First-Level DWT Decomposition of Images

First, we need to perform first-level Discrete Wavelet Transform (DWT) decomposition on the image. This can be implemented using wavelet decomposition functions like wavedec2() in MATLAB or PyWavelets in Python, which separates the image into four subbands: LL (approximation coefficients), LH (horizontal details), HL (vertical details), and HH (diagonal details).

Next, we use two uncorrelated pseudo-random sequences to represent 0 and 1 in the watermark information. These sequences are typically generated using cryptographic methods or random number generators with specific seeds to ensure orthogonality. The sequences are then embedded into the LH and HL subbands obtained from the wavelet decomposition through coefficient modification techniques, such as additive embedding or quantization index modulation.

It's important to note that the complete algorithm comprises two main components: watermark embedding and watermark extraction. During the embedding process, the watermark information is inserted into the host image by modifying the selected wavelet coefficients. In the extraction process, the embedded watermark is recovered from the watermarked image using correlation detection or reverse transformation methods, typically requiring the original pseudo-random sequences for proper decoding.