Image Watermarking Using Adaptive Local Noise Reduction Filter

We presents in this paper a method of image watermarking using adaptive local noise reduction filter. For watermark embedding, the method modifies the reflectance component contained in the blue color component of a host color image in order to embed a binary image. For watermark extraction, adaptive local noise reduction filter is utilized to remove the watermark from the watermarked component, and the difference between the result and watermarked component is used to acquire an extracted watermark. The weight Peak Signal to Noise Ratio (wPSNR) is utilized to evaluate quality of the watermarked image, while Bit Correction Rate (BCR) is utilized to evaluate accuracy of the extracted watermark. Performance of our method is compared with the related previous method to demonstrate the improvements. As shown in the experimental results, improvements in terms of BCR at equivalent $wPSNR are obtained under various types of attacks from Stirmark benchmark.