Images captured from different viewpoints or devices have often exhibited significant geometric and photometric differences due to factors such as environmental variations, camera technology differences, and shooting conditions’ instability. To address this problem, homography estimation has attracted much attention as a method to describe the geometric projection relationship between images. Researchers have proposed numerous homography estimation methods for single-source and multimodal images in the past decades. However, the comprehensive review and analysis of homography estimation methods, from feature-based to deep learning-based, is still lacking. Therefore, we provide a comprehensive overview of research advances in homography estimation methods. First, we provide a detailed introduction to homography estimation’s core principles and matrix representations. Then, we review homography estimation methods for single-source and multimodal images, from feature-based to deep learning-based methods. Specifically, we analyze traditional and learning-based methods for feature-based homography estimation methods in detail. For deep learning-based homography estimation methods, we explore supervised, unsupervised, and other methods in-depth. Subsequently, we specifically review several metrics used to evaluate these methods. After that, we analyze the relevant applications of homography estimation and show the broad application prospects of this technique. Finally, we discuss current challenges and future research directions, providing a reference for computer vision researchers and engineers.

The homography estimation of infrared and visible images, a key technique for assisting perception, is an integral element within the 6G Space–Air–Ground Integrated Network (6G SAGIN) framework. It is widely applied in the registration of these two image types, leading to enhanced environmental perception and improved efficiency in perception computation. However, the traditional estimation methods are frequently challenged by insufficient feature points and the low similarity in features when dealing with these images, which results in poor performance. Deep-learning-based methods have attempted to address these issues by leveraging strong deep feature extraction capabilities but often overlook the importance of precisely guided feature matching in regression networks. Consequently, exactly acquiring feature correlations between multi-modal images remains a complex task. In this study, we propose a feature correlation transformer method, devised to offer explicit guidance for feature matching for the task of homography estimation between infrared and visible images. First, we propose a feature patch, which is used as a basic unit for correlation computation, thus effectively coping with modal differences in infrared and visible images. Additionally, we propose a novel cross-image attention mechanism to identify correlations between varied modal images, thus transforming the multi-source images homography estimation problem into a single-source images problem by achieving source-to-target image mapping in the feature dimension. Lastly, we propose a feature correlation loss (FCL) to induce the network into learning a distinctive target feature map, further enhancing source-to-target image mapping. To validate the effectiveness of the newly proposed components, we conducted extensive experiments to demonstrate the superiority of our method compared with existing methods in both quantitative and qualitative aspects.

Homography estimation for infrared and visible images is a critical and fundamental task in multimodal image processing. Recently, the coarse-to-fine strategy has been gradually applied to the homography estimation task and has proved to be effective. However, current coarse-to-fine homography estimation methods typically require the introduction of additional neural networks to acquire multi-scale feature maps and the design of complex homography matrix fusion strategies. In this paper, we propose a new unsupervised homography estimation method for infrared and visible images. First, we design a novel coarse-to-fine strategy. This strategy utilizes different stages in the regression network to obtain multi-scale feature maps, enabling the progressive refinement of the homography matrix. Second, we design a local correlation transformer (LCTrans), which aims to capture the intrinsic connections between local features more precisely, thus highlighting the features crucial for homography estimation. Finally, we design an average feature correlation loss (AFCL) to enhance the robustness of the model. Through extensive experiments, we validated the effectiveness of all the proposed components. Experimental results demonstrate that our method outperforms existing methods on synthetic benchmark datasets in both qualitative and quantitative comparisons.

Single-resolution homography estimation of infrared and visible images is a significant and challenging research area within the field of computing, which has attracted a great deal of attention. However, due to the large modal differences between infrared and visible images, existing methods are difficult to stably and accurately extract and match features between the two image types at a single resolution, which results in poor performance on the homography estimation task. To address this issue, this paper proposes an end-to-end unsupervised single-resolution infrared and visible image homography estimation method based on graph neural network (GNN), homoViG. Firstly, the method employs a triple attention shallow feature extractor to capture cross-dimensional feature dependencies and enhance feature representation effectively. Secondly, Vision GNN (ViG) is utilized as the backbone network to transform the feature point matching problem into a graph node matching problem. Finally, this paper proposes a new homography estimator, residual fusion vision graph neural network (RFViG), to reduce the feature redundancy caused by the frequent residual operations of ViG. Meanwhile, RFViG replaces the residual connections with an attention feature fusion module, highlighting the important features in the low-level feature graph. Furthermore, this model introduces detail feature loss and feature identity loss in the optimization phase, facilitating network optimization. Through extensive experimentation, we demonstrate the efficacy of all proposed components. The experimental results demonstrate that homoViG outperforms existing methods on synthetic benchmark datasets in both qualitative and quantitative comparisons.