Multi-focus image fusion using adaptive patch rendering anisotropic diffusion filter

Multi-focus image fusion (MFIF) extracts various focused regions from partially focused images of the same scene which are subsequently merged to create a composite image in which all objects are visualised precisely. Because of the diffusion of spatial intensities at the edges, many conventional MFIF methods face difficulties with respect to spatially inconsistent structures, visual distortion, ghost artifacts, and preserving edge information in the fused image. To address these issues, we proposed the Adaptive Patch Rendering Anisotropic Diffusion Filter (APRADF) for MFIF. To acquire large- and small-scale intensity variations, we first decompose the two source images into a base layer and a detail layer. We further capture the focus and blur patches of the source images to detect the diffusion of the edges at the boundaries. The proposed APRADF is then applied on the weight maps, base layer and detail layer. At last, the final fused image will be produced by adding the fused detail layer to the fused base layer linearly. Both qualitative and quantitative investigations were conducted on publicly accessible databases to evaluate the performance of the proposed method. The experimental results reveal that APRADF-based fusion outperforms state-of-the-art algorithms in subjective and objective analysis.