SAR Segmentation using Anisotropic Diffusion

DOI Web Site Web Site 参考文献34件
  • Intajag Sathit
    King Mongkut's Institute of Technology Ladkrabang, Department of Instrumentation Engineering, Faculty of Engineering
  • Tipsuwanporn Vittaya
    King Mongkut's Institute of Technology Ladkrabang, Department of Instrumentation Engineering, Faculty of Engineering
  • Cheevasuwit Fusak
    King Mongkut's Institute of Technology Ladkrabang, Department of Instrumentation Engineering, Faculty of Engineering

この論文をさがす

抄録

Speckle effects are commonly observed in synthetic aperture radar (SAR) images. The human eye is capable of deriving meaningful information from SAR images; however, an automatic or semi-automatic processing algorithm has difficulty in distinguishing objects in the images because of noise effects present in those images. This paper presents a segmentation method for SAR images, which employs an anisotropic diffusion algorithm. In the proposed scheme, a SAR image is transformed into a logarithmic domain where the diffusion process is used to grow homogeneous regions in the noise environment until the regions reach some criteria for homogeneity; consequently, the segmented image in the logarithm domain is converted to the intensity domain by an exponential function. To grow homogeneous regions the adaptive diffusion method is introduced with a tensor technique in which tensor data are varied with the neighboring pixels. The diffusion algorithm will stop itself by a standard deviation divided by the mean, which is provided according to the homogeneity criteria. Results are shown on both synthetic and satellite SAR images. The evaluation of the proposed method employs the theoretical gain of equivalent numbers of looks (ENL).

収録刊行物

参考文献 (34)*注記

もっと見る

キーワード

詳細情報 詳細情報について

問題の指摘

ページトップへ