论文部分内容阅读
针对高分辨率遥感影像分类与信息提取中存在的难点,基于不同目标地物在高分辨率影像上具有对应最优分割尺度的基本思想,该文在分析现有最优分割尺度确定方法的基础上,提出了加权均值法结合最大面积的最优分割尺度的确定方法;利用该方法,进行了高分辨率影像分割实验,获取了对应典型地物的最优分割尺度数值范围,实现了典型地物的信息提取;并运用样本点检验的方法,计算并分析了分类的精度结果。结果表明:基于加权均值与最大面积相结合的最优分割尺度计算方法,应用于面向对象高分辨率影像信息的提取具有较为理想的精度。
Aiming at the difficulties in classification and information extraction of high-resolution remote sensing images, based on the basic idea that different target objects have the corresponding optimal segmentation scale in high-resolution images, this paper analyzes the basis of the existing methods for determining the optimal segmentation scale , A method of determining the optimal segmentation scale based on the weighted average method combined with the maximum area is proposed. By using this method, the high resolution image segmentation experiment is carried out, and the optimal segmentation scale value range corresponding to the typical features is obtained. Extract the information of the object; and use the sample point test method to calculate and analyze the classification accuracy results. The results show that the optimal segmentation method based on weighted average and maximum area is more suitable for the extraction of object-oriented high-resolution image information.