基于块特征和混沌序列的图像篡改检测与恢复

为提高图像篡改检测与恢复性能,提出一种基于块特征与混沌序列的图像认证水印方案。对2×2分块离散余弦变换系数编码产生块特征,嵌入到其他图像块像素的低两位,块嵌入的映射关系通过混沌序列产生;结合奇偶检测法对常规的认证方式进行改进,并通过提取有效块的特征信息实现被篡改区域的自恢复。试验表明,该方案嵌入水印信息量少,安全性高,对常规篡改可进行精确定位,并能较好的实现自恢复,还可以有效地检测出仅图像内容的攻击。

Images tamper detection and recovery based on block features and chaotic sequence

To improve the performance of images tamper detection and recovery, an authentication watermarking scheme based on block features and chaotic sequence was presented. For each block of size 2×2 pixels,discrete cosine transform(DCT)coefficients were used to generate watermarking and embedded into the two least significant bits of its mapping block which was determined by chaotic sequence. The conventional authentication methods were improved by combining parity detection and the tampered region could be self-recovered by extracting characteristic information of the valid block. Experimental results demonstrated that the proposed method not only improved the quality and security of watermarked images, but also could detect tampered regions accurately and reconstruct images while keeping the recognition quality. Additionally, the scheme was not vulnerable to the content-only attack.