A novel time-frequency concentration-keeping method of cross-terms suppression

Abstract

At present, there are mainly two types of methods in cross-terms suppression of time-frequency distribution: one is improving of Cohen class kernel functions, the other is combination with linear time-frequency analysis. Their advantages and disadvantages are analyzed with examples respectively in this paper, and then time-frequency image features are analyzed. By introducing edge detection and ridge extraction methods in image processing, a new method of cross-terms suppression based on image feature matching is proposed to deal with time-frequency image of Gabor transform and Wigner-Ville distribution. For multi-component signal, theoretical analysis and simulation of time-frequency distribution methods of cross-terms suppression show that, this novel method can effectively suppress cross terms at the same time maintain Wigner distribution's time-frequency concentration.