Improved DCT-based detection of copy-move forgery in images

Area of Science:

• Digital Image Forensics
• Computer Vision
• Image Processing

Background:

• Digital image tampering techniques are increasingly sophisticated.
• Copy-move forgery, where parts of an image are copied and pasted within the same image, is a common manipulation.
• Existing detection methods may struggle with compressed or noisy images.

Purpose of the Study:

• To develop an improved DCT-based method for detecting copy-move forgery.
• To enhance the robustness of forgery detection against common image distortions.

Main Methods:

• The proposed method divides images into overlapping blocks.
• Discrete Cosine Transform (DCT) is applied to extract feature vectors from each block.
• Feature vector dimensionality is reduced via truncation, followed by lexicographical sorting for efficient comparison.

Main Results:

• Duplicated image blocks are located adjacent to each other in the sorted feature list.
• A similarity judgment scheme enhances the robustness of the detection.
• Experimental results confirm the method's effectiveness on images with JPEG compression, blurring, and additive white Gaussian noise.

Conclusions:

• The improved DCT-based method offers a robust solution for detecting copy-move image forgery.
• The technique demonstrates resilience against common image processing distortions.
• This contributes to more reliable digital image authentication.