A New Hybrid Approach with Perona-Malik Improvement for Multi-focus Image Fusion

Multi-focus image fusion methods combine two or more images, which have blurred and defocused parts, and create a clearer image called as all-in-focused image. A new hybrid approach for multi-focus image fusion is applied in this paper. Firstly, the stationary wavelet transform (SWT) is applied to each source image in order to analyze it easily. Four sub-bands, which are LL, LH, HL and HH, are created by using the SWT. Then the Perona - Malik diffusion approach is implemented to improve each sub-band of the source images. This method helps removing noise from the source images besides preserving image edges. The fusion rule is used to decide which parts of source images are important for fused image. The gradient-based fusion rule which preserves edges and important information of source images is applied as a fusion rule. After the fusion rule, the sub-bands of all-in-focused image are created. Finally, the inverse SWT is implemented to reconstruct all-in-focused image. The proposed method is compared with other methods in the literature based on subjective quality metrics, which are RMSE, PSNR and the running time criterion. The experimental results showed that the proposed method is more successful in this area.