A Fast Bidimensional Empirical Mode Decomposition Based on Partial Differential Equation and Its Application on Image Processing

Based on partial differential equation(PDE),a fast bidimensional empirical mode decomposition algorithm(BEMD) is proposed to solve problems of the boundary effect and the low decomposition speed in the current BEMD algorithms.Firstly,a fourth order PDE is constructed for the envelope which is constrained by the extreme points,then the PDE is calculated iteratively by the difference to obtain the image's up and low envelopes.After the image shifted,the intrinsic mode functions(IMFs) can be achieved,and then the mode decomposition is realized.The residual image after decomposing is also applied to the margin detection and the preprocessing of face recognition in this paper.The experimental results show that the fast PDE-BEMD algorithm can not only effectively reduce the time and the space complexity but also avoid the boundary effect.It can obtain the intrinsic mode functions(IMFs) with the distinct margin information and at the same time,the residual image is not blurred.The algorithm has good performances in the margin detection and the image denoising.