Image Feature Extraction Based on the Two-Dimensional Empirical Mode Decomposition

This paper presents the approach to extract the local features of the two-dimensional Intrinsic Mode Function (IMF) obtained by 2EMD and detect the edge features using the monogenic signal. The monogenic signal is a 2D-generalization of the analytic signal using the Riesz transform taking place of the Hilbert Transform. Analyzing each 2D-IMF by studying local properties (amplitude, phase, orientation and frequency) extracted from the monogenic signal of each one of them. According to these properties, we further to analyze the performance of 2EMD. The edge extraction based on the local properties is carried on and is demonstrated with both texture and natural images. The edge extracting time is faster than the typical methods.