IVOMFuse: An image fusion method based on infrared-to-visible object mapping

Due to the significant difference in the structure of infrared and visible images, most of the existing algorithms use globally consistent strategy, which may weaken the contrast of the target region. To solve this problem, we propose an infrared and visible image fusion method based on infrared-to-visible object mapping (IVOMFuse) First of all, we use the probability induced intuitionistic fuzzy C-means algorithm (PIIFCM) to process the infrared image, and extract the target region of the infrared image according to the image structure features. However, because of the poor contrast of the target of the visible image, it can not be segmented directly. For this reason, we map the infrared target region to the visible image, and the target region with the same coordinates of the two images can be obtained very quickly. Secondly, due to the significant differences in image structure, we set a hybrid fusion strategies for different regions. We use the fusion strategy based on the Expectation-Maximization (EM) algorithm to iterate and optimize the probability model for the target region. The fourth-order partial differential equation (FPDE) is used for the background region to decompose the source image into high-frequency texture and low-frequency approximation images respectively. Then, the principal component analysis (PCA) and the average fusion strategy are used to obtain the fused image. Finally, we evaluated our algorithm with fourteen other state-of-the-art image fusion methods on three commonly used datasets, namely TNO, CVC14, and RoadScene. Among the ten evaluation metrics, our algorithm achieved the best six times. (c) 2023 Elsevier Inc. All rights reserved.

Automated summary

Due to the difference in structure between infrared and visible images, existing algorithms often use globally consistent strategies, which may compromise the contrast of target regions. To address this issue, we propose an infrared and visible image fusion method called IVOMFuse, which utilizes infrared-to-visible object mapping. Our method involves processing the infrared image using the PIIFCM algorithm to extract the target region. The target region is then mapped to the visible image, allowing for quick retrieval of the corresponding target region. Additionally, we employ a hybrid fusion strategy for different regions based on the EM algorithm and FPDE to optimize the probability model and decompose the source image into high-frequency and low-frequency approximation images. Finally, our algorithm achieves superior performance compared to 14 state-of-the-art fusion methods on three commonly used datasets.