Noise Addition to Host Images After Wavelet Transform-Based Watermark Embedding and Grayscale Watermark Extraction

After completing watermark embedding using wavelet transform techniques, we can enhance image complexity by introducing noise processing to the host image. This approach significantly improves watermark security and stability during grayscale watermark extraction. Noise processing can be implemented through two primary methods: adding random noise patterns (using functions like MATLAB's imnoise() with 'gaussian' or 'salt & pepper' parameters) or applying digital filters (employing filter2() or medfilt2() for mean/median filtering). These additional processing steps increase computational complexity but substantially enhance watermark robustness against attacks. The implementation typically involves: 1) Performing DWT (Discrete Wavelet Transform) using wavelet decomposition functions like wavedec2(), 2) Modifying coefficient matrices for watermark insertion, 3) Applying inverse DWT with waverec2(), 4) Introducing controlled noise using randn() for Gaussian noise or imnoise() for impulse noise, and 5) Extracting watermarks through coefficient correlation analysis despite noise interference. This layered approach strengthens resistance to common image processing operations while maintaining perceptual quality.