Color Image Fusion in Quaternion Spectral Domain

Most of the traditional image fusion technologies based on grayscale images have shown good fusion performance. But few studies have focused on the fusion of color images. There are more and more works showing the benefits of quaternion representation-based color image processing. A novel color image fusion method in quaternion spectral domain is proposed in our paper. It represents color images as quaternion matrices, and converts them to spectral domain by discrete quaternion Fourier transform (DQFT). The fusion mechanism considers the high-pass and low-pass information of the integrated images. The low-pass part is integrated based on quaternion sparse representation. The high-pass information is fused by the max-absolute rule. Eventually, we reconstruct the two parts and get the fused image in spatial domain. The experimental results verify that the proposed method is effective in medical image fusion task.