测绘通报 ›› 2018, Vol. 0 ›› Issue (9): 59-63.doi: 10.13474/j.cnki.11-2246.2018.0280

Previous Articles     Next Articles

Edge Detection Method for SAR Image Based on Improved Ball Tensor Voting

ZHANG Ying, LI Aixia, ZHAO Hong   

  1. College of Surveying & Municipal Engineering, Zhejiang University of Water Resources and Electric Power, Hangzhou 310018, China
  • Received:2018-04-02 Revised:2018-04-16 Online:2018-09-25 Published:2018-09-29
  • Contact: 李爱霞。E-mail:liax_zj@sina.com E-mail:liax_zj@sina.com

Abstract:

Tensor voting is a robust method for feature extraction,which collects the tensor information in neighborhood based on the principles of smoothness,proximity and connecting,and then detects the geometric structures.According to the results of ball tensor voting,we propose an edge detection method for SAR image based on the iterative ball tensor voting in the paper.The edge detection method votes for the iterative ball tensor on the basis of a certain scale range.Using a generalized nonlinear mapping function,the eigenvalues of each pixel were enhanced in the process of iteration,which increased the accuracy of the edge detection.The proposed method has been tested by the real SAR images,and the results of the experiments demonstrate its effectiveness and superiority in terms of the accuracy and integrity of the edge location.

Key words: SAR, edge detection, tensor voting, image enhancement, multi-scale

CLC Number: