方差齐性检验与模糊评价的遥感影像变化检测

发布时间：2018-05-07 04:41

本文选题：高分辨率影像 + 变化检测　；参考：《测绘科学》2017年04期

【摘要】：针对高分辨率遥感影像地物空间结构复杂、光谱纹理信息丰富的特点,该文提出一种基于Levene检验及模糊评价的遥感影像变化检测方法。首先,对预处理后的两期影像叠置分割得到图斑对象,提取其光谱、纹理等特征;其次,利用改进的Levene检验法统计各图斑变化信息得到候选变化图斑,经IHS变换获取影像色调信息,结合纹理特征进行变化向量分析;最后,通过迭代的方式确定阈值并计算候选图斑的变化隶属度,确定最终变化区域。选取江苏省某地区ZY-3高分辨率遥感影像进行实验,结果表明:改进的Levene检验法很大程度地减少了后续的计算量,提高了变化检测精度,对于相对辐射差异较大的非变化图斑具有良好的容错性;采用色调-纹理特征及模糊评价法可以有效地降低漏检及虚检率,提高变化检测精度。
[Abstract]:In view of the complex spatial structure and rich spectral texture information of high resolution remote sensing images, a method of remote sensing image change detection based on Levene test and fuzzy evaluation is proposed in this paper. First of all, after the preprocessing of the two stages of image overlay segmentation to obtain the image spot object, extract its spectral, texture and other features; secondly, using the improved Levene test method to calculate the change of each image spot information to obtain candidate change patterns, The color information of the image is obtained by IHS transform, and the change vector analysis is carried out with the texture feature. Finally, the threshold value is determined by iterative method and the membership degree of the candidate image spot is calculated to determine the final change region. The ZY-3 high-resolution remote sensing images in a certain area of Jiangsu Province are selected for experiments. The results show that the improved Levene method greatly reduces the amount of subsequent calculation and improves the accuracy of change detection. The method of hue texture feature and fuzzy evaluation can effectively reduce the false detection rate and improve the accuracy of change detection.
【作者单位】：中国矿业大学环境与测绘学院;国家测绘地理信息局卫星测绘应用中心;
【基金】：高分辨率对地观测系统重大专项(AH1601) 测绘地理信息公益性行业科研专项(201412007,201512027)
【分类号】：P237

【相似文献】