Image watermark detection techniques using quadtrees

The quadtree, a hierarchical data structure for the representation of spatial information based on the principle of recursive decomposition, is widely used in digital image processing and computer graphics. This paper demonstrates the detection of invisible watermarked images generated by popular watermarking techniques, including CDMA, DCT, DWT, and Least Significant Bit (LSB) using quadtree. Results corresponding to typical (512 × 512) pixel images show differences among these methods when they are used. Each time we use the same image, the original images and invisible watermarked image to test the four methods in conjunction with quadtree decomposition. In addition to the subjective method represented by quadtree, many objective evaluations such as Pearson correlation, mean square error (MSE), Structural SIMilarity Index (SSIM) and false positive and false negative were used to give the comparison criteria between original and watermarked images. In results, the quadtree decomposition considered a promise subjective method to recognize among these watermark techniques.