An Image Watermarking Algorithm Based on Embedding Pixel Values Directly with DCT and Block Selection

The current image watermarking algorithms have developed from embedding binary images to grayscale and color images. However, most existing algorithms employ quantization index modulation (QIM) and spread transform dither modulation(STDM) for embedding, which require converting watermark pixel values into bit sequences. Although QIM and STDM can provide good robustness, they increase the embedding payload. Unlike embedding bit sequences using QIM and STDM, this paper proposes a discrete cosine transform (DCT) based methods that directly embeds watermark pixel values, combined with an image block selection algorithm. The DCT-based block texture selection is utilized to prioritize embedding in smoother regions to reduce vulnerability to filtering attacks. Watermark pixel values are normalized against DC coefficients before replacing selected DCT low-frequency components. Furthermore, discrete wavelet transform (DWT) and an image geometric correction algorithm are integrated to enhance robustness. Experimental results demonstrate that the proposed algorithm matches or exceeds QIM-based algorithms against certain attacks, particularly against image enhancement attacks. It also outperforms existing DCT-based pixel values embedding algorithms.