Image forgery is a crucial part of the transmission of misinformation,which may be illegal in some jurisdictions.The powerful image editing software has made it nearly impossible to detect altered images with the nake...Image forgery is a crucial part of the transmission of misinformation,which may be illegal in some jurisdictions.The powerful image editing software has made it nearly impossible to detect altered images with the naked eye.Images must be protected against attempts to manipulate them.Image authentication methods have gained popularity because of their use in multimedia and multimedia networking applications.Attempts were made to address the consequences of image forgeries by creating algorithms for identifying altered images.Because image tampering detection targets processing techniques such as object removal or addition,identifying altered images remains a major challenge in research.In this study,a novel image texture feature extraction model based on the generalized k-symbolWhittaker function(GKSWF)is proposed for better image forgery detection.The proposed method is divided into two stages.The first stage involves feature extraction using the proposed GKSWF model,followed by classification using the“support vector machine”(SVM)to distinguish between authentic and manipulated images.Each extracted feature from an input image is saved in the features database for use in image splicing detection.The proposed GKSWF as a feature extraction model is intended to extract clues of tampering texture details based on the probability of image pixel.When tested on publicly available image dataset“CASIA”v2.0(ChineseAcademy of Sciences,Institute of Automation),the proposed model had a 98.60%accuracy rate on the YCbCr(luminance(Y),chroma blue(Cb)and chroma red(Cr))color spaces in image block size of 8×8 pixels.The proposed image authentication model shows great accuracy with a relatively modest dimension feature size,supporting the benefit of utilizing the k-symbol Whittaker function in image authentication algorithms.展开更多
为了提高动作的识别精度与鲁棒性,降低冗余特征,提高算法效率,设计了一种基于最大相关-最小冗余(Max-Correlation and Min-Redundancy,MCMR)的动作识别算法.首先,为了消除噪声影响,减少计算成本,利用符号聚集近似(SAX)技术将连续图像序...为了提高动作的识别精度与鲁棒性,降低冗余特征,提高算法效率,设计了一种基于最大相关-最小冗余(Max-Correlation and Min-Redundancy,MCMR)的动作识别算法.首先,为了消除噪声影响,减少计算成本,利用符号聚集近似(SAX)技术将连续图像序列转换为离散符号;其次,为避免出现时间漂移问题,利用动态时间归整(Dynamic Time Warping,DTW)来计算符号特征的距离,提取符号序列的特征;然后,为了消除冗余的特征,定义了一个特征权重,根据权重对特征进行降序排列,引入最大相关-最小冗余技术消除相关性弱的特征,筛选出具有高相关性和低冗余的特征;最后,为了完成动作识别,根据筛选出的特征,利用k-近邻(K-Nearest Neighbor,KNN)进行分类器学习.结果表明:与当前动作识别算法相比,本文算法能够有效完成动作的识别与理解,具有较高的识别率,有效地降低了冗余特征,提高了算法的效率和鲁棒性.展开更多
文摘Image forgery is a crucial part of the transmission of misinformation,which may be illegal in some jurisdictions.The powerful image editing software has made it nearly impossible to detect altered images with the naked eye.Images must be protected against attempts to manipulate them.Image authentication methods have gained popularity because of their use in multimedia and multimedia networking applications.Attempts were made to address the consequences of image forgeries by creating algorithms for identifying altered images.Because image tampering detection targets processing techniques such as object removal or addition,identifying altered images remains a major challenge in research.In this study,a novel image texture feature extraction model based on the generalized k-symbolWhittaker function(GKSWF)is proposed for better image forgery detection.The proposed method is divided into two stages.The first stage involves feature extraction using the proposed GKSWF model,followed by classification using the“support vector machine”(SVM)to distinguish between authentic and manipulated images.Each extracted feature from an input image is saved in the features database for use in image splicing detection.The proposed GKSWF as a feature extraction model is intended to extract clues of tampering texture details based on the probability of image pixel.When tested on publicly available image dataset“CASIA”v2.0(ChineseAcademy of Sciences,Institute of Automation),the proposed model had a 98.60%accuracy rate on the YCbCr(luminance(Y),chroma blue(Cb)and chroma red(Cr))color spaces in image block size of 8×8 pixels.The proposed image authentication model shows great accuracy with a relatively modest dimension feature size,supporting the benefit of utilizing the k-symbol Whittaker function in image authentication algorithms.
文摘为了提高动作的识别精度与鲁棒性,降低冗余特征,提高算法效率,设计了一种基于最大相关-最小冗余(Max-Correlation and Min-Redundancy,MCMR)的动作识别算法.首先,为了消除噪声影响,减少计算成本,利用符号聚集近似(SAX)技术将连续图像序列转换为离散符号;其次,为避免出现时间漂移问题,利用动态时间归整(Dynamic Time Warping,DTW)来计算符号特征的距离,提取符号序列的特征;然后,为了消除冗余的特征,定义了一个特征权重,根据权重对特征进行降序排列,引入最大相关-最小冗余技术消除相关性弱的特征,筛选出具有高相关性和低冗余的特征;最后,为了完成动作识别,根据筛选出的特征,利用k-近邻(K-Nearest Neighbor,KNN)进行分类器学习.结果表明:与当前动作识别算法相比,本文算法能够有效完成动作的识别与理解,具有较高的识别率,有效地降低了冗余特征,提高了算法的效率和鲁棒性.
