Multi-Focus Image Fusion Using the Local Fractal Dimension

Abstract

Abstract With the development of sensor and image-processing technology, image fusion has become a promising research field. Multi-focus image fusion is an important issue in multi-sensor image fusion. To make full use of the texture features of an image and take into account the inherent advantages of fractal theory in multi-focus image fusion, a new image fusion algorithm using the local fractal dimension (LFD) is proposed. The algorithm first calculates the LFD of each source image pixel-wise by using a blanket method and generates LFD maps of the source images. Then the local energy of each LFD is calculated to generate a decision map, in order to decide pixels of the fused image are from which source image. Finally, the fused image is reconstructed from the source images according to the decision map. Experimental results show that the proposed algorithm outperforms classic LP-based and DWT-based methods according to both visual and objective evaluations.