Edge detection using neighborhood limited empirical mode decomposition and time-frequency analysis

One novel edge detection algorithm based on neighborhood limited empirical mode decomposition (NLEMD) and its time-frequency analysis is proposed. Firstly, the image is decomposed by NLEMD, then bi-orthant Hilbert transform is applied in IMFpsilas and instantaneous frequency with instantaneous energy are extracted. Secondly, the IMFpsilas and instantaneous frequency with instantaneous energy are fused under certain principle to obtain fused image. Thirdly, using local adaptive threshold to extract edges. One can obtain different edges for multi-resolution, and experiments proved the validity of the new algorithm in this paper.