A Hybrid CNN–RNN Framework for Watermark Preserving Image Smoothening, Denoising, and Video Authentication

Abstract

The rapid growth of digital multimedia content has increased the need for robust techniques that ensure both visual quality enhancement and content security. Conventional image smoothening and denoising methods often degrade embedded watermark information, while existing watermarking approaches frequently struggle to maintain robustness under image processing operations and video transformations. To address these challenges, this study proposes a Hybrid CNN–RNN Framework for Watermark-Preserving Image Smoothening, Denoising, and Video Authentication. The proposed framework employs Convolutional Neural Networks (CNNs) to perform adaptive feature extraction, image smoothening, and noise reduction while preserving the structural integrity of embedded watermarks. In addition, Recurrent Neural Networks (RNNs) are utilized to analyze temporal dependencies in video sequences, enabling reliable video authentication and tamper detection. The framework integrates watermark embedding, denoising, smoothening, and authentication within a unified deep learning architecture. Performance evaluation is conducted using standard image and video datasets with metrics such as Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), Mean Squared Error (MSE), Bit Error Rate (BER), and authentication accuracy. Experimental analysis demonstrates that the proposed approach effectively enhances image quality, improves watermark robustness, and maintains temporal consistency in video authentication tasks. The results indicate that the hybrid CNN–RNN model provides a practical and scalable solution for secure multimedia processing, digital copyright protection, and content verification applications.