一种用于图像内容鉴别的数字签名方案

提出了一种稳健的数字签名方案用于图像内容鉴别．通过对图像的预处理，从图像行与列中提取出对JPEG压缩稳健的原始信息序列，然后利用Hash函数对原始信息序列进行加密并提取签名比特．产生的数字签名可以有效地鉴别图像内容的真伪并对蓄意的修改进行交叉定位．由于数字签名对于JPEG压缩是稳健的，从而把JPEG压缩操作与对图像的恶意修改区分开来．理论分析结果表明这种数字签名具有较高的修改检测概率．仿真结果充分证明了该方案的正确性和有效性．

A Digital Signature for Image Content Authentication

A robust digital signature scheme is proposed for image content authentication. Different from conventional cryptographic signature scheme, the proposed signature scheme has the ability of detecting and localizing small tampered area, while distinguishing malicious tamper from moderate JPEG compression. Through the preprocessing, original message sequences robust against moderate JPEG compression are extracted from image rows and columns, which are then hashed to get the signature bits. In authentication, the tampered area can be effectively detected and localized by cross checking technique. The performance analysis of the proposed signature scheme is made and the probability model of tamper detection is derived, which demonstrates the good property of the proposed signature. The signature behavior under JPEG compression is also analyzed in detail to show its robustness when different compression quality factors are applied. Finally, experimental results are presented to demonstrate the correctness of theoretical analysis and the effectiveness of the proposed signature scheme.