论文部分内容阅读
由于全变分(Total Variation,TV)模型具有较好的去噪、增强和扩散等功能,在过去的几十年中,TV模型在图像去噪、增强和超分辨率重建等方面得到了深入研究与广泛应用。鉴于TV模型的理论与分割理论具有一致性,因此本文主要研究TV模型用于高分辨率遥感影像的分割,并针对地物多尺度特征,提出了自适应的TV(ATV)模型;且与目前流行的面向对象的影像分析软件eCognition中的FNEA分割方法进行了比较。实验采用2幅高分辨率遥感影像,同时采用了面向对象的分割和分类评价,得出各方法各具优缺点的结论。
Since the Total Variation (TV) model has some functions such as denoising, enhancement and diffusion, the TV model has been deeply studied in image de-noising, enhancement and super-resolution reconstruction in the past few decades Research and extensive application. In view of the consistency between theory and segmentation theory of TV model, this dissertation mainly studies the TV model for high-resolution remote sensing image segmentation and proposes an adaptive TV (ATV) model for multi-scale features. The popular object-oriented image analysis software eCognition FNEA segmentation method is compared. Experiments using two high-resolution remote sensing images, using object-oriented segmentation and classification evaluation, draw the conclusions of the advantages and disadvantages of each method.