国土资源遥感2016，Vol.28Issue(4)：49-58,10.DOI:10.6046/gtzyyg.2016.04.08
面向对象的遥感影像最优分割尺度监督评价
Supervised evaluation of optimal segmentation scale with object-oriented method in remote sensing image
摘要
Abstract
The object - oriented classification quality of the remote sensing images depends not only on the classification algorithm but also on the goodness of the segmentation results. The quality of image segmentation determines the accuracy of subsequent classification of the remote sensing images. The quantitative method for determining the optimal segmentation scale and eliminating the interference of subjective factors becomes the focus of the image segmentation quality assessment. However, the importance of object recognition in image segmentation quality evaluation is often ignored in the previous segmentation quality evaluation method. After analyzing the complex spatial relations between the image objects and the actual image region, a new optimal segmentation scale evaluation index based on the area and position of the image object was proposed to evaluate the optimal segmentation scale. Based on the evaluation index, a WorldView2 multispectral image was used to be researched and the optimal segmentation parameters were determined. The results show that the segmentation scale evaluation index is effective in image segmentation quality assessment and parameter optimization. The experimental results have also shown the effectiveness of the method proposed in this paper for both segmentation quality assessment and optimal parameter selection. Also, the procedure of segmentation quality assessment can be conducted with less human intervention, making the result more objective.关键词
面向对象/最优分割尺度/监督评价/遥感影像Key words
object-oriented/optimal segmentation scale/supervised evaluation/remote sensing image分类
信息技术与安全科学引用本文复制引用
庄喜阳,赵书河,陈诚,丛佃敏,曲永超..面向对象的遥感影像最优分割尺度监督评价[J].国土资源遥感,2016,28(4):49-58,10.基金项目
国家重点研发计划项目(编号:2016YFB0502500)和中国科学院战略性先导科技专项“应对气候变化的碳收支认证及相关问题”(编号：XDA05050106)共同资助。 （编号:2016YFB0502500）
