Hash-based identification of sparse image tampering.

Area of Science:

• Digital Image Forensics
• Multimedia Security
• Information Theory

Background:

• The proliferation of digital content creation and editing tools has outpaced advancements in content protection.
• Unauthorized content modification and dissemination pose significant challenges to digital integrity.
• Multimedia hashing offers a robust solution for verifying content authenticity and detecting tampering.

Purpose of the Study:

• To develop an image hashing algorithm for content authentication and tampering identification.
• To provide proof of legitimacy for digital images and pinpoint semantic alterations.
• To create a universal hash resistant to various tampering methods.

Main Methods:

• An image hashing algorithm based on compressive sensing principles.
• Quantization of a limited number of random projections for hash generation with a small bit budget.
• Utilizing distributed source codes for efficient hash encoding and decoding.
• Employing hash to estimate mean square error distortion and identify sparse tampering in orthonormal bases.

Main Results:

• The proposed algorithm effectively authenticates images and identifies tampering.
• A universal hash signature can detect and identify diverse types of image modifications.
• The algorithm demonstrates robustness against content-preserving transformations like cropping, scaling, and rotation.
• The method maintains a small hash size through distributed source coding.

Conclusions:

• The compressive sensing-based image hashing algorithm offers an effective solution for digital content authentication and tampering detection.
• The universality and robustness of the hash enhance its practical applicability in multimedia security.
• The algorithm balances security requirements with efficiency, making it suitable for widespread use.