High quality multi-focus polychromatic composite image fusion algorithm based on filtering in frequency domain and synthesis in space domain

摘要

A novel multi-focus polychromatic image fusion algorithm based on filtering in the frequency domain using fast Fourier transform (FFT) and synthesis in the space domain (FFDSSD) is presented in this paper. First, the original multi-focus images are transformed into their frequency data by FFT for easy and accurate clarity determination. Then a Gaussian low-pass filter is used to filter the high frequency information corresponding to the image saliencies. After an inverse FFT, the filtered images are obtained. The deviation between the filtered images and the original ones, representing the clarity of the image, is used to select the pixels from the multi-focus images to reconstruct a completely focused image. These operations in space domain preserve the original information as much as possible and are relatively insensitive to misregistration scenarios with respect to transform domain methods. The polychromatic noise is well considered and successfully avoided while the information in different chromatic channels is preserved. A natural, nice-looking fused microscopic image for human visual evaluations is obtained in a dedicated experiment. The experimental results indicate that the proposed algorithm has a good performance in objective quality metrics and runtime efficiency.