Efficient and Model-Based Infrared and Visible Image Fusion via Algorithm Unrolling

Infrared and visible image fusion (IVIF) expects to obtain images that retain thermal radiation information from infrared images and texture details from visible images. In this paper, a model-based convolutional neural network (CNN) model, referred to as Algorithm Unrolling Image Fusion (AUIF), is proposed to overcome the shortcomings of traditional CNN-based IVIF models. The proposed AUIF model starts with the iterative formulas of two traditional optimization models, which are established to accomplish two-scale decomposition, i.e., separating low-frequency base information and high-frequency detail information from source images. Then the algorithm unrolling is implemented where each iteration is mapped to a CNN layer and each optimization model is transformed into a trainable neural network. Compared with the general network architectures, the proposed framework combines the model-based prior information and is designed more reasonably. After the unrolling operation, our model contains two decomposers (encoders) and an additional reconstructor (decoder). In the training phase, this network is trained to reconstruct the input image. While in the test phase, the base (or detail) decomposed feature maps of infrared/visible images are merged respectively by an extra fusion layer, and then the decoder outputs the fusion image. Qualitative and quantitative comparisons demonstrate the superiority of our model, which can robustly generate fusion images containing highlight targets and legible details, exceeding the state-of-the-art methods. Furthermore, our network has fewer weights and faster speed.

I. Introduction

Image fusion, as an image enhancement technology, is a hot issue in the image processing research community. By merging the images obtained by different sensors in the same scene, we expect to obtain images that highlight the advantages of each source image and are robust to perturbations at the same time [1]–[5]. Image fusion can effectively improve the utilization of image information, eliminate conflicts and redundancies among multiple sensors, while form a clear and complete description of targets to facilitate recognition and tracking in subsequence [6]–[8]. Infrared and visible image fusion, abbreviated as IVIF, is a typical topic in image fusion [9]–[12]. By incorporating prior knowledge to the images during the preprocessing stage, IVIF is effective to make full use of information in images and widely used in fire control [13], autonomous driving [14], [15] and face recognition [16], etc.