遥感影像融合的自适应变化检测

• 作 者: 魏立飞 钟燕飞 张良培 李平湘
• 机 构: 武汉大学测绘遥感信息工程国家重点实验室,
• 出 处: 遥感学报
• 发表日期: 2010年06期
• 关键词: 变化检测 遥感影像融合 自适应 阈值范围 融合影像 变化区域 segmentation iteration pixel algebraic

提出一种基于影像融合和自适应阈值选择的遥感影像变化检测方法。首先利用经过改进的融合技术对原始数据的差值影像和比值影像进行处理,构造融合影像,在该融合影像的基础上进行自适应迭代运算得到初步变化阈值范围,然后通过分析阈值范围两侧影像像元的离散程度,求解最终的阈值范围,从而得到更优变化阈值,提取变化区域。实验结果表明,本文方法的检测精度优于传统的变化检测方法,同时具有一定的稳定性和智能性。

A remote sensing image change detection method based on image fusion and adaptive threshold selection is proposed. Firstly, the improved fusion technique is used to process the difference image and the ratio image of the original data, the fusion image is constructed, and the initial variation threshold range is obtained through the adaptive iteration operation based on the fusion image. Then the images on both sides of the threshold range are analyzed Pixel discrete degree, solving the final threshold range, so as to obtain a better threshold of change, to extract the changing area. The experimental results show that the proposed method is superior to the traditional method of change detection and has some stability and intelligence.