Digital image authentication and tampering detection have always been critical research topics within the domain of information security. A key challenge lies in that the imperceptibility of embedded watermarks tends to decrease when aiming for high embedding capacity. To address this issue, this paper proposes a self-recovery watermarking algorithm for color images based on a three-dimensional reference matrix. The algorithm utilizes a three-dimensional reference matrix for watermark embedding and extraction. The recovery bits are set as the average value of 2 \(\times\) 2-sized image blocks, reducing the error in tampering detection compared with common algorithms that use 8 \(\times\) 8 or 4 \(\times\) 4-sized image blocks. To ensure the integrity and security of watermarks, authentication bits in the algorithm are generated by the Secure Hash Algorithm 256. Experimental results demonstrate that the proposed algorithm achieves a high embedding capacity of 2 bits per pixel. The visual quality of watermarked images exceeds 48 dB for different image verifications. Meanwhile, the algorithm can accurately detect tampering, and the visual quality of recovered images can reach 38 dB even at a 50% tampering rate. This research provides an effective solution for balancing embedding capacity and imperceptibility while ensuring the security of watermarked images, which has important practical significance for digital image authentication and tampering detection.
