Enhancing BEMD decomposition using adaptive support size for CSRBF functions

Abstract

Despite their widespread development, the Fourier transform and wavelet transform are still unsuitable for analyzing non-stationary and non-linear signals. To address this limitation, bidimensional empirical mode decomposition (BEMD) has emerged as a promising technique. BEMD effectively extracts structures at various scales and frequencies but faces significant computational complexity, primarily during the extremum interpolation phase. To mitigate this, different interpolation functions were presented and suggested, with BEMD using compactly supported radial basis functions (BEMD-CSRBF) showing promising results in reducing computational cost while maintaining decomposition quality. However, the choice of support size for CSRBF functions significantly impacts the quality of BEMD. This article presents an enhancement to the BEMD-CSRBF algorithm by adjusting the CSRBF support size based on the extrema distribution of the image. Our method’s results show a significant improvement in the BEMD-CSRBF algorithm’s quality. Furthermore, when compared to the other two approaches to BEMD, it shows higher accuracy in terms of both intrinsic mode function (IMF) quality and computational efficiency.