利用篡改后JPEG图像量化表不一致的特性，提出一种针对JPEG图像的篡改盲检测新方法。通过智能选取若干图像块，迭代估计出待测图像的原始量化表，大致定位出篡改区域。然后用估计出的原始量化表对篡改区域再进行一次JPEG压缩，由压缩前后图像像素值的差值最终确定篡改位置。实验结果表明，提出的估计量化表算法复杂度小，精度高。检测算法能有效地检测出多种篡改类型的JPEG图像，且对篡改区域和未篡改区域压缩因子相差较小的JPEG合成类篡改，检测正确率更高。

A new approach is proposed to detect JPEG image forgery through the inconsistencies of quantization table. The original quantization table is estimated iteratively by selecting qualified blocks intelligently and further used to position the approximately doctored areas. Then the doctored area is compressed by the original quantization table estimated. Using the difference between the double compressed image and the test image, doctored areas will be more precisely determined. Simulation results show that the quantization table estimation algorithm proposed is less time consuming and of higher precision. The detection algorithm can expose various kinds of JPEG image forgeries and it is more effective for small difference quality factor between the doctored area and undoctored area’s JPEG images.