SAR Image De-noising Based on Curvelet Domain Hidden Markov Tree Models
Hai Jin · Chinese Journal of Computers · 2007
Based on the statistical property of SAR image speckle noise and combining curvelet transform with HMT models,a method of SAR image de-noising based on curvelet domain hidden Markov tree(HMT) models is presented in this paper.Using HMT models to capture the scale dependencies among curvelet coefficients,it implements the image de-noising and reduces SAR speckle noise effectively,furthermore,analyze the algorithm mechanism and computation complexity.De-noising performance is evaluated through subjective inspection,as well as objective measurements—flatness index and edge save index.Results clearly demonstrate the superiority of this new approach when compared to conventional wavelet domain HMT and curvelet transform.FI value obtained is appropriate and ESI vertically and horizontally are averagely increased by about from 0.2 to 0.3.