基于面向对象的高分辨率遥感影像目标信息提取

Target information extraction of high resolution remote sensing image with Object-Oriented

分析比较多光谱遥感影像分割任务中应用较多的3种分割算法,得出分形网络演化(FNEA)方法在影像对象获取中的优势明显。综合考虑目标地物的光谱、形状、纹理结构等属性特征,上下文语义特征,空间关系特征,辅以地物的边缘信息,提取研究区高分辨率遥感影像地物信息,并对分类结果进行精度评价,实验表明,基于面向对象的FNEA算法在高分辨率遥感影像目标信息提取方面具有良好的应用价值。

Three algorithms used in multi-spectral remote sensing image segmentation were analyzed, and Fractal Net Evolution Approach was believed to have obvious advantages in high resolution image segmentation. Spectral properties, shape, texture structure was used to extract the target information with the semantic feature, spatial relation. The classification results of high resolution remote sensing imagery was evaluated by the accuracy. Experimental results shew that image target information extraction based on FNEA Object-Oriented has good application value.