Lattice Fusion Networks for Image Denoising

A novel method for feature fusion in convolutional neural networks is proposed in this work. Different feature fusion techniques are suggested to facilitate the flow of information and improve the training of deep neural networks. Some of these techniques as well as the proposed network can be considered a type of Directed Acyclic Graph (DAG) Network, where a layer can receive inputs from other layers and have outputs to other layers. In the proposed general framework of Lattice Fusion Network (LFN), feature maps of each convolutional layer are passed to other layers based on a lattice graph structure, where nodes are convolutional layers. To investigate the performance of the proposed network, two specific designs based on the general framework of LFN were implemented for the task of image denoising. Results were compared with state of the art methods. The proposed network achieved better results with far fewer learnable parameters, which shows the effectiveness of LFNs for training of deep neural networks. LFN is able to outperform the stat of the art DnCNN with half (52%) the number of learnable parameters.