Empirical Mode Decomposition And Its Application On Image Fusion And Coding

Empirical Mode Decomposition method is introduced by Norden E. Huang of NASA in 1998. It is a new processing method for nonlinear and non-stationary signal. This method will gradually decompose a complex nonlinear and non-stationary signal into some stable data layers and the remaining final trend. Each of smooth layers is called as intrinsic mode function, which represents the signal characteristics of different scales. EMD method can make time-frequency localized and is a self-adaptive method. It can control the pollution made by the odd and even signal in a small scope.As a new data analysis method, EMD method decomposes the signal from a new perspective. The development history of EMD is just a short time, and the method has the potential for image processing. So perfecting its theory and making it into image processing fields effectively have attracted the attentions of more and more scholars and have become an issue that has important theoretical and practical value.The paper first introduces the background and the development of EMD and expounds the application of EMD and analyzes the prospect that EMD is applied to image processing; then the paper makes a detailed explanation for EMD algorithms and introduces one-dimensional EMD and discusses how to extend EMD into image processing and gives a solution for the problems involved in EMD such as the extrema point locating, interpolation method selected, the border effect processing, sifting termination criteria and decomposition termination criteria selected. The paper introduces two fast EMD methods including extrema point optimizing and image blocking; combining with the multi-scale processing method and the natures of EMD, the paper introduces two image fusion methods based on EMD including frequency decompositon and multi-scale decomposition; finally aiming at image coding, the paper introduces several image coding methods based on EMD including entropy coding, extrema points extracting coding, sifting process coding and transform coding. The paper has made an elementary study on EMD and its application on image fusion and image coding and has made some innovation.