中文

Journal of Applied Sciences ›› 2015, Vol. 33 ›› Issue (1): 21-31.doi: 10.3969/j.issn.0255-8297.2015.01.003

• Signal and Information Processing • Previous Articles     Next Articles

Segmentation of SAR Image of Rivers Based on Shearlet Transform and Krawtchouk Moment Invariants

  

  1. 1. College of Electronic and Information Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China
    2. Engineering Technology Research Center of Wuhan Intelligent Basin, Changjiang River Scientific Research Institute, Changjiang Water Resources Commission, Wuhan 430010, China 3. Key Laboratory of the Yellow River Sediment of Ministry of Water Resource, Yellow Institute of Hydraulic Research, Yellow River Water Resources Commission, Zhengzhou 450003, China
    4. State Key Laboratory of Urban Water Resource and Environment, Harbin Institute Technology, Harbin 150090, China
    5. Key Laboratory of Port, Waterway and Sedimentation Engineering of the Ministry
    of Transport, Nanjing Hydraulic Research Institute, Nanjing 210024, China
  • Received:2014-09-13 Revised:2014-12-22 Online:2015-01-30 Published:2015-01-04

PDF

887

Abstract: Segmentation of synthetic aperture radar (SAR) images is a key procedure in river detection and recognition. To further improve accuracy of SAR image segmentation for rivers, a segmentation method is proposed based on Shearlet transform, Krawtchouk moment invariants, and fuzzy local information C-means (FLICM). The SAR image is first decomposed with Shearlet transform, and its texture features are extracted as the first part of the feature vector. Krawtchouk moment invariants of the image are then calculated to obtain corresponding shape features used as the second part of the feature vector. Finally,
the image is clustered based on FLICM algorithm using the extracted feature vector. Thus segmentation of the river SAR image is obtained. A large number of experiments are performed. The results are compared with several recently proposed methods based on pulse coupled neural network (PCNN) combined with maximum variance ratio, and Gabor wavelet transform combined with fuzzy C-means (FCM) and FLICM clustering. It has been shown that the proposed method has clear advantages both in subjective visual effects and in terms of objective evaluation indicator such as segmentation error rate. The method can
provide river segmentation with better accuracy.
 

Key words: river detection, SAR image segmentation, Shearlet transform, Krawtchouk moment invariant, fuzzy local information C-means clustering (FLICM)

CLC Number: