Ensuring digital media security through robust image watermarking is essential to prevent unauthorized distribution,tampering,and copyright infringement.This study introduces a novel hybrid watermarking framework that...Ensuring digital media security through robust image watermarking is essential to prevent unauthorized distribution,tampering,and copyright infringement.This study introduces a novel hybrid watermarking framework that integrates Discrete Wavelet Transform(DWT),Redundant Discrete Wavelet Transform(RDWT),and Möbius Transformations(MT),with optimization of transformation parameters achieved via a Genetic Algorithm(GA).By combining frequency and spatial domain techniques,the proposed method significantly enhances both the imper-ceptibility and robustness of watermark embedding.The approach leverages DWT and RDWT for multi-resolution decomposition,enabling watermark insertion in frequency subbands that balance visibility and resistance to attacks.RDWT,in particular,offers shift-invariance,which improves performance under geometric transformations.Möbius transformations are employed for spatial manipulation,providing conformal mapping and spatial dispersion that fortify watermark resilience against rotation,scaling,and translation.The GA dynamically optimizes the Möbius parameters,selecting configurations that maximize robustness metrics such as Peak Signal-to-Noise Ratio(PSNR),Structural Similarity Index Measure(SSIM),Bit Error Rate(BER),and Normalized Cross-Correlation(NCC).Extensive experiments conducted on medical and standard benchmark images demonstrate the efficacy of the proposed RDWT-MT scheme.Results show that PSNR exceeds 68 dB,SSIM approaches 1.0,and BER remains at 0.0000,indicating excellent imperceptibility and perfect watermark recovery.Moreover,the method exhibits exceptional resilience to a wide range of image processing attacks,including Gaussian noise,JPEG compression,histogram equalization,and cropping,achieving NCC values close to or equal to 1.0.Comparative evaluations with state-of-the-art watermarking techniques highlight the superiority of the proposed method in terms of robustness,fidelity,and computational efficiency.The hybrid framework ensures secure,adaptive watermark embedding,making it highly suitable for applications in digital rights management,content authentication,and medical image protection.The integration of spatial and frequency domain features with evolutionary optimization presents a promising direction for future watermarking technologies.展开更多
针对现有深度学习算法在壁画修复时,存在全局语义一致性约束不足及局部特征提取不充分,导致修复后的壁画易出现边界效应和细节模糊等问题,提出一种双向自回归Transformer与快速傅里叶卷积增强的壁画修复方法.首先,设计基于Transformer...针对现有深度学习算法在壁画修复时,存在全局语义一致性约束不足及局部特征提取不充分,导致修复后的壁画易出现边界效应和细节模糊等问题,提出一种双向自回归Transformer与快速傅里叶卷积增强的壁画修复方法.首先,设计基于Transformer结构的全局语义特征修复模块,利用双向自回归机制与掩码语言模型(masked language modeling,MLM),提出改进的多头注意力全局语义壁画修复模块,提高对全局语义特征的修复能力.然后,构建了由门控卷积和残差模块组成的全局语义增强模块,增强全局语义特征一致性约束.最后,设计局部细节修复模块,采用大核注意力机制(large kernel attention,LKA)与快速傅里叶卷积提高细节特征的捕获能力,同时减少局部细节信息的丢失,提升修复壁画局部和整体特征的一致性.通过对敦煌壁画数字化修复实验,结果表明,所提算法修复性能更优,客观评价指标均优于比较算法.展开更多
现有的基于卷积神经网络的超分辨率重建方法由于感受野限制,难以充分利用遥感图像丰富的上下文信息和自相关性,导致重建效果不佳.针对该问题,本文提出了一种基于多重蒸馏与Transformer的遥感图像超分辨率(remote sensing image super-re...现有的基于卷积神经网络的超分辨率重建方法由于感受野限制,难以充分利用遥感图像丰富的上下文信息和自相关性,导致重建效果不佳.针对该问题,本文提出了一种基于多重蒸馏与Transformer的遥感图像超分辨率(remote sensing image super-resolution based on multi-distillation and Transformer,MDT)重建方法.首先结合多重蒸馏和双注意力机制,逐步提取低分辨率图像中的多尺度特征,以减少特征丢失.接着,构建一种卷积调制Transformer来提取图像的全局信息,恢复更多复杂的纹理细节,从而提升重建图像的视觉效果.最后,在上采样过程中添加全局残差路径,提高特征在网络中的传播效率,有效减少了图像的失真与伪影问题.在AID和UCMerced两个数据集上的进行实验,结果表明,本文方法在放大至4倍超分辨率任务上的峰值信噪比和结构相似度分别最高达到了29.10 dB和0.7807,重建图像质量明显提高,并且在细节保留方面达到了更好的视觉效果.展开更多
文摘Ensuring digital media security through robust image watermarking is essential to prevent unauthorized distribution,tampering,and copyright infringement.This study introduces a novel hybrid watermarking framework that integrates Discrete Wavelet Transform(DWT),Redundant Discrete Wavelet Transform(RDWT),and Möbius Transformations(MT),with optimization of transformation parameters achieved via a Genetic Algorithm(GA).By combining frequency and spatial domain techniques,the proposed method significantly enhances both the imper-ceptibility and robustness of watermark embedding.The approach leverages DWT and RDWT for multi-resolution decomposition,enabling watermark insertion in frequency subbands that balance visibility and resistance to attacks.RDWT,in particular,offers shift-invariance,which improves performance under geometric transformations.Möbius transformations are employed for spatial manipulation,providing conformal mapping and spatial dispersion that fortify watermark resilience against rotation,scaling,and translation.The GA dynamically optimizes the Möbius parameters,selecting configurations that maximize robustness metrics such as Peak Signal-to-Noise Ratio(PSNR),Structural Similarity Index Measure(SSIM),Bit Error Rate(BER),and Normalized Cross-Correlation(NCC).Extensive experiments conducted on medical and standard benchmark images demonstrate the efficacy of the proposed RDWT-MT scheme.Results show that PSNR exceeds 68 dB,SSIM approaches 1.0,and BER remains at 0.0000,indicating excellent imperceptibility and perfect watermark recovery.Moreover,the method exhibits exceptional resilience to a wide range of image processing attacks,including Gaussian noise,JPEG compression,histogram equalization,and cropping,achieving NCC values close to or equal to 1.0.Comparative evaluations with state-of-the-art watermarking techniques highlight the superiority of the proposed method in terms of robustness,fidelity,and computational efficiency.The hybrid framework ensures secure,adaptive watermark embedding,making it highly suitable for applications in digital rights management,content authentication,and medical image protection.The integration of spatial and frequency domain features with evolutionary optimization presents a promising direction for future watermarking technologies.
文摘针对现有深度学习算法在壁画修复时,存在全局语义一致性约束不足及局部特征提取不充分,导致修复后的壁画易出现边界效应和细节模糊等问题,提出一种双向自回归Transformer与快速傅里叶卷积增强的壁画修复方法.首先,设计基于Transformer结构的全局语义特征修复模块,利用双向自回归机制与掩码语言模型(masked language modeling,MLM),提出改进的多头注意力全局语义壁画修复模块,提高对全局语义特征的修复能力.然后,构建了由门控卷积和残差模块组成的全局语义增强模块,增强全局语义特征一致性约束.最后,设计局部细节修复模块,采用大核注意力机制(large kernel attention,LKA)与快速傅里叶卷积提高细节特征的捕获能力,同时减少局部细节信息的丢失,提升修复壁画局部和整体特征的一致性.通过对敦煌壁画数字化修复实验,结果表明,所提算法修复性能更优,客观评价指标均优于比较算法.
文摘现有的基于卷积神经网络的超分辨率重建方法由于感受野限制,难以充分利用遥感图像丰富的上下文信息和自相关性,导致重建效果不佳.针对该问题,本文提出了一种基于多重蒸馏与Transformer的遥感图像超分辨率(remote sensing image super-resolution based on multi-distillation and Transformer,MDT)重建方法.首先结合多重蒸馏和双注意力机制,逐步提取低分辨率图像中的多尺度特征,以减少特征丢失.接着,构建一种卷积调制Transformer来提取图像的全局信息,恢复更多复杂的纹理细节,从而提升重建图像的视觉效果.最后,在上采样过程中添加全局残差路径,提高特征在网络中的传播效率,有效减少了图像的失真与伪影问题.在AID和UCMerced两个数据集上的进行实验,结果表明,本文方法在放大至4倍超分辨率任务上的峰值信噪比和结构相似度分别最高达到了29.10 dB和0.7807,重建图像质量明显提高,并且在细节保留方面达到了更好的视觉效果.
