COLOR IMAGE WATERMARKING WITH QR DECOMPOSITION AND WAVELET: AN IN-DEPTH ANALYSIS

Abstract

Watermarking digital photos protects intellectual property. This work provides a blind watermarking method that balances undetectability with strength using RDWT and SVD. Changes in host image entropy determine where the watermark is inserted. An orthogonal matrix U for watermark embedding is evaluated using RDWT and SVD. The suggested method mixes up the binary watermark image with the Arnold chaotic map for safety. Many signal processing and geometrical attack issues were employed to test our technique. The experiments show that the suggested strategy reduces picture quality loss and is more resilient to JPEG2000 compression, cropping, scaling, and other adjustments. A mixed method for labeling photographs uses RDWT, DCT, SVD, and trigonometric functions. Putting all functions in one domain allows for visible, long-lasting, and undoable watermarking. The approach was tested with different host pictures and watermark intensities. A robustness tolerance of 0.8 is utilized to test the correlation-based extraction approach. We also compare the watermarked and original images using the Peak Signal-to-Noise Ratio (PSNR). The experiments show that the suggested method works whether the image is rotated, resized, blurred, contrast changed, JPEG files compressed, histogram equalized, image converted affinely, mean filtered, and Gaussian noise present. Normalized Cross-Correlation (NCC) always exceeds the tolerance level, even in skewed images. In this method, the retrieved image matches the original. The technology can safeguard copyright, settle ownership disputes, check material, authenticate users, and handle delicate situations requiring content integrity and undoing alterations.