Using overlapping arrays for an improved spatial mean quantization watermarking

This paper presents a new method for watermarking digital images. To achieve a high robustness to noise, lossy compression, blurring and sharpening attacks, spatial domain array mean values are quantized. To recover the embedded data, the original image is not required. Based on this approach, low image quality degradations with a high robustness can be achieved. To reduce the image degradations, our method uses overlapping arrays to embed the information. The method bases on the uniform distribution of quantization errors and uses the intersections of overlapping arrays to optimize the PSNR value. Theoretical analyses and practical results show the advantage over the non- overlapping-array method. Experimental results confirm the theoretical assumptions.