目的图像和视频合成技术在媒体后期处理领域广泛应用,随着技术门槛的降低,大量合成素材被发布并迅速传播。然而,部分合成内容可能含有误导性信息,威胁视听内容的真实性和安全性。传统合成检测方法主要依赖合成痕迹或画面异常检测,但随...目的图像和视频合成技术在媒体后期处理领域广泛应用,随着技术门槛的降低,大量合成素材被发布并迅速传播。然而,部分合成内容可能含有误导性信息,威胁视听内容的真实性和安全性。传统合成检测方法主要依赖合成痕迹或画面异常检测,但随着合成技术的不断进步,现有方法在检测精度和适应性方面仍存在优化空间,需要改进以应对日益复杂的合成内容检测需求。方法本文提出一种融合物理与深度学习的合成图像检测方法,创新性地结合光照和阴影一致性分析。通过特征提取与融合网络,实现光照图与光照强度的一致性分析,判断物体采集环境;利用交比估计检测光照方向一致性,有效提升了检测精度和适应性。同时构建了具有物理属性的数据集,为合成图像检测提供数据支持。结果在NIST 16(National Institute of Standards and Technology Database 16)、Coverage和CASIA(Chinese Academy of Sciences Institute of Automation Database)数据集上的实验表明,本文方法在AUC(area under the curve)指标上分别达到94.2%、93.6%和90.3%,F1分数分别达到80.2%、79.3%和58.1%,优于对比方法。在噪声攻击实验中,本文方法对尺寸变化、高斯模糊、高斯噪声和JPEG(Joint Photographic Experts Group)压缩表现出更强的适应性,平均AUC为84.03%。此外,本文提出的数据集在训练过程中表现出高可用性,使用该数据集训练的模型AUC平均提升18.1%。结论本文方法在准确性和鲁棒性方面均优于对比方法,构建的数据集能够有效支持合成图像检测模型的训练、验证和测试,为该领域的研究提供了重要参考。数据集下载链接:https://doi.org/10.57760/sciencedb.j00240.00069.展开更多
Before joining the Beijing Dragon Shadow Puppetry Art Theatre,this group of young people rarely had the chance to meet people like themselves,lacked friends,faced obstacles at work,and felt alienated from society.Now,...Before joining the Beijing Dragon Shadow Puppetry Art Theatre,this group of young people rarely had the chance to meet people like themselves,lacked friends,faced obstacles at work,and felt alienated from society.Now,they have gained appreciation,recognition,confidence,friendships,and love.展开更多
Shadow is one of the major problems in remotely sensed imagery which hampers the accuracy of information extraction and change detection.In these images,shadow is generally produced by different objects,namely,cloud,m...Shadow is one of the major problems in remotely sensed imagery which hampers the accuracy of information extraction and change detection.In these images,shadow is generally produced by different objects,namely,cloud,mountain and urban materials.The shadow correction process consists of two steps:detection and de-shadowing.This paper reviews a range of techniques for both steps,focusing on urban regions(urban shadows),mountainous areas(topographic shadow),cloud shadows and composite shadows.Several issues including the problems and the advantages of those algorithms are discussed.In recent years,thresholding and recovery techniques have become important for shadow detection and de-shadowing,respectively.Research on shadow correction is still an important topic,particularly for urban regions(in high spatial resolution data) and mountainous forest(in high and medium spatial resolution data).Moreover,new algorithms are needed for shadow correction,especially given the advent of new satellite images.展开更多
In this paper we study shadowing property for sequences of mappings on compact metric spaces, i.e., nonautonomous discrete dynamical systems. We investi- gate the relations of various expansivity properties with shado...In this paper we study shadowing property for sequences of mappings on compact metric spaces, i.e., nonautonomous discrete dynamical systems. We investi- gate the relations of various expansivity properties with shadowing and h-shadowing property.展开更多
文摘目的图像和视频合成技术在媒体后期处理领域广泛应用,随着技术门槛的降低,大量合成素材被发布并迅速传播。然而,部分合成内容可能含有误导性信息,威胁视听内容的真实性和安全性。传统合成检测方法主要依赖合成痕迹或画面异常检测,但随着合成技术的不断进步,现有方法在检测精度和适应性方面仍存在优化空间,需要改进以应对日益复杂的合成内容检测需求。方法本文提出一种融合物理与深度学习的合成图像检测方法,创新性地结合光照和阴影一致性分析。通过特征提取与融合网络,实现光照图与光照强度的一致性分析,判断物体采集环境;利用交比估计检测光照方向一致性,有效提升了检测精度和适应性。同时构建了具有物理属性的数据集,为合成图像检测提供数据支持。结果在NIST 16(National Institute of Standards and Technology Database 16)、Coverage和CASIA(Chinese Academy of Sciences Institute of Automation Database)数据集上的实验表明,本文方法在AUC(area under the curve)指标上分别达到94.2%、93.6%和90.3%,F1分数分别达到80.2%、79.3%和58.1%,优于对比方法。在噪声攻击实验中,本文方法对尺寸变化、高斯模糊、高斯噪声和JPEG(Joint Photographic Experts Group)压缩表现出更强的适应性,平均AUC为84.03%。此外,本文提出的数据集在训练过程中表现出高可用性,使用该数据集训练的模型AUC平均提升18.1%。结论本文方法在准确性和鲁棒性方面均优于对比方法,构建的数据集能够有效支持合成图像检测模型的训练、验证和测试,为该领域的研究提供了重要参考。数据集下载链接:https://doi.org/10.57760/sciencedb.j00240.00069.
文摘Before joining the Beijing Dragon Shadow Puppetry Art Theatre,this group of young people rarely had the chance to meet people like themselves,lacked friends,faced obstacles at work,and felt alienated from society.Now,they have gained appreciation,recognition,confidence,friendships,and love.
基金Under the auspices of National Technology Research and Development Program of China(No.2006BAJ05A02)National Natural Science Foundation of China(No.31172023)
文摘Shadow is one of the major problems in remotely sensed imagery which hampers the accuracy of information extraction and change detection.In these images,shadow is generally produced by different objects,namely,cloud,mountain and urban materials.The shadow correction process consists of two steps:detection and de-shadowing.This paper reviews a range of techniques for both steps,focusing on urban regions(urban shadows),mountainous areas(topographic shadow),cloud shadows and composite shadows.Several issues including the problems and the advantages of those algorithms are discussed.In recent years,thresholding and recovery techniques have become important for shadow detection and de-shadowing,respectively.Research on shadow correction is still an important topic,particularly for urban regions(in high spatial resolution data) and mountainous forest(in high and medium spatial resolution data).Moreover,new algorithms are needed for shadow correction,especially given the advent of new satellite images.
文摘In this paper we study shadowing property for sequences of mappings on compact metric spaces, i.e., nonautonomous discrete dynamical systems. We investi- gate the relations of various expansivity properties with shadowing and h-shadowing property.
