结合颜色直方图和LBP纹理的遥感影像分割

基金项目:
国家自然科学基金项目"基于多尺度一体化分割的高分辨率SAR与光学图像变化检测"

江苏省资源环境信息工程重点实验室开放基金资助项目"高光谱影像全尺度分割与仿生模式分类方法"

国家863计划资助项目"多传感器信息实时接入"

详细信息

刊出日期: 2017-09-15

Combinational color histogram and LBP textural features for remote sensing image segmentation

Publish Date: 15 September 2017

摘要: 提出了一种结合颜色直方图特征和LBP纹理特征的高分辨率遥感影像分割方法.首先对梯度图像进行双阈值联合的自适应标记,进而通过快速分水岭变换来获得初始分割结果,然后以一种结合颜色直方图特征和LBP纹理特征的区域相似性度为指标对初始分割区域进行合并获得最终的分割结果.通过在高分辨率遥感影像上的分割对比实验,证明了该方法能充分利用遥感影像的光谱和纹理信息,分割效果良好.
- 标记分水岭分割 /
- 颜色直方图 /
- LBP纹理 /
- 区域合并
Abstract: A novel segmentation method combining color histogram with LBP textural features for segmentation of high-resolution remote sensing images is presented.This method starts with an adaptive marker-based watershed algorithm to obtain an initial segmentation result, and the markers are constructed by dual-threshold joint segmentation of the gradient image.And then a regional similarity indicator combining color histogram with LBP textural features is adopted to guide regional merging procedure and obtain the final result.Comparative experiments on high-resolution remote sensing images have proved the effectiveness of the method.
- marker-based watershed /
- color histogram /
- LBP textural feature /
- region merging

中国自然资源航空物探遥感中心	主办
地质出版社	出版