Double JPEG compression is the most common process to hide image manipulation. Therefore, it becomes necessary to detect the double JPEG compression. Several approaches have been developed for double JPEG compression detection with high accuracy, but they do not provide a unified solution in terms of real-time applicability. To address this issue, a new 953-dimensional unified detector is proposed. The unified detector is a combination of 44 spatial domain features and 909 frequency domain features. Extensive experiments are performed on UCID and RAISE databases to evaluate the robustness of the proposed detector. In addition, the proposed detector is evaluated and compared with a state-of-the-art method under a multi-class (9-class) classification.

双 JPEG 压缩是隐藏图像处理的最常见过程。因此，有必要检测双重JPEG压缩。已经开发了几种用于高精度双JPEG压缩检测的方法，但它们在实时适用性方面没有提供统一的解决方案。为了解决这个问题，提出了一种新的 953 维统一检测器。统一检测器是 44 个空间域特征和 909 个频域特征的组合。在UCID和RAISE数据库上进行了广泛的实验，以评估所提出的检测器的鲁棒性。此外，所提出的检测器在多类（9类）分类下与最先进的方法进行了评估和比较。