基于最优尺度和规则的高分辨率影像分类研究

高分辨率影像的广泛应用推进面向对象影像分析(OBIA)的发展,而分割作为面向对象分类的关键步骤,其尺度的选择直接关系到地物信息的提取。空间尺度是地物的固有属性,在合适的分割尺度下可以更好地挖掘地物信息。本文结合最大面积法和分割质量评价模型对张山营镇影像进行分割实验,先通过分析对象最大面积初步得到最优尺度范围,后结合分割质量评价模型以确定最优分割尺度层次。在此基础上,综合样本提取的光谱、纹理等特征进行规则训练,最终完成面向对象的土地覆被分类研究。结果显示:基于多层次最优尺度的规则分类方法获得更好的分类结果,其总体精度为88.8%,Kappa系数为0.861,而基于单一尺度的最邻近法总体精度81.4%,Kappa系数0.773,基于单一尺度的规则分类法总体精度为83.2%,Kappa系数为0.85。

Classification of high-resolution image based on optimal scale and rule

The development of object-oriented classification technology has been promoted by extensive application of high-resolution image, and with segmentation as one key step of object-oriented classification, the selection of segmentation scale determines the result of information extraction.Spatial scale is an inherent attribute of the object, and the information of the object can be better recognized under the appropriate segmentation scale.Based on QuickBird image of Zhangshanying town, a segmentation experiment has been carried out in the scope of 30~200scale.An optimal scale range is obtained by analyzing the maximum area of the objects, and then the final scale level is determined by the segmentation quality assessment model.According to the spectrum and texture feature provided by the samples, the rules of object-oriented classification are built to finish the extraction of land cover type.The result shows that the method based on optimal scales and rules is more effective in high-resolution image information extraction than that based on single scale, which overall accuracy reaches 88.8%, and Kappa coefficient is 0.861.While the overall accuracy of the nearest neighbor method based on single scale is 81.4% and the overall accuracy of the method based on rules and single scale is 83.2%.