolution segmentation;image classification

0 引言

随着遥感技术的进步，遥感影像的分辨率已达到分米级，影像上的地物信息更加丰富，显示也更加清晰1。（剩余5772字）