摘要
本文针对羽化操作的篡改方式,提出了一种新的图像盲识别算法,这种算法使用纹理分析方法。该方法的基本思想是首先进行图像分解,根据纹理图像进行轮廓构建。在轮廓构建中先对纹理图像进行类似纹理再提取,去除当中少量的噪声,其中使用了双树复小波变换。利用得到的数据,进行轮廓构建。实验表明,该算法在没有任何数字水印和数字签名的前提下,能提高篡改区域的检测精度,有效的检测大面积光滑的图像。
A novel method is presented for blinding detection of digital forgeries based on employing the eclosion when someone forges the digital image. The basic idea of the presented method in this article is that analyzing the texture of the image. In other words, at the first place, we decompose the image and then reconstruct the profile. In order to remove a small amount of noise in the texture image, we employ the dual tree complex wavelet transform, and then reconstruct the profile. The experimental results show that the algorithm can locate the edge eclosion exactly without any embedded information such as digital watermark and digital signature.
出处
《信号处理》
CSCD
北大核心
2009年第12期1922-1927,共6页
Journal of Signal Processing
基金
国家自然科学基金(批准号:60772040
60772042)资助的课题
关键词
羽化
图像分解
双树复小波变换
软阈值
edge eclosion
image decomposition
dual tree complex wavelet transform
soft tbreshold
