Benefiting from the growing mass of big data and the rapid development of computing efficiency, a series of advanced models of machine learning and deep learning have been proposed one after another, and have achieved great success in many application scenarios. The research purpose of this paper is to fuse the image multi-features of the restricted Boltzmann machine and the neural network in the experiment, using the convolutional restricted Boltzmann machine, to establish a model and investigate the three datasets in the ship, oil, and machine. Image fusion performance evaluation results on the test set. The results show that the model can improve the performance of image fusion to a certain extent.