Gradient constrained bi-dimensional empirical mode decomposition and its application

In order to avoid the shortcomings of the capacity of getting the image details through the traditional bi-dimensional empirical mode decomposition (BEMD), an improved bi-dimensional Empirical Mode Decomposition method is proposed based on the gradient and local extrema. It can gain the high frequency edge information of the image by the gradient´s strong mining capacity to the image detail information. In addition, a new fusion strategy is realized by using non negative matrix factorization method as the fusion rule. The result shows that this method owns better detail capture capability than traditional enhancement and fusion algorithm.