Extended depth of focus through image fusion using the wavelet transform
Jesse Berent
Section Microtechnique, EPFL
Semester Project June 2003
This project consists in an extended depth of focus algorithm to obtain in focus microscopic images of 3D objects and organisms. The idea is to merge a stack of photographs taken at different focal lengths into a single totally focused composite image. The approach used to determine which parts of the original pictures are focused is based on the wavelet transform. Using a maximum absolute coefficient selection rule provides good results; however we have added post-processing in order to determine the topology and maintain the convexity property, thus avoiding saturation and accumulation of noise. In order to process colour pictures efficiently, a method to convert RGB images to greyscale preserving maximum contrast and intensity changes for the detection of salient features has been implemented such that the topographical map can be produced using only the greyscale images. The colour pixels are then retrieved from the original stack.