<p>Traditional color image watermarking schemes often use transform domain techniques to enhance robustness, but often at the expense of real-time performance and practical applicability. To address this limitation, this paper proposes a novel watermarking scheme optimized for rapid execution and effective resistance against common attacks, particularly scaling. The method uses the maximum eigenvalue of the <i>S</i>-matrix derived from polar decomposition to embed watermark information directly into spatial blocks to computationally intensive frequency transformations at first time, which improves the real-time performance. A block expansion strategy is introduced to distribute watermark data uniformly across the host image, ensuring smoothness in embedded regions and enhancing the invisibility. Moreover, a fusion domain embedding mechanism is designed to counteract scaling attacks by analyzing the variation coefficient of the maximum eigenvalue. Experimental results demonstrate the proposed scheme effectively balances imperceptibility, robustness, and computational efficiency, offering a practical solution for secure and real-time image.</p>