UFC-Net with Fully-Connected Layers and Hadamard Identity Skip Connection for Image Inpainting

Image inpainting is an interesting technique in computer vision and artificial intelligence for plausibly filling in blank areas of an image by referring to their surrounding areas. Although its performance has been improved significantly using diverse convolutional neural network (CNN)-based models, these models have difficulty filling in some erased areas due to the kernel size of the CNN. If the kernel size is too narrow for the blank area, the models cannot consider the entire surrounding area, only partial areas or none at all. This issue leads to typical problems of inpainting, such as pixel reconstruction failure and unintended filling. To alleviate this, in this paper, we propose a novel inpainting model called UFC-net that reinforces two components in U-net. The first component is the latent networks in the middle of U-net to consider the entire surrounding area. The second component is the Hadamard identity skip connection to improve the attention of the inpainting model on the blank areas and reduce computational cost. We performed extensive comparisons with other inpainting models using the Places2 dataset to evaluate the effectiveness of the proposed scheme. We report some of the results.     

基于傅里叶卷积和多特征调制的图像修复网络

目的 解决大面积破损难以修复且修复过程中感受野、特征空间信息利用不足,导致修复后的孔洞区域与背景之间出现结构、纹理、风格不一致的问题。方法 基于傅里叶卷积和多特征调制的修复网络FFC-MFMGAN,傅里叶卷积在网络的浅层便具有较大的感受野,尤其是在宽掩码时能够跳过掩码区域,捕获到有效特征,多特征调制生成网络能够分别利用完整区域的信息和随机样式操纵,增强与未受损区域的语义连贯性,以及大空洞率下修复的多样性。结果 在Place 2数据集上,将文中方法与其他图像修复方法进行了对比实验,经过测试,各类指标均得到明显改善,峰值信噪比提高了1.4%,结构相似性提高了4.5%,平均绝对误差降低了12.6%,基于学习的感知图像块相似性降低了9.1%。结论 FFC-MFMGAN网络能够较好地修复大面积不规则孔洞,同时增强修复图像的全局结构性和清晰度,对实际包装印刷图像的缺陷修复也有一定参考价值。     

Better Visual Image Super-Resolution with Laplacian Pyramid of Generative Adversarial Networks

Although there has been a great breakthrough in the accuracy and speed of super-resolution (SR) reconstruction of a single image by using a convolutional neural network, an important problem remains unresolved: how to restore finer texture details during image super-resolution reconstruction? This paper proposes an Enhanced Laplacian Pyramid Generative Adversarial Network (ELSRGAN), based on the Laplacian pyramid to capture the high-frequency details of the image. By combining Laplacian pyramids and generative adversarial networks, progressive reconstruction of super-resolution images can be made, making model applications more flexible. In order to solve the problem of gradient disappearance, we introduce the Residual-in-Residual Dense Block (RRDB) as the basic network unit. Network capacity benefits more from dense connections, is able to capture more visual features with better reconstruction effects, and removes BN layers to increase calculation speed and reduce calculation complexity. In addition, a loss of content driven by perceived similarity is used instead of content loss driven by spatial similarity, thereby enhancing the visual effect of the superresolution image, making it more consistent with human visual perception. Extensive qualitative and quantitative evaluation of the baseline datasets shows that the proposed algorithm has higher mean-sort-score (MSS) than any state-of-the-art method and has better visual perception.     

Image Super-Resolution Based on Generative Adversarial Networks: A Brief Review

Single image super resolution (SISR) is an important research content in the field of computer vision and image processing. With the rapid development of deep neural networks, different image super-resolution models have emerged. Compared tosome traditional SISR methods, deep learning-based methods can complete the superresolution tasks through a single image. In addition, compared with the SISR methodsusing traditional convolutional neural networks, SISR based on generative adversarial networks (GAN) has achieved the most advanced visual performance. In this review, we first explore the challenges faced by SISR and introduce some common datasets and evaluation metrics. Then, we review the improved network structures and loss functionsof GAN-based perceptual SISR. Subsequently, the advantages and disadvantages of different networks are analyzed by multiple comparative experiments. Finally, we summarize the paper and look forward to the future development trends of GAN-based perceptual SISR.     

基于生成对抗网络的压缩感知图像重构方法

目的为了解决传统压缩感知图像重构方法存在的重构时间长、重构图像质量不高等问题,提出一种基于生成对抗网络的压缩感知图像重构方法。方法基于生成对抗网络思想设计一种由具有稀疏采样功能的鉴别器和具有图像重构功能的生成器组成的深度学习网络模型,利用对抗损失和重构损失2个部分组成的新的损失函数对网络参数进行优化,完成图像压缩重构过程。结果实验表明,文中方法在12.5%的低采样率下重构时间为0.009s,相较于常用的OMP算法、CoSaMP算法、SP算法和IRLS算法,其峰值信噪比(PSNR)提高了10～12 dB。结论文中设计的方法应用于图像重构时重构时间短,在低采样率下仍能获得高质量的重构效果。     

多网络联合的红外与可见光图像融合算法研究

目的针对红外与可见光图像在融合过程中,融合图像失真以及可见光图像信息融合不足的问题,提出一种联合多网络结构的红外与可见光图像融合算法。方法首先采用基于密集残差连接的编码器对输入的红外与可见光图像进行特征提取,然后利用融合策略对得到的特征图进行融合,最后将融合后的特征图送入基于GAN网络的解码器中。结果通过与可见光图像对抗优化训练,使得融合后的图像保留了更多可见光图像的细节、背景信息,增强了图像的视觉效果。结论实验表明,与现有的融合算法相比,该算法达到了更好的实验效果,在主观感知和客观评价上都具有更好的表现力。     

Double Encryption Using Trigonometric Chaotic Map and XOR of an Image

In the most recent decades, a major number of image encryption plans have been proposed. The vast majority of these plans reached a high-security level; however, their moderate speeds because of their complicated processes made them of no use in real-time applications. Inspired by this, we propose another efficient and rapid image encryption plan dependent on the Trigonometric chaotic guide. In contrast to the most of current plans, we utilize this basic map to create just a couple of arbitrary rows and columns. Moreover, to additionally speed up, we raise the processing unit from the pixel level to the row/column level. The security of the new plot is accomplished through a substitution permutation network, where we apply a circular shift of rows and columns to break the solid connection of neighboring pixels. At that point, we join the XOR operation with modulo function to cover the pixels values and forestall any leaking of data. High-security tests and simulation analyses are carried out to exhibit that the scheme is very secure and exceptionally quick for real-time image processing at 80 fps (frames per second).     

A 3-D Image Detection Method of a Surface Opening Crack in Concrete Using Ultrasonic Transducer Arrays

The detection of the size and the location of existing three dimensional cracks in a concrete structure is an important topic in civil engineering. In this paper, a multisource, multireceiver method that considers the travel times diffracted by a crack tip is introduced, to backcalculate a 3-D image of the crack tip of a surface opening crack. The possible location of the crack tip front is on the surface of an ellipsoid, which is constructed by a fixed travel time length measured from the source to the receiver, by letting the source and receiver points be the foci of the corresponding ellipsoid. If the locations of the source and the receiver, together with the associated measured travel time of the diffracted echo between each source-receiver pair are known, the image of the tip can be determined by counting the number of intersections of the ellipsoidal surfaces in an image construction cellular structure. The backcalculated crack tip image, as seen from experimental data, match the dimensions of the real crack very well, demonstrating the capability and accuracy of this newly proposed multisource, multireceiver method for concrete NDE.