Contents
2023
Volume: 45 Issue 7
14 Article(s)
Export citation format
Infrared and Visible Image Fusion Combining Information Perception and Multiscale Features
Haicheng QU, Qianqian HU, and Xuecong ZHANG
Existing image fusion algorithms based on deep learning are unable to satisfy the demands of computational efficiency and fusion effect. Most have also adopted a fusion strategy based on a single-scale model, which cannot effectively extract the contextual information in images. This study proposes an end-to-end image Existing image fusion algorithms based on deep learning are unable to satisfy the demands of computational efficiency and fusion effect. Most have also adopted a fusion strategy based on a single-scale model, which cannot effectively extract the contextual information in images. This study proposes an end-to-end image fusion network based on information perception and multiscale features. The network consists of an encoder, a fusion strategy, and decoder. Specifically, the multiscale features of the infrared and visible images were extracted by the encoder, and a feature complementary enhancement module was designed to fuse different modal multiscale features. Finally, the lightweight decoder was designed to combine the low-level details and high-level semantic information. In addition, the information entropy of the source image was used to construct an information-sensing loss function to train the fusion network and generate the fused image with rich information. The proposed fusion framework was evaluated on the TNO and MSRS datasets. The results show that compared with existing fusion methods, the proposed network was superior to other methods in terms of both subjective visual description and objective index evaluation, with higher computational efficiency..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 685 (2023)
A Panchromatic and Multispectral Image Fusion Method Combining Energy and Structural Information
Jiawen SONG, Daming ZHU, Xiaoqing ZUO, Zhitao FU, and Sijing CHEN
Component substitution is a classical method for remote-sensing image fusion that has good spatial fidelity but is prone to spectral distortion. Therefore, a panchromatic and multispectral image fusion method that combines structural and energy information is proposed. First, the method decomposes the spatial and spectComponent substitution is a classical method for remote-sensing image fusion that has good spatial fidelity but is prone to spectral distortion. Therefore, a panchromatic and multispectral image fusion method that combines structural and energy information is proposed. First, the method decomposes the spatial and spectral information of multispectral images by hyperspherical color-space transformation. Second, a two-layer decomposition scheme is introduced through joint bilateral filtering. The panchromatic image and intensity components are then decomposed into structural and energy layers. Finally, the structural layer is fused by the neighborhood spatial frequency strategy, and the pure energy layer of the intensity component is used as the energy layer of the pre-fusion image. The intensity component defines the color intensity. By combining the pre-fused structural layer with the energy layer of the intensity component, the spatial and spectral information of the source image can be effectively combined, thereby reducing the spectral distortion of the pansharpened image. In this study, several experiments were conducted on the Pléiades and QuickBird datasets, and the experimental results were qualitatively and quantitatively analyzed. The results show that the proposed method has certain advantages over existing methods..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 696 (2023)
Infrared and Visible Image Fusion Based on Saliency Detection and Latent Low-Rank Representation
Zhiliang LONG, Yueming DENG, Runmin WANG, and Jun DONG
To address the problems of missing detail and low contrast in the fusion of infrared and visible images, this study proposes a fusion method based on saliency detection and latent low-rank representation. First, a pre-fusion image is obtained by saliency detection for the infrared and visible images. Then, the infraredTo address the problems of missing detail and low contrast in the fusion of infrared and visible images, this study proposes a fusion method based on saliency detection and latent low-rank representation. First, a pre-fusion image is obtained by saliency detection for the infrared and visible images. Then, the infrared, visible, and pre-fused images are decomposed into low-rank and detail layers by the multilevel latent low-rank representation method. The detail layer is fused by combining the hyperspherical L2 norm and structural similarities, while the low-rank layer is fused using an approach based on the energy property. The final fused image is obtained by adding the fusion results of the low-rank and detail layers. The proposed method is compared with 11 representative image fusion methods by conducting subjective and objective evaluations of multiple groups of fused images. The results show that the image fusion method enhances the effective detail information and improves the image contrast, yielding a fusion result that is more in line with people's visual understanding..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 705 (2023)
Night Vision Anti-halation Method Based on Partition Fusion of Visible and Infrared Images
Jiahao LIANG, Quanmin GUO, and Hanlei WANG
The imperfect visual effects in fused images are caused by the large difference between the regional features of night-vision halation images. To address this problem, a partition fusion method for visible and infrared images is proposed. First, the halation threshold of the low-frequency coefficient, determined by theThe imperfect visual effects in fused images are caused by the large difference between the regional features of night-vision halation images. To address this problem, a partition fusion method for visible and infrared images is proposed. First, the halation threshold of the low-frequency coefficient, determined by the adaptive threshold iteration method, divided the low-frequency coefficient into halation and nonhalation regions. In the halation region, the proposed nonlinear adjustment method for the infrared coefficient weights eliminated halation according to the degree of halation in the image. In the nonhalation region, the weight adjustment method based on the prior grayscale mean was applied to improve the weight of brighter images participating in the fusion to enhance the visibility of dark areas. The experimental results show that the proposed method can be applied to night-vision halation scenes of different degrees to eliminate halation and improve the quality of night-vision image fusion..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 714 (2023)
Infrared and Visible Light Image Fusion Method Based on Swin Transformer and Hybrid Feature Aggregation
Bicao LI, Jiaxi LU, Zhoufeng LIU, Chunlei LI, and Jie ZHANG
The fusion of infrared and visible light images can generate images containing more information in line with human visual perception compared with the original images, and is also beneficial for downstream tasks. Traditional image fusion methods based on signal processing have problems such as poor generalization abiliThe fusion of infrared and visible light images can generate images containing more information in line with human visual perception compared with the original images, and is also beneficial for downstream tasks. Traditional image fusion methods based on signal processing have problems such as poor generalization ability and reduced performance of complex image fusion. Deep learning is capable of features extraction and provides good results. However, its results have problems such as reduced preservation of textural details and blurred images. To address these problems, this study proposes a fusion network model of infrared and visible light images based on the multiscale Swin Transformer and an attention mechanism. Swin Transformers can extract long-distance semantic information from a multiscale perspective, and the attention mechanism can weaken the insignificant features in the proposed features to retain the main information. In addition, this study proposes a new hybrid fusion strategy and designs brightness enhancement and detail retention modules according to the respective characteristics of the infrared and visible images to retain more textural details and infrared target information. The fusion method has three parts: the encoder, fusion strategy, and decoder. First, the source image was input into the encoder to extract multiscale depth features. Then, a fusion strategy was designed to fuse the depth features of each scale. Finally, the fused image was reconstructed using a decoder based on nested connections. The experimental results on public datasets show that the proposed method has a better fusion performance compared with other state-of-the-art methods. Among the objective evaluation indicators, EI, AG, QP, EN, and SD were optimal. From a subjective perspective, the proposed infrared and visible light image fusion method can preserve additional edge details in the results..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 721 (2023)
Infrared and Visible Image Fusion Based on Dilated Convolution and Dual Attention Mechanism
Le HE, Zhongwei LI, Cai LUO, Peng REN, and Hao SUI
The multiscale features extraction method in infrared and visible image fusion algorithms loses detail information. Existing fusion strategies also cannot balance the visual detail and infrared target features. Therefore, a fusion network via a dilated convolution and dual-attention mechanism (DCDAM) is proposed. FirstThe multiscale features extraction method in infrared and visible image fusion algorithms loses detail information. Existing fusion strategies also cannot balance the visual detail and infrared target features. Therefore, a fusion network via a dilated convolution and dual-attention mechanism (DCDAM) is proposed. First, the network extracts the original features from the image through a multiscale encoder. The encoder systematically aggregates the multiscale context information through dilated convolution instead of using downsampling operator. Second, a dual-attention mechanism is introduced into the fusion strategy, and the original features are input into the attention module for feature enhancement to obtain the attention features. The original and attention features were combined into the final fusion feature. The mechanism captured the typical information without losing details and suppressed the noise during the fusion process. Finally, the decoder used a full-scale jump connection and dense network to decode the fusion features and generate the fused image. The experimental results show that the DCDAM is better than other representative methods in qualitative and quantitative index evaluations and has a good visual effect..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 732 (2023)
Multimodal Fusion Detection of UAV Target Based on Siamese Network
Ziqiang HAN, Mingkai YUE, Cong ZHANG, and Qi GAO
To address the threat of small drones "black flying" to the public domain. Based on the multimodal image information of an unmanned aerial vehicle (UAV) target, a lightweight multimodal adaptive fusion Siamese network is proposed in this paper. To design a new adaptive fusion strategy, this module assigns difTo address the threat of small drones "black flying" to the public domain. Based on the multimodal image information of an unmanned aerial vehicle (UAV) target, a lightweight multimodal adaptive fusion Siamese network is proposed in this paper. To design a new adaptive fusion strategy, this module assigns different modal weights by defining two model training parameters to achieve adaptive fusion. The structure is reconstructed on the basis of a Ghost PAN, and a pyramid fusion structure more suitable for UAV target detection is constructed. The results of ablation experiments show that each module of the algorithm in this study can improve the detection accuracy of the UAV targets. Multi-algorithm comparison experiments demonstrated the robustness of the algorithm. The mAP increased by 9% when the detection time was basically unchanged..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 739 (2023)
Multiscale Infrared Target Detection Based on Attention Mechanism
Xiangrong LI, and Lihui SUN
To address the problems of poor textural detail, low contrast, and poor target detection in infrared images, a multiscale infrared target detection model that integrates a channel attention mechanism is proposed based on Yolov4 (You Only Look Once version 4). First, the number of model parameters is reduced by reducingTo address the problems of poor textural detail, low contrast, and poor target detection in infrared images, a multiscale infrared target detection model that integrates a channel attention mechanism is proposed based on Yolov4 (You Only Look Once version 4). First, the number of model parameters is reduced by reducing the depth of the backbone feature extraction network. Second, to supplement the shallow high-resolution feature information, the multiscale feature fusion module is reconstructed to improve the utilization of the feature information. Finally, before the multiscale feature map is generated, the channel attention mechanism is integrated to further improve the infrared feature extraction ability and reduce noise interference. The experimental results show that the size of the algorithm model in this study was only 28.87% of the Yolov4. The detection accuracy of the infrared targets also significantly improved..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 746 (2023)
Weak and Small Infrared Target Detection Combined With Frame Difference Kernel Correlation Filtering
Yuping WANG, and Yi ZENG
To improve the performance of infrared target detection, weak and small infrared target detection combined with frame difference kernel correlation filtering is proposed. First, the current frame is trained by kernel correlation filtering to obtain the maximum regression value. Then, the difference value is calculated To improve the performance of infrared target detection, weak and small infrared target detection combined with frame difference kernel correlation filtering is proposed. First, the current frame is trained by kernel correlation filtering to obtain the maximum regression value. Then, the difference value is calculated relative to the interval frame to perform a cyclic shift to compensate for the background motion between frames. The relative motion features of the current frame are extracted using the interframe difference method, which enhances the ability to distinguish weak and small targets from the infrared background. Finally, threshold segmentation is performed on the relative motion features to obtain the final detection results. Simulation experiments show that the proposed algorithm effectively detected weak and small infrared targets in complex environments. Compared with similar algorithms, the proposed algorithm suppressed clutter and point-shaped interference sources, and achieved a higher target detection rate. Simultaneously, a large number of operations are placed in the frequency domain, and the operational efficiency is better than that of other algorithms..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 755 (2023)
Digital Twin Guidance Law for Missile to Hit Weak Ground Infrared Target
Liguang WANG, Junjie SHUI, Luhui XU, Jiong ZHAO, Changqing YU, and Yiming FAN
When an infrared-guided missile attacks a ground target, natural or human factors can cause the infrared characteristics of the target to weaken or even disappear. The seeker cannot or intermittently detects the target, which significantly affects the guidance accuracy. To solve this problem, a digital twin guidance laWhen an infrared-guided missile attacks a ground target, natural or human factors can cause the infrared characteristics of the target to weaken or even disappear. The seeker cannot or intermittently detects the target, which significantly affects the guidance accuracy. To solve this problem, a digital twin guidance law is proposed for hitting a weak infrared target on the ground. On the basis of an infrared seeker in the physical world, digital twin models of the target and guidance laws are developed in the digital world. The state parameters of the missile motion and control in the guidance process at each point in time are obtained by simulation, and saved as the digital twin of the guidance process. During guidance process, when the seeker cannot obtain the measurement signal, its digital twin is activated immediately to take over and provide the control system with the acceleration order. Simulations show that the digital twin of the seeker can provide the missile control system with the maneuver order to accurately guide the missile when the infrared seeker is unable to capture the signal. The digital twin guidance law is robust against infrared camouflage, interference, and bad weather, and has broad application prospects..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 768 (2023)
Study on Dither-caused Clutter for Staring Infrared Camera in Geostationary Orbit
Boyuan BIAN, Feng ZHOU, Xiaoman LI, Libing JIN, Hui GONG, and Minlong LIAN
For a staring infrared camera in geostationary orbit, the dither-caused clutter results from the combined effects of background features, camera parameters, camera line-of-sight dither characteristics, and background suppression algorithms. To quantitatively evaluate the intensity of dither-caused clutter, several timeFor a staring infrared camera in geostationary orbit, the dither-caused clutter results from the combined effects of background features, camera parameters, camera line-of-sight dither characteristics, and background suppression algorithms. To quantitatively evaluate the intensity of dither-caused clutter, several time-related factors, such as the dither spectrum, detector integration time, frame period, and interframe differential background suppression algorithm are considered. They are combined into a background-independent dither-equivalent angle, and the model of dither-caused clutter is established by multiplying the dither-equivalent angle and gradient statistics of the background radiation intensity. Based on ground measurement experiments on the camera line-of-sight dither characteristics, the spectrum of the camera line-of-sight dither caused by the cryocooler and momentum wheels are analyzed. The dither-caused clutter is also simulated and calculated to verify the theoretical model. The results show that the relative deviation between the calculated and simulated results was less than 15%. This indicates the high versatility and efficiency of the of model, which is suitable for the iterative optimization of camera design..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 775 (2023)
Research on Thin Silicon Windows Processing Technology for Infrared
Guoliang YIN, Rukun DONG, Changyu YING, Qiming XIE, Erping ZHANG, Hongmei ZHANG, and Wen YU
To meet the requirements of reduced load and high precision in photoelectric systems, thin silicon infrared windows were designed by increasing the surface accuracy compared with ordinary infrared window parts, and doubling the size of ordinary thin optical window parts (diameter to thickness ratio of 10:1). If the conTo meet the requirements of reduced load and high precision in photoelectric systems, thin silicon infrared windows were designed by increasing the surface accuracy compared with ordinary infrared window parts, and doubling the size of ordinary thin optical window parts (diameter to thickness ratio of 10:1). If the conventional process is used for processing, the parts are unable to meet the technical specifications. This study combines the processing advantages of classical polishing with the characteristics of silicon single-crystal materials; optimizes the adhesive ratio; adopts the adhesive point distribution method of bonding; and solves the processing problems of thin silicon infrared windows by changing relevant technical parameters, such as the spindle speed of the equipment and parts processing temperature. The processing of silicon windows by conventional and precision annealing were also compared..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 784 (2023)
Redundant Object Damage and Prevention Method for Infrared Detectors
Yongchang LEI, Jianlin LI, Wei DONG, Jiading ZHOU, Likun HOU, and Kunlun QIAN
Damage caused by foreign objects is a common risk faced by aerospace and aviation products, which has contributed to many incidents. The dynamic displacement of redundant objects can cause instantaneous circular image failure in infrared (IR) detectors, which interferes with the detection and tracking of small targets Damage caused by foreign objects is a common risk faced by aerospace and aviation products, which has contributed to many incidents. The dynamic displacement of redundant objects can cause instantaneous circular image failure in infrared (IR) detectors, which interferes with the detection and tracking of small targets with low infrared radiation. Moreover, the dynamic displacement of redundant objects can collide with the infrared focal plane array, causing ineffective pixels and affecting the minimum resolvable temperature difference and operating range of the IR imager. The faulty device must be dismantled step-by-step to eliminate the macroparticles. Electron microscopy was used to test the redundant objects. The source and generation of redundant objects during infrared focal plane array manufacturing and application were determined. Mechanical and optical analyses show that macroparticles can damage the intensity distribution of the imaging beam on the focal plane. The time required for the displacement to pass through the field of view is less than 50 ms, the linearity is less than 1 mm, and the diffraction phenomenon is significant. Most of the circular images occurred in the Fraunhofer diffraction area. A fringe diffraction spot is produced if the 10.m redundant object is close to the focal plane array..
Infrared Technology
- Publication Date: Jan. 01, 1900
- Vol. 45, Issue 7, 790 (2023)
© Copyright 2018-2021 | Chinese Laser Press. All Rights Reserved 沪ICP备15018463号-20