MRI Image Fusion Based on Optimized Dictionary Learning and Binary Map Refining in Gradient Domain

The insufficient ability of edge feature extraction and high complexity limit the ability of sparse representation to obtain better medical image fusion performance. In this letter, we propose a novel multimodal medical image fusion method with optimized dictionary learning and binary map refining. The optimized dictionary learning uses loop iterations between separable FISTA and manifold-based conjugate gradient algorithm to catch detail texture features in detail layer, and the binary map refining solution adopts Gabor energy measurement with GDGIF to reserve structure and brightness characteristics in base layer. Experimental results of various medical images and clinical applications indicate the effectiveness of the proposed method.