Autofocus and Fusion using Nonlinear Correlation

摘要

In this work a new algorithm is proposed for auto focusing and images fusion captured by microscope's CCD. The proposed algorithm for auto focusing implements the spiral scanning of each image in the stack f(x,y)_w to define the V_w vector. The spectrum of the vector FV_w is calculated by fast Fourier transform. The best in-focus image is determined by a focus measure that is obtained by the FV_I nonlinear correlation vector,of the reference image, with each other FV_w images in the stack. In addition, fusion is performed with a subset of selected images f(x, y)_(SBF) like the images with best focus measurement. Fusion creates a new improved image f(x, y)_F with the selection of pixels of higher intensity.