The wide application of the image super-resolution algorithms significantly improves the visual quality of infrared images. In this paper, an infrared image super-resolution reconstruction method based on a closed-loop regression network is proposed. The residual channel attention block is introduced into the up-sampling module group, which effectively improves the capacity of the network and increases the number of feature maps, enhances the extraction and recovery ability of infrared image features, and is conducive to the recovery of image details. Compared with other infrared information recovery methods previously proposed, the proposed method has obvious advantages in high-resolution detail acquisition capability. Neural network through closed-loop regression, this scheme overcomes the LR image to HR image defects in nonlinear mapping function, by introducing additional constraints on the LR data to reduce the space of the possible functions.
Terahertz waves refer to electromagnetic waves with frequencies ranging from 0.1THz to 10THz.Due to the ability to penetrate many non-polar materials, terahertz waves can be used to detect hidden objects. A convolutional neural network structure called Attention U-Net to achieve super-resolution of terahertz images is proposed in this paper. The function of the convolutional layer and pooling layer in the encoding path is reducing the size and extracting the edge features of the image, while the role of the deconvolution layer in the decoding path is to up-sample the image and restore the image content. The introduction of skip connection on the feature map of the symmetrical encoding path and decoding path maximizes the utilization of feature information in each layer of the network and effectively solves the problem of gradient disappearance. This network also replaces the convolution on the codec path with the attention mechanism block, including the spatial attention mechanism and the channel attention mechanism, which makes the extracted features more directional, obtains more detailed information about the target that needs to be focused and suppresses other Useless information. The network and algorithm proposed in this paper have good results in experiments and have a wide range of application prospects in the field of security inspection.
We propose a novel network for low-light-level and visible image fusion enhancement task, which is based on the feature extraction convolution neural network. By extracting the high-frequency information of visible light detector under low illumination, and combining the advantages of wide activation network and channel attention mechanism, the network can automatically filter and extract the useful information in the image to complete the super-resolution reconstruction of low light level image. It makes up for the lack of visible light information and low resolution (LR) of low light level detector at night and can realize all-weather real-time imaging. The experimental results show that our method has better numerical performance than the traditional super-resolution network structure, and also retains more abundant image texture information, which is more in line with the feeling of human eyes.
In recent decades, with the rapid development of image sensor technology, image acquisition has gradually evolved from a single sensor mode to a multi-sensor mode. The data information obtained by a single sensor is limited, and the use of multi-source data fusion can provide a more accurate understanding of the observation scene. This paper proposes a network structure of infrared visible color night vision image fusion based on deep learning. The network adopts a fusion-encoding-decoding structure for end-to-end learning to achieve the purpose of color night vision image fusion, making the image more in line with human visual effects. The fusion structure contains a multi-scale feature extraction block and a channel attention block, which perform feature extraction on low-resolution infrared images and visible images respectively. The multi-scale feature block can expand the receptive field and avoid losing too much feature information. The channel attention block can improve the sensitivity of the network to channel characteristics. A certain number of convolutional layers and deconvolutional layers are used in the network to realize the encoding and decoding of the feature map to achieve the purpose of restoring the color fusion image. After experimental verification, our method has a good fusion effect, rich colors, and conforms to the human visual effect.
Single image super-resolution (SISR) is a notoriously challenging ill-posed problem that aims to obtain a high-resolution output from one of its low-resolution versions. Recently, powerful deep learning algorithms have been applied to SISR and have achieved state-of-the-art performance. In this paper, a high-efficiency infrared image super-resolution algorithm based on a cascaded deep network is proposed. In this method, the low-resolution infrared image is directly processed without the preprocessing of bicubic interpolation up-sampling that can reduce the complexity of the network and the amount of computation. The network structure consists of two layers of the network. The sub-pixel convolution in each layer can enlarge the image size by twice and make the input image size to reach the final high-resolution image size. Besides, we utilize multi-scale feature extraction blocks to extract features from the same feature image by using multiple convolution kernels of different sizes, which makes the feature image information more abundant. The experimental results show that the test speed of each image in our network is 0.046 seconds, which manifests our proposed algorithm has high efficiency of infrared image super-resolution.
In recent years, the convolution neural network has been widely used in single image super-resolution and has an excellent super-resolution ability. In this paper, a novel convolutional neural network structure based on symmetric skip connection is proposed, which contains multiple convolution layers and deconvolution layers. The role of the convolution layer is to extract the details of image content, and the function of the deconvolution layer is to make the image upsampling and restore the image content details. In addition, we use skip connection between the convolution layer and the deconvolution layer of network structure, which can transfer image information from the front end to the back end. Meanwhile, skip connection can also effectively solve the problem of gradient vanishing. Besides, the residual block is introduced to deepen the network structure. The deeper network structure can learn more complex changes. Different from other papers, this paper uses the method of adding the number of channels for feature fusion. This method can greatly increase the number of feature images, which is helpful to restore image details by deconvolution layer. A large number of experiments show that our network has efficient super-resolution ability of infrared image details.
Convolution neural network has been successfully applied to the super-resolution method of the visible image. In this paper, we propose an infrared image super-resolution imaging algorithm based on auxiliary convolution neural network, which uses the detail information provided by the visible image under low-light conditions for super-resolution imaging of infrared image. In this algorithm, infrared image and visible image are input into the convolution neural network at the same time to obtain high resolution infrared image. The results show that the super-resolution infrared image has more detailed information. Compared with other super-resolution methods, the proposed network can obtain the high super-resolution reconstruction efficiency.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.