【摘要】 为验证图像内容是否被恶意篡改,利用规整化PCET矩具有的旋转缩放不变性,提出了一种可容忍几何操作的图像内容认证方法.该方法利用SUSAN算子的响应值对图像中重要的视觉内容进行加权并计算PCET矩作为内容认证的特征,再用确定性自适应量化方法对特征进行量化并生成认证码,然后以量化索引调制的方式将认证码嵌入到原始图像的PCET幅值中;对图像进行验证时,从待测图像中提取水印,通过比较所提取的水印与待测图像的认证码来判断图像是否被篡改.实验结果表明,该方法对旋转缩放带来的几何失真具有鲁棒性,能抵御压缩、增强、滤波、噪声污染等操作,并对小篡改具有敏感性.更多还原

【Abstract】 In order to find out whether the image content is maliciously tampered,a technique of geometrically-robust image content authentication is proposed based on the rotaion scale invariance of the normlized PCET moments.First,all images are weighted by using the response values of SUSAN operator to enhance the important visual image contents,and the PCET moments of the enhanced image are calculated as the content authentication features.Next,the deterministic adaptive quantization method is used to quantify the features,thus generating authentication codes.Then,these codes are embedded into the PCET amplitude of the original image by means of the quantization index modulation method.In the process of verifying the images,watermark is extracted from the detected image and are compared with the authentication codes of the detected image to prove whether the image is tampered.Experimental results show that the proposed method is robust to the geometric distortion caused by rotation and scale,and can resist operations,such as compression,enhancement,filtering and noise pollution,and that it is sensitive to slight tampering.更多还原