共查询到20条相似文献，搜索用时 0 毫秒
1.
RUI Zhichao;GUO Yanyan 《测试技术学报》2024,38(1):34-40
For an image of large irregular missing regions, recent deep learning-based image inpainting approaches often generate content with blurred textures and distorted structures. To solve these problems, a three-stage model that separates the inpainting problem into contextual feature-based structure prediction and image completion is proposed. In the first stage, our model utilizes surrounding image features to predict missing regions during a dilated convolutional encoder-decoder network training. In the second stage, an encoder-decoder network based on the self-attention mechanism takes edge features extracted from the first stage predictions as inputs and recovers the texture structure of the missing regions. In the third stage, the outputs of the first two stages are passed to a refined inpainting network using the improved U-net architecture to guide the repair process. The proposed algorithm is compared with the existing classic algorithm on the public datasets. Experiments show that our method outperforms existing methods in terms of subjective vision and objective evaluation. 相似文献
2.
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. 相似文献
3.
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. 相似文献
4.
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. 相似文献
5.
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。 相似文献
6.
目的为了有效解决文物图像的不易保存和物理方法修复困难等问题,提出一种基于生成对抗网络的图像修复算法。方法文中算法主要分为2个阶段,第1阶段通过Canny边缘检测器提取图像已知部分的边缘信息,利用1个生成器和1个鉴别器修复图像缺失边缘。第2阶段将第1阶段生成的边缘作为先验信息,通过1个生成器和2个鉴别器修复图像缺失部分。2个鉴别器由整体鉴别器和局部鉴别器组成,整体鉴别器用来评估修复后的图像整体连贯性,局部鉴别器用来查看待修复区域为中心的小区域局部一致性。结果与传统算法对比,文中算法在提高生成图片纹理质量的基础上保证了全局语义结构一致性,在客观指标(峰值信噪比和结构相似性)上,较其他方法有更好的效果。结论文中算法可以有效修复文物图像的缺损部分,尤其是结构复杂的大范围缺失,取得了良好的视觉效果,表明该算法有良好的修复性能。 相似文献
7.
目的 以创新汽车产品辅助设计方法研究为主线,对研究进展、相关理论框架、关键技术进行概述.方法 结合文献梳理法和产业调研法分析人工智能在汽车工业设计中的重要意义、发展趋势及应用进展.结果 提出了基于图库的辅助设计、基于语义的辅助设计、造型设计评价三种应用场景的关键技术框架.结论 人工智能在汽车创新设计中的应用不充分,针对研究空缺提出了三种亟待突破的关键技术框架,为汽车企业提供了思路和借鉴. 相似文献
8.
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。 相似文献
9.
Deep Learning (DL) is such a powerful tool that we have seen tremendous success in areas such as Computer Vision, Speech Recognition, and Natural Language Pro-cessing. Since Automated Modulation Classification (AMC) is an important part in Cognitive Radio Networks, we try to explore its potential in solving signal modula-tion recognition problem. It cannot be overlooked that DL model is a complex mod-el, thus making them prone to over-fitting. DL model requires many training data to combat with over-fitting, but adding high quality labels to training data manually is not always cheap and accessible, especially in real-time system, which may counter unprecedented data in dataset. Semi-supervised Learning is a way to exploit unla-beled data effectively to reduce over-fitting in DL. In this paper, we extend Genera-tive Adversarial Networks (GANs) to the semi-supervised learning will show it is a method can be used to create a more data-efficient classifier. 相似文献
10.
目的 为减少ZB47包装机小包拉线缺陷投诉,基于无监督深度神经网络构建一种小包拉线缺陷视觉智能检测方法。方法 首先,在ZB47包装机CH转塔部位设计并加装小包图像采集装置,获得实时高清晰度小包图像。其次,将小包图像根据拉线位置进行固定位置的裁剪,从而减轻不同工况的环境背景影响并且加快检测速度。然后,构建自编码器–编码器结构的主干网络,同时叠加生成对抗网络中的判别器模块组成缺陷判别模型,并综合采用图像间、图像隐空间以及图像特征间的信息构建模型的损失函数。最后,使用裁剪后的正常小包拉线图像对构建的缺陷判别模型进行训练,并基于所有的正常小包图像得到异常阈值。结果 实际验证阶段,待检测图像的得分大于异常阈值即判断为异常图像,触发CH转塔部位的小包剔除装置将该缺陷小包剔除。生产现场测试表明,所提方法可以对典型小包缺陷进行快速准确检测,缺陷检测准确率为99.99%。结论 该方法能够满足生产现场卷烟小包拉线缺陷检测的准确性和实时性要求。 相似文献
11.
Recently, many researchers have concentrated on using neural networks to learn features for Distant Supervised Relation Extraction (DSRE). These approaches generally use a softmax classifier with cross-entropy loss, which inevitably brings the noise of artificial class NA into classification process. To address the shortcoming, the classifier with ranking loss is employed to DSRE. Uniformly randomly selecting a relation or heuristically selecting the highest score among all incorrect relations are two common methods for generating a negative class in the ranking loss function. However, the majority of the generated negative class can be easily discriminated from positive class and will contribute little towards the training. Inspired by Generative Adversarial Networks (GANs), we use a neural network as the negative class generator to assist the training of our desired model, which acts as the discriminator in GANs. Through the alternating optimization of generator and discriminator, the generator is learning to produce more and more discriminable negative classes and the discriminator has to become better as well. This framework is independent of the concrete form of generator and discriminator. In this paper, we use a two layers fully-connected neural network as the generator and the Piecewise Convolutional Neural Networks (PCNNs) as the discriminator. Experiment results show that our proposed GAN-based method is effective and performs better than state-of-the-art methods. 相似文献
12.
在艺科融合的新文科建设背景及人机耦合的设计趋势下,以算法、人工智能作为创造性技术工具的生成式字体设计,在语义与美学层面重新诠释了互动性中形式与语境的相互渗透。旨围绕算法字体及生成式对抗网络驱动的人工智能字体展开形式分析、设计方法与技术路径的探讨。主要论述基于数学定理、计算几何的算法字体,利用数据和参数来编码形状语法与视觉系统;基于生成式对抗网络GAN的深度学习,驱动人工智能进行字体生成与风格迁移。在字体设计实践中,允许人工智能探索人类创造力的协作维度,这种方法是否能作为一种创新策略嵌入到字体设计流程中,保持系统的开放性,赋予技术系统与人类主体之间不断发展的对话。同时,使字体设计突破其工具性使命,将设计挑战作为进化和创新的契机,为互动系统及其社会技术环境的设计提供了一种新方法。 相似文献
13.
Fatma Harby Adel Thaljaoui Durre Nayab Suliman Aladhadh Salim EL Khediri Rehan Ullah Khan 《计算机、材料和连续体（英文）》2023,74(3):5191-5207
In the machine learning (ML) paradigm, data augmentation servesas a regularization approach for creating ML models. The increase in thediversification of training samples increases the generalization capabilities,which enhances the prediction performance of classifiers when tested onunseen examples. Deep learning (DL) models have a lot of parameters, andthey frequently overfit. Effectively, to avoid overfitting, data plays a majorrole to augment the latest improvements in DL. Nevertheless, reliable datacollection is a major limiting factor. Frequently, this problem is undertakenby combining augmentation of data, transfer learning, dropout, and methodsof normalization in batches. In this paper, we introduce the application of dataaugmentation in the field of image classification using Random Multi-modelDeep Learning (RMDL) which uses the association approaches of multiDL to yield random models for classification. We present a methodologyfor using Generative Adversarial Networks (GANs) to generate images fordata augmenting. Through experiments, we discover that samples generatedby GANs when fed into RMDL improve both accuracy and model efficiency.Experimenting across both MNIST and CIAFAR-10 datasets show that,error rate with proposed approach has been decreased with different randommodels. 相似文献
14.
Andrew Kusiak 《国际生产研究杂志》2020,58(5):1594-1604
Manufacturing is undergoing transformation driven by the developments in process technology, information technology, and data science. A future manufacturing enterprise will be highly digital. This will create opportunities for machine learning algorithms to generate predictive models across the enterprise in the spirit of the digital twin concept. Convolutional and generative adversarial neural networks have received some attention of the manufacturing research community. Representative research and applications of the two machine learning concepts in manufacturing are presented. Advantages and limitations of each neural network are discussed. The paper might be helpful in identifying research gaps, inspire machine learning research in new manufacturing domains, contribute to the development of successful neural network architectures, and getting deeper insights into the manufacturing data. 相似文献
15.
《Journal of Modern Optics》2012,59(19):1880-1888
ABSTRACTImage completion is an approach to repair a damaged region (a hole) in an image. In this study, a novel image completion algorithm is proposed using texture direction to guide the completion process. We first analyse and detect the dominant texture direction using a gray-level co-occurrence matrix (GLCM). Based on the detected direction, we then generate a direction map that serves as a soft constraint during the completion process. Further, we propose an objective function that can balance the global texture structure and local coherency in order to maintain the structure consistency. Experimental results show that the proposed method has potential performance in maintaining the structure continuity and texture smoothness. 相似文献
16.
目的探究传统文化元素的生成方法,将之运用至文创产品的设计实践中,在传播中华优良传统文化方面进行尝试。通过技术方法上的创新和运用解决非物质文化遗产传承人不足,文化元素陈旧匮乏等现实问题。方法以瑞昌竹编的传统文化元素为例,综合文献研究、田野调查、包括对瑞昌竹编国家级非遗传承人的深度访谈,将采集的竹编传统文化元素导入模型系统,利用生成式对抗网络技术训练计算机GPU,使其衍生出新的文化元素,建立瑞昌竹编文化元素库。结论利用生成式对抗网络模型迭代计算出的瑞昌竹编传统文化元素既具有传统纹样的艺术性又兼具信息时代对设计衍生的时效性要求,同时还符合当代社会对时尚偏好的追求。将其运用于现代文创产品的设计当中,使文创产品成为传播我国优良传统文化元素的有效载体。 相似文献
17.
目的 为了更好地实现包装设计中符合用户美学感知需求的动漫图像元素,生成美感更高的彩色线稿图像,提出一种基于美学质量评价的条件生成对抗网络线稿图像彩色化方法(Conditional Image Colorization with Image Aesthetics GANs,IM-GAN).方法 采用Mish函数作为生成模型激活函数,并使用美学质量评价优化生成模型损失函数,实现线稿图像的自动彩色化任务.结果 使用图像美学质量评分作为客观评价指标,观察者打分作为主观评价指标,对算法进行评价.IM-GAN生成的彩色图像具有更高美学质量评分和主观打分.结论 文中方法能够完成线稿图像自动彩色化任务,在包装设计动漫形象应用方面具有一定的参考和使用价值. 相似文献
18.
目的 针对包装产品上QR码在采集过程中的运动模糊、失焦模糊,长期磨损形成的自模糊和环境中的噪声等因素,导致QR码无法识别的问题,提出一种基于生成对抗网络的QR码去模糊算法。方法 采用深度学习模型生成对抗网络对模糊核和环境噪声具有的强大拟合和估计能力,提取模糊QR码图像与真实图像的深层特征和差距,并通过生成器与判别器不断迭代对抗,使生成器具有由输入的模糊QR码产生与之对应的去模糊QR码图像的能力。结果 生成器能较好地对模糊核和环境噪声进行估计,而且能够实现对数据集内多种不同模糊程度QR码的去模糊,去模糊QR码图像效果较好,处理时间快,识别率较高。结论 采用基于生成对抗网络的QR码去模糊算法能够广泛应用于包装产品外壳上QR码的预处理过程,泛化能力较好,能有效提高扫描识别率。 相似文献
19.
目的 针对剪纸智能生成中剪纸的地域性和风格性问题,探索一种能够生成特定风格剪纸的智能生成方法.方法 以侗族剪纸为研究对象,对其进行纹样分析,得到其中应用比较广泛的花草纹样.然后,根据花草纹构造特征进行分解,重构出数量较多的保留侗族剪纸风格的图案,同时通过数据增广的方式获得侗族剪纸花草纹数据集.基于该数据集,提出使用生成对抗网络来拟合花草纹数据集的数据分布,生成侗族剪纸风格的纹样.结果 生成的纹样图像质量较高,线条流畅、优美,符合侗族剪纸风格.结论 通过生成对抗网络学习侗族的剪纸风格,取得了较好效果,为特定风格剪纸的智能生成和非物质文化遗产传承提供了新的思路. 相似文献
20.
The increasing penetration rate of electric kickboard vehicles has been popularized and promoted primarily because of its clean and efficient features. Electric kickboards are gradually growing in popularity in tourist and education-centric localities. In the upcoming arrival of electric kickboard vehicles, deploying a customer rental service is essential. Due to its free-floating nature, the shared electric kickboard is a common and practical means of transportation. Relocation plans for shared electric kickboards are required to increase the quality of service, and forecasting demand for their use in a specific region is crucial. Predicting demand accurately with small data is troublesome. Extensive data is necessary for training machine learning algorithms for effective prediction. Data generation is a method for expanding the amount of data that will be further accessible for training. In this work, we proposed a model that takes time-series customers’ electric kickboard demand data as input, pre-processes it, and generates synthetic data according to the original data distribution using generative adversarial networks (GAN). The electric kickboard mobility demand prediction error was reduced when we combined synthetic data with the original data. We proposed Tabular-GAN-Modified-WGAN-GP for generating synthetic data for better prediction results. We modified The Wasserstein GAN-gradient penalty (GP) with the RMSprop optimizer and then employed Spectral Normalization (SN) to improve training stability and faster convergence. Finally, we applied a regression-based blending ensemble technique that can help us to improve performance of demand prediction. We used various evaluation criteria and visual representations to compare our proposed model’s performance. Synthetic data generated by our suggested GAN model is also evaluated. The TGAN-Modified-WGAN-GP model mitigates the overfitting and mode collapse problem, and it also converges faster than previous GAN models for synthetic data creation. The presented model’s performance is compared to existing ensemble and baseline models. The experimental findings imply that combining synthetic and actual data can significantly reduce prediction error rates in the mean absolute percentage error (MAPE) of 4.476 and increase prediction accuracy. 相似文献