Two-Scale Image Fusion Algorithm Based on Improved PCNN and DCT

To address the problems of the difficulty in setting the parameters of the pulse-coupled neural network model and the dependence of the image fusion algorithm based on discrete cosine transform on the block size, which affect the efficiency and robustness of image fusion, a two-scale image fusion algorithm based on the improved pulse coupled neural network and the discrete cosine transform is proposed. The algorithm combines the input information to improve the traditional pulse coupled neural network model framework, introduces the sine-cosine algorithm to set the network parameters. Then, it improves the fusion algorithm based on the discrete cosine transform to fuse the image, and reconstructs the fused image. Finally, it proposes the information compensation algorithm to compensate for some positions of the reconstructed image and obtains the final fusion result. The results of fusion experiments with seven algorithms on five datasets (multi-focused image dataset, TNO dataset and three brain image datasets with different modalities) show that the proposed algorithm performs better robustness for fusion of information-centric images, it is better than the other seven algorithms in the fusion efficiency of images of different sizes, and it has advantages in the fusion of information-centric images. © 2022 Institute of Computing Technology. All rights reserved.