一种改进SLIC分割的多尺度面向对象CVA遥感影像变化检测方法

A Kind of Change Detection Method for CVA Remote Sensing Image with Multiscale Object Orientation Based on Improved SLIC Segmentation

依托无锡市江阴市地理国情普查项目,对江阴市示范区开展遥感影像变化检测方法研究,针对基于像元变化检测提取特征单一化以及面向对象方法严重依赖分割尺度的问题,融合多特征提出一种基于改进简单的线性迭代聚类(SLIC)分割的多尺度面向对象的变化矢量分析(CVA)变化检测方法。首先对预处理后的影像进行多尺度分割形成矢量层,然后用矢量层分割影像得到像斑,接着对2期像斑进行特征提取构成变化矢量,并对其进行分析得到变化强度,最后进行阈值分割和精度评定。实验结果证明,该方法综合利用各类型特征的优势,提高了算法的检测精度,为变化检测提供了新思路,可以作为地理国情监测的有效手段,同时也对城市遥感影像数据的快速更新具有一定参考价值。

Relying on the geographical information survey projrct in Jiangyin City,Wuxi Province,the study on change detection method for remote sensing image was carried out in Jiangyin Municipal Demonstration Area.For problems such as single extraction feature of change detection based on image element and heavy dependence of object orientation on segmentation scale,a kind of CVA(change vector analysis)change detection method with multiscale object orientation was proposed on the basis of improved SLIC(simple linear iterative clustering)segmentation by integrating multiple features.First of all,the preprocessed images were segmented at multiple scales to form a vector layer then used to segment the image for disk,and later the characteristics of the two-phase image disk was extracted to form the change vector that was analyzed to obtain the change intensity.Finally,the threshold was segmented and precision evaluated.The experimental results show that this method makes full use of the advantages of various types of features,and improves the detection accuracy of the algorithm,providing new ideas for change detection,therefore it can be used as an effective means,to monitor geographical information and is of reference value for the rapid update of urban remote sensing image.