Multimodal Image Fusion based on Improved Pulse-Coupled Neural Network and Convolutional Sparse Representation in NSST Domain

Multi-mode medical image fusion combines image functions of different modes and plays a vital role in the diagnosis of various diseases, treatment planning and follow-up research. In view of this situation, an image fusion method based on non-subsampled shearlet transform (NSST) and an improved pulse coupled neural network model was proposed. In the proposed method, the non-subsampled shearlet transform (NSST) method was used to decompose the source image into low frequency sub-bands and several high frequency sub-bands. The improved sparse representation was used to fuse low-frequency sub-bands, which could eliminate detailed features through the sobel operator and the guided filter, thereby improving the ability to effectively conserve energy. At the same time, the high frequency sub-bands were fused by the parameter adaptive pulse coupled neural network (PAPCNN). With four other different types of medical image fusion methods to verify the effectiveness of the proposed method. Experimental results show that this method is better than the other four comparison methods in subjective visual performance and objective evaluation.