Relativeenergy And Entropy Based Wavelet Packetdecomposition Of Images
Multiresolution signal analysis via orthogonal discrete wavelet transform (ODWT) separates signals into lower and higher frequency contents. For images as 2-D signals, each decomposition level consists of a low pass filtered partrepresenting the approximation of the image and high pass filtered parts regarding the details such as noise, edges, sharp transitions between intensity values of pixels etc. of the image. In this study, new method is proposed for the extraction of the necessary information from high and mid frequency bands. For this reason, relative energy and entropy concept is applied to the subimages of the original image in order to understand which frequency bands should be further decomposed via multiresolution signal analysis. This is accomplished by searching the relevant information from the child subimage, which has the same positional edge information with its parent image or the image it is revealed from after decomposition.
Index Terms- discrete wavelet transform,entropy,multiresolution signal analysis, relative energy