基于差异图融合的变化检测分析-change detection and analysis based on difference graph fusion.docxVIP

  • 89
  • 0
  • 约3.81万字
  • 约 59页
  • 2018-05-18 发布于上海
  • 举报

基于差异图融合的变化检测分析-change detection and analysis based on difference graph fusion.docx

基于差异图融合的变化检测分析-change detection and analysis based on difference graph fusion

摘要 遥感图像变化检测是指通过对同一地区不同的时期的两幅或多幅遥感图像进 行比较分析,来得到所需的地物变化信息。在农作物生长状况监测、森林采伐监 测、国土资源勘查、灾情的估计等民用方面以及战区打击效果的评估等军用方面 都有着广泛而重要的应用。本文介绍了遥感图像变化检测的研究背景以及存在的问题,对已有变化检测 技术进行了总结,并以差异图融合为主要研究内容,针对两时相的单波段遥感图 像进行了变化检测方法的研究。(1)提出了一种基于方差信息的差异图融合方法。该方法首先提取已有差值 差异图的方差信息,在利用方差信息的基础上实现差值差异图与比值差异图的融 合,一定程度上克服了差值图对噪声敏感和比值图易丢失边缘信息的缺点,更好 地保持了细节信息,减小信息丢失和累积误差,提高了检测精度。(2)提出了一种基于 Treelet 差异图融合的变化检测方法,首先提取差值差异 图和比值差异图,然后对两种差异图进行中值滤波与均值滤波,产生中值滤波后 的比值差异图和差值差异图,均值滤波后的比值差异图和差值差异图,用 Treelet 方法对这四种差异图进行降维处理,以实现图像的融合。并将该方法与未融合的 单幅图像分割结果以及仅融合两种差异图的方法进行比较,显示了该方法综合了 各个差异图的优势,对不同的变化检测图像数据均取得了较好的结果,提高了遥 感图像变化检测的检测精度。(3)为了进一步提高基于 Treelet 差异图融合的变化检测方法的精度,取得更 好的检测效果,本文将小波变换与 Treelet 方法相结合,进行差异图融合。并通过 实验,将该方法与 Treelet 差异图融合的变化检测方法进行了对比,该方法取得了 较好的检测结果,验证了该方法的可行性。本文工作得到如下基金资助:国家自然科学基金(No.;高等学校 学科创新引智计划(111 计划)(No. B07048);中央高校基本科研业务费专项资 金资助(Nos.K50510020015, K5051202019)。关键词:遥感图像 变化检测 图像融合 Treelet 图像聚类分割 小波分解AbstractChange detection in multitemporal remote sensing image is a process that analyzes a series of images of the same region acquired at different time to obtain the changes occurred in the area. It has been widely used in the civil economy, such as crop growth condition monitoring, land use and land cover, assessments of environment disaster, etc. It also plays an important role in national defense, such as the assessment of military strike.In this paper, we introduce the background and problems of change detection in remote sensing images, and summarize the traditional methods proposed in this area. Our work is focused on difference image fusion to get a higher quality diffe rence image. It can be summarize as follows:Difference method and ratio method are weak at noise tolerance and preserving detail information respectively. A new change detection method based on image fusion to overcome the above shortcomings is proposed. The two difference images acquired by difference method and ratio method are fused based on the variance of difference image of difference method. The experiments carried ou

您可能关注的文档

文档评论(0)

1亿VIP精品文档

相关文档