Combining Level Set and orthogonal transforms for speckle reduction and detail preservation in SAR images

Abstract

This paper introduces a new approach for SAR image processing by combining a Level Set method and DCT or Haar based filtering with small-sized object (target) preservation. The approach exploits the Level Set method ability to detect small-sized targets and other heterogeneities in one-look amplitude SAR image It is demonstrated that the proposed approach provides efficient noise suppression in homogeneous image regions with simultaneous preservation of useful information (details) in SAR images. Moreover, we perform quantitative analysis for synthetic images and qualitative study for real data.