Attention Network with Information Distillation for Super-Resolution

Resolution is an intuitive assessment for the visual quality of images, which is limited by physical devices. Recently, image super-resolution (SR) models based on deep convolutional neural networks (CNNs) have made significant progress. However, most existing SR models require high computational costs with network depth, hindering practical application. In addition, these models treat intermediate features equally and rarely explore the discriminative capacity hidden in their abundant features. To tackle these issues, we propose an attention network with information distillation(AIDN) for efficient and accurate image super-resolution, which adaptively modulates the feature responses by modeling the interactions between channel dimension and spatial features. Specifically, gated channel transformation (GCT) is introduced to gather global contextual information among different channels to modulate intermediate high-level features. Moreover, a recalibrated attention module (RAM) is proposed to rescale these feature responses, and RAM concentrates the essential contents around spatial locations. Benefiting from the gated channel transformation and spatial information masks working jointly, our proposed AIDN can obtain a more powerful ability to identify information. It effectively improves computational efficiency while improving reconstruction accuracy. Comprehensive quantitative and qualitative evaluations demonstrate that our AIDN outperforms state-of-the-art models in terms of reconstruction performance and visual quality.     

基于密集连接结构的超分辨精简网络

近年来, 随着深度神经网络的发展并被应用在超分辨领域, 图像超分辨率重建的效果得到了明显的提升。但是之前的工作大都把精力放在如何设计深度模型来提高重建的效果上, 而忽视了设计模型需要大量参数与计算量这一问题, 严重制约了深度学习方法在图像超分辨率重建方面的实际应用范围。针对该问题, 基于密集连接结构设计了一种新的网络。在以下3个方面进行了算法改进:1)提出了一种基于密集连接结构的新模型; 2)加入1×1卷积层作为特征选择层, 同时进一步减少计算量; 3)探讨了通道数量与重建精度、计算量之间的关系。实验结果表明本文提出的模型取得了与其他卷积神经网络模型相近的复原精度, 同时计算速度只有之前最快深度模型FSRCNN的一半以下。     

基于深度反向投影的感知增强超分辨率重建模型

以SRCNN(super-resolution convolutional neural network)模型为代表的超分辨率重建模型通常都有很高的PSNR(peak signal to noise ratio)和SSIM(structural similarity)值,但其在视觉感知上并不令人满意,而以SRGAN为代...     

基于GoogLeNet和空间谱变换的高光谱图像超分辨率方法

为了提高高光谱图像的空间分辨率,提出了一种基于GoogLeNet和空间谱变换的高光谱图像超分辨率(SR)方法.设计出遥感图像的光谱SR框架,对图像中不同反射光谱进行提取;采用GoogLeNet的稀疏编码对粗像素光谱进行放大,并投影到高分辨率字典上,将潜在SR表示进行反转,以获得超分辨光谱;为了提高图像重构的保真度,利用...     

多尺度融合超分辨率算法在无人机探测中的应用

基于光电传感器的低慢小无人机探测系统能够快速准确地发现并识别无人机目标,但远距离非合作无人机目标在图像中像素比重过小,特征退化较明显,使识别率大大降低.图像超分辨技术能够从低分辨率目标图像区域中获得高分辨率图像并恢复更多的细节特征,现有超分辨技术很难在保证推理速度的前提下兼容图像的高低频特征,因此为了满足探测系统的需求...     

宽带电磁图像卷积神经网络盲恢复方法研究

在利用抛物反射面对电磁干扰源成像过程中,由于系统衍射受限及成像频带较宽,导致干扰源成像模糊,分辨率低,难以分辨,不同频率不同区域干扰源所成图像分辨率不同,采用已有超分辨算法难以提高分辨率。为了实现宽带电磁图像的盲复原, 应用卷积神经网络的方法。网络训练是直接输入模糊图像,不假设任何特定的模糊和噪声模型情况下,重建出高质量图像。实验和仿真结果证明了卷积神经网络盲恢复方法在宽频带不同成像区域下表现了优于其他盲恢复算法的优势。     

Single infrared image super-resolution combining non-local means with kernel regression

In many infrared imaging systems, the focal plane array is not sufficient dense to adequately sample the scene with the desired field of view. Therefore, there are not enough high frequency details in the infrared image generally. Super-resolution (SR) technology can be used to increase the resolution of low-resolution (LR) infrared image. In this paper, a novel super-resolution algorithm is proposed based on non-local means (NLM) and steering kernel regression (SKR). Based on that there are a large number of similar patches within an infrared image, NLM method can abstract the non-local similarity information and then the value of high-resolution (HR) pixel can be estimated. SKR method is derived based on the local smoothness of the natural images. In this paper the SKR is used to give the regularization term which can restrict the image noise and protect image edges. The estimated SR image is obtained by minimizing a cost function. In the experiments the proposed algorithm is compared with state-of-the-art algorithms. The comparison results show that the proposed method is robust to the noise and it can restore higher quality image both in quantitative term and visual effect.     

Improving Image Super-Resolution Based on Multiscale Generative Adversarial Networks

Convolutional neural networks have greatly improved the performance of image super-resolution. However, perceptual networks have problems such as blurred line structures and a lack of high-frequency information when reconstructing image textures. To mitigate these issues, a generative adversarial network based on multiscale asynchronous learning is proposed in this paper, whereby a pyramid structure is employed in the network model to integrate high-frequency information at different scales. Our scheme employs a U-net as a discriminator to focus on the consistency of adjacent pixels in the input image and uses the LPIPS loss for perceptual extreme super-resolution with stronger supervision. Experiments on benchmark datasets and independent datasets Set5, Set14, BSD100, and SunHays80 show that our approach is effective in restoring detailed texture information from low-resolution images.     

Point target detection utilizing super-resolution strategy for infrared scanning oversampling system

To improve the resolution of remote sensing infrared images, infrared scanning oversampling system is employed with information amount quadrupled, which contributes to the target detection. Generally the image data from double-line detector of infrared scanning oversampling system is shuffled to a whole oversampled image to be post-processed, whereas the aliasing between neighboring pixels leads to image degradation with a great impact on target detection. This paper formulates a point target detection method utilizing super-resolution (SR) strategy concerning infrared scanning oversampling system, with an accelerated SR strategy proposed to realize fast de-aliasing of the oversampled image and an adaptive MRF-based regularization designed to achieve the preserving and aggregation of target energy. Extensive experiments demonstrate the superior detection performance, robustness and efficiency of the proposed method compared with other state-of-the-art approaches.     

Rapid Diffusion Weighted Imaging with Enhanced Resolution

In this paper, a new, fast compressively sensed diffusion magnetic resonance image enhancement technique is presented. This algorithm aims to overcome two major obstacles—image resolution limitation and algorithm reconstruction time efficiency-by combining a highly sparse k–q-space sampling pattern with super-resolution (SR) image enhancement. Similar to the RoSA (rotating single-shot acquisition) acceleration scheme, the presented algorithm takes advantage of simultaneous k–q-space sampling procedures being able to implement directly with no hardware modifications. The method sequentially processes compressively sensed k-space’s semi-PROPELLER blades with respect to appropriately synchronized diffusion directions. The dMR image structure is expressed as a kind of minimum-spanning tree. It fades out distortions of the image’s features. Moreover, as contrasted with numerous other super-resolution algorithms, the presented method overcomes the simplifying motion model as well as blur kernel and noise estimation issues. The simulation and experimental studies have been conducted using a dMRI scanner as well as a phantom input. Combining super-resolution with time-efficient data sets resulted in a reduction of motion artifacts, improving edge delineation as well as spatial resolution.     

基于亚像素扫描的超分辨技术在高分辨X射线显微镜中的应用

在不改变现有硬件条件的情况下,开展超分辨扫描重建方法,可以在不增加系统成本的基础上提高高分辨X射线显微镜的成像性能.设计了基于亚像素扫描的超分辨扫描模式,按照设计的调制方式进行亚像素位移的移动,采集多幅具有互补信息的低分辨率图像;然后基于系统的点扩散函数,对高分辨率图像进行复原;最后结合POCS超分辨重建算法重建出高分辨图像.实验结果表明,10倍光耦探测器下的衬度噪声比提高了20%左右,空间分辨力提高了0.2μm(约15%),细节分辨能力超过探测器像素尺寸1.35μm的限制,可以看到在低分辨率图像中看不到的细节.实验说明用超分辨技术提高高分辨X射线显微镜的分辨率是有意义的.     

稀疏阈值的超分辨率图像重建

为了解决基于字典学习的超分辨重构算法耗时过长的问题,提出了基于稀疏阈值模型的图像超分辨率重建方法。首先,将联合字典理论与图像块稀疏阈值方法相结合,训练得到高、低分辨率过完备图像字典对。接着,通过稀疏阈值OMP算法对图像特征块进行稀疏表示。然后,通过高分辨率字典重构出初始的超分辨图像。最后,通过改进迭代反投影算法对初始的超分辨图像进行全局优化,从而进一步提高图像重构质量。实验结果表明,超分辨图像重构平均峰值信噪比(PSNR)为30.1 d B,平均结构自相似度(SSIM)为0.937 9,平均计算时间为10.2 s。有效提高了超分辨重构的速度,改善了重构高分辨图像的质量。     

Deep learning and expert knowledge based underwater acoustic OFDM receiver

In this paper, a deep learning and expert knowledge based receiver is proposed for underwater acoustic (UWA) orthogonal frequency division multiplexing (OFDM). Different from the existing deep learning based UWA OFDM receivers, the proposed receiver combines deep learning with the classical expert knowledge of block-based signal processing in UWA OFDM to improve system performance and interpretability. It performs joint channel estimation and signal detection by designing skip connection (SC) convolutional neural network (CNN) cascaded attention mechanism (AM) enhanced bi-directional long short-term memory (BiLSTM) network, abbreviated as SC-CNN-AM-BiLSTM network (SCABNet). Specifically, the channel estimation subnet is designed with SC-CNN to utilize the thought of image super-resolution to reconstruct the entire channel frequency response of all subcarriers. The signal detection subnet is designed with AM-BiLSTM to extract the correlations of received sequential data for signal detection. Especially with the AM, the signal detection subnet can focus more on effective information of the received distorted signal to train the optimal network weights to improve the accuracy of data recovery. The proposed SCABNet is evaluated by experimental data, and the results have demonstrated that the SCABNet has the lowest BER and robust performance compared to the traditional linear algorithm, deep learning based black-box receiver, and ComNet receiver. And the proposed SCABNet is effective and robust when multiple nonideal factors co-exist.     

Single-image super-resolution with joint-optimization of TV regularization and sparse representation

A super-resolution (SR) reconstruction framework is proposed using regularization restoration combined with learning-based resolution enhancement via sparse representation. With the viewpoint of conventional learning methods, the original image can be split into low frequency (LF) and high frequency (HF) components. The reconstruction mainly focuses on the process of HF part, while the LF one is founded simply by typical interpolation function. For the severely blurred single-image, we first use regularization restoration technology to recover it. Then the regularized output remarkably betters the quality of LF used in traditional learning-based methods. Last, image resolution enhancement with characteristic of edge preserving can implement based on the acquired relatively sharp intermediate image and the pre-constructed over-complete dictionary for sparse representation. Specifically, the regularization can favorably weaken the dependence of atoms on the course of degradation. With both techniques, we can noticeably eliminate the blur and the edge artifacts in the enlarged image simultaneously. Various experimental results demonstrate that the proposed approach can produce visually pleasing resolution for severely blurred image.     

An adaptive super-resolution method based on regional pixel information and ringing artifacts suppression

Multi-frame image super-resolution (SR) aims to utilize information from a set of low-resolution (LR) images to compose a high-resolution (HR) one. In this paper, a novel multi-frame image super-resolution algorithm is proposed based on regional pixel information and ringing artifacts suppression. Firstly, a new regularization term which adopts Regional Adaptive Weight Coefficients (RAWC) is produced to keep edges and flat regions. After detailed analysis, an iterative process is given for image reconstruction. Then an adaptive term according to the local variance of iterative correction image is designed to evaluate the ringing artifacts. Finally, the original iteration is updated by adding the restraint term for better visual effects and lower noise of reconstructive HR image. Thorough experimental results show the proposed algorithm is effective for SR reconstruction and ringing artifacts suppression.     

基于MAP的高光谱图像超分辨率方法

高光谱图像得到了越来越广泛的应用,但较低的空间分辨率严重地影响着它的应用效果;其超分辨率方法受到学术界的高度重视,但一直没有得到很好的解决。为此重点研究了建立低分辨率资源图像与高分辨率目标图像之间的关系模型;引入关联感兴趣光谱端元的算子进行空间变换;应用最大后验概率(MAP)算法实现超分辨率复原。实验表明,该超分辨率方法具有超分辨率效果好、复杂度低、抗噪声性能强和保护感兴趣类别等优点。     

Magnetic resonance image enhancement using highly sparse input

Lately, the Magnetic Resonance scans have struggled with its own inherent limitations, such as spatial resolution as well as long examination times. In this paper, a novel, rapid compressively-sensed magnetic resonance high resolution image resolution algorithm is presented. This technique addresses these two key issues by employing a highly-sparse sampling scheme and super-resolution reconstruction (SRR) method. Due to highly challenging requirements for the accuracy of diagnostic images registration, the presented technique exploits image priors, deblurring, parallel imaging, and a discrete dense displacement sampling for the deformable human body and motion analysis. The clinical trials as well as phantom based studied have been conducted. It has been proven that the proposed algorithm is able to enhance image spatial resolution, reduce motion artefacts and scan times.     

多局部残差连接注意网络的图像去模糊

针对现有的基于卷积神经网络的图像去模糊算法存在图像纹理细节恢复不清晰的问题,提出了一种基于多局部残差连接注意网络的图像去模糊算法。首先,采用一个卷积层进行浅层特征提取;其次,设计了一种新的基于残差连接和并行注意机制的多局部残差连接注意模块,用于消除图像模糊并提取上下文信息;再次,采用一个基于扩张卷积的成对连接模块进行细节恢复;最后,利用一个卷积层重建清晰图像。实验结果表明：在GoPro数据集上的PSNR (peak signal to noise ratio)和SSIM (structure similarity)分别为31.83 dB、0.927 5,在定性和定量两方面都表明所提方法能够有效地恢复模糊图像的纹理细节,网络性能优于对比方法。     

Triple-D network for efficient undersampled magnetic resonance images reconstruction

Compressed sensing (CS) theory can help accelerate magnetic resonance imaging (MRI) by sampling partial k-space measurements. However, conventional optimization-based CS-MRI methods are often time-consuming and are based on fixed transform or shallow image dictionaries, which limits modeling capabilities. Recently, deep learning models have been used to solve the CS-MRI problem. However, recent researches have focused on modeling in image domain, and the potential of k-space modeling capability has not been utilized seriously. In this paper, we propose a deep model called Dual Domain Dense network (Triple-D network), which consisted of some k-space and image domain sub-network. These sub-networks are connected with dense connections, which can utilize feature maps at different levels to enhance performance. To further promote model capabilities, we use two strategies: multi-supervision strategies, which can avoid loss of supervision information; channel-wise attention layer (CA layer), which can adaptively adjust the weight of the feature map. Experimental results show that the proposed Triple-D network provides promising performance in CS-MRI, and it can effectively work on different sampling trajectories and noisy settings.     

基于多任务卷积神经网络的红外与可见光多分辨率图像融合

红外与可见光图像融合一直是图像领域研究的热点,融合技术能弥补单一传感器的不足,为图像理解与分析提供良好的成像基础。因生产工艺以及成本的限制,红外探测器的分辨率远低于可见光探测器,并在一定程度上因源图像分辨率的差异阻碍了实际应用。针对红外与可见光图像分辨率不一致的问题,提出了用于红外图像超分辨率重建与融合的多任务卷积网络框架,应用于多分辨率图像融合。在网络结构方面,首先设计了双通道网络分别提取红外与可见光特征,使算法不受源图像分辨率的限制;其次提出了特征上采样模块,先用双线性插值方法增加像素个数,再通过多层感知器精细化拟合像素平滑空间与高频空间的映射关系,无需重新训练模型即可实现任意尺度的红外图像上采样;接着将线性注意力引入网络,学习特征空间位置间的非线性关系,抑制无关信息并增强网络对全局信息的表达。在损失函数方面,提出了梯度损失,保留红外与可见光图像中绝对值较大的滤波器响应值,并计算该值与重建的融合图像响应值的Frobenius范数,无需理想的融合图像作为真值监督网络学习就能生成融合图像;此外,在梯度损失、像素损失的共同作用下对多任务模型进行优化,可以同时重建融合图像和高分辨率红外图像...