Efficient fragile watermarking for image tampering detection using adaptive matrix on chaotic sequencing

• A novel adaptive matrix designed to reduce the complexity of singular decomposition. • Achieves detection performance comparable to SVD, Hadamard, and Walsh methods, with higher efficiency. • Reduces computation time by 20% during embedding and 15% during extraction processes. • Enhances tampering detection's stability and reliability in regular and irregular attack scenarios. This paper introduces a novel fragile watermarking technique for image forgery detection using adaptive matrices derived from Walsh and Hadamard transforms. The proposed method overcomes the limitations of traditional SVD, Hadamard, and Walsh methods by eliminating negative coefficients, simplifying the algorithm structure, and optimizing the computational complexity. The embedding process uses a two-stage authentication mechanism with a 16-bit validation scheme, ensuring precise forgery localization. This adaptive approach is intelligently designed to adapt the matrix pattern to the image characteristics. At the same time, the utilization of logistic sequencing enables the generation of non-periodic and non-convergent patterns, which significantly improves authentication efficiency and accuracy. Performance evaluation shows an average PSNR value of 55.90 dB and SSIM above 0.99, indicating a high degree of imperceptibility. In addition, this method achieves detection accuracy comparable to previous approaches, with an overall recall value of 1.00 and a TPR exceeding 0.96 across multiple forgery scenarios. This method offers the best efficiency compared to SVD, Hadamard, and Walsh methods and consistent authentication performance stability across multiple forgery levels. These advantages allow the proposed method to be developed in tampering detection applications that require speed and reliability.