Research the efficiency of image processing algorithms in zero watermark schemes

• Published in: Sučasnij stan naukovih doslìdženʹ ta tehnologìj v promislovostì (Online) no. 1(31); pp. 102 - 114
• Main Authors: Poddubnyi, Vadym, Sievierinov, Oleksandr, Nepokrytov, Dmytro
• Format: Journal Article
• Language: English
• Published: 31.03.2025
• ISSN: 2522-9818; 2524-2296
• DOI: 10.30837/2522-9818.2025.1.102

Subject matter: image transformation algorithms used in zero-watermarking techniques for development authentication algorithm. Objectives: identify effective image transformation algorithms for use in zero-watermarking schemes. The study addresses the following tasks: examining the range of existing image transformation algorithms, formulating informal requirements for image transformation algorithms to be used in zero-watermarking-based authentication schemes, and making assumptions about the feasibility of using each analyzed image transformation algorithm. To achieve these objectives, the following methods are employed: modeling – software implementation of each studied algorithm, empirical methods – application of algorithms and observation of transformation results, mathematical methods – calculation of normal correlation metrics and peak signal-to-noise ratio (PSNR). Results: an analysis was conducted on a set of algorithms that could potentially be used in zero-watermarking schemes for authentication purposes. A methodology was developed to evaluate algorithms while considering image dimensionality reduction due to compression. Additionally, requirements for image processing algorithms in zero-watermarking-based authentication were established. Conclusions: the study identified the most effective image transformation algorithms for use in zero-watermarking authentication schemes: DWT (Discrete Wavelet Transform), SVD (Singular Value Decomposition), DCT (Discrete Cosine Transform), and K-means clustering. For low-resolution images, DCT is a viable option. The most effective algorithm combinations are DWT + DCT and DWT + K-means, as these combinations ensure optimal robustness to noise while maintaining sensitivity to distinguish similar images. Future authentication schemes based on these algorithms may be useful (in combination with IoT devices), including for user authentication in enterprises and organizations.