Automatic focus and fusion image algorithm using nonlinear correlation: Image quality evaluation

We propose a new autofocus and fusion algorithm to enhance the characteristics of the image in less run-time. The autofocus method selects the best-focused image (BFI) in a stack of images captured at different distances from the object. A vector is defined for each image in the stack, which contains elements selected by spiral scanning of the image. The spectrum of each vector is calculated using the Fourier transform then applying non-linear correlation to the reference vector spectrum and those of each of the corresponding to the images in the stack the BFI is determined. The fusion is carried out with a subset of images that have a focus measure value close to the BFI ones. The parabolic filter is applied to determine the relevant elements of the images that will be included in the fusion. The results evaluation is conducted using a multi-image metric, a quality measure that represents the percentage improvement of the fused image. The comparison was made with other fusion methods such as standard wavelets, resulting in that the autofocus and fusion algorithm (AFA) method obtained the highest quality indices. It is concluded that the AFA improves the quality of the images in less time than the conventional fusion methods.