Multi-Focus Image Fusion using Gradients of Wavelet Coefficients

A new image fusion method is developed to fuse multi-focused images by manipulating gradients of wavelet coefficients of the source images. The proposed approach consists of two phases. The first phase generates a binary decision map based on input images' activity levels. In second phase, the decision map is processed so that it separates the focused and defocused parts of the input images. Then using the processed decision map, fused image is constructed by copying the focused regions from each of the source images. The experimental results show that the proposed technique can achieve better performance than classical wavelet transform based methods.