Multifocus image fusion in wavelet domain

Abstract

How to select wavelet filters, decomposition levels and fusion schemes is of great concern fur multifocus image fusion in wavelet domain. We presented a simple image fusion algorithm based on wavelet transform. By wavelet transform, an image can be represented by a low frequency approximation, which contain the average information of the image, and several high frequency details with different scales and directions, which contain the texture or edge feature of the image. For the multifocus images, there are some areas unclear in certain source images which correspond to small wavelet coefficients, and clear in other source images which correspond to large coefficients. So we simply took the coefficients with greater modulus as the final coefficients to get the fusion image, which contains more details and is clearer in visual and more convenient for image analysis and understanding. We also discussed the influence of the wavelet filters, decomposition levels and fusion schemes on the fusion results, compared wavelet based fusion with other multiresolution fusions, such as Laplacian pyramid, gradient pyramid, contrast pyramid and ratio pyramid. Some results are obtained, which are of great value for research and experiment in this field.