Digital Image Watermarking has emerged as an effective and trustworthy solution as a proof of one’s ownership and copyright protection even in the cases where the protected data is attacked in different possible manners. Watermarking procedure consists of four stages: structuring watermark, watermark embedding, processing watermarked image, extracting watermark. Watermarking for digital images can be categorized into frequency domain watermarking or spatial domain watermarking. The most frequently used frequency domain transformations are DCT and DWT. The well-known and widely used spatial domain methods are least significant bit (LSB) and histogram based. Digital watermarking performed in spatial-domain consists of easy operations and are generally computational cost efficient but gives poor results after common signal processing like filtering and JPEG etc. and also after geometric distortions like scaling and cropping etc. Whereas, watermarking techniques in frequency domain proves to be greatly robust against different attacks but consists of computational complexities.
To increase efficiency of watermarking, the frequency domain transformation used should be optimized first. Thus, the authors of presented an efficient architecture in terms of memory for lifting based DWT supporting dual mode: 9/7 lossy and 5/3 lossless coding showing remarkable reduction in latency and transpose memory with regular signal flow. The researchers proposed a new frequency domain DWT based watermarking algorithm having an additional watermark generation stage where watermark was constructed using host image and was disordered using Arnold transform. The writers reviewed various watermarking techniques with special focus on hardware implementation and presented the differences and similarities between these different types based on their assessment. The experimenters proposed an effective blind digital watermarking algorithm where the original image is DWT transformed and using a secret key a robust watermarked image is generated.
- Area is high
- Power consumption is high
Watermarking procedure consists of four stages: structuring watermark, watermark embedding, processing watermarked image, extracting watermark.
The DWT and IDWT approach followed is based on Haar wavelet transform which uses the following equations:
The equations used for determining the approximation (high) and detailed (low) component during forward transform and even and odd components (pixels) of the image during the inverse transform are:
The paper implements the algorithm in three stages:
Stage 1: This watermark embedding stage consists of generating watermarked image from the host image and watermark image with the help of Haar based DWT and BPS
Stage 2: The processing watermarked image stage where the watermarked image is attacked to check for its robustness under different extreme conditions prevailing over transmission media.
Stage 3: The watermark extraction stage consists of extracting watermark image from watermarked image without the usage of original image.
The flowchart depicting the procedure for embedding watermark in host image during image watermarking process using HL coefficients is shown in Fig 4 and same is followed for LH embedding. In this process the original gray scale image is transformed using frequency domain DWT to obtain HH, HL, LL, LH coefficients and then watermark is inserted in the suitable coefficients of transformed host image using spatial domain bit plane slicing. Then it is again back transformed or inversely transformed to generate watermarked image.
The HH and LL quadrants are not used for embedding watermark as they do not provide a robust and reliable solution and also the extraction of watermark from watermarked image is not possible under various circumstances. This paper draws a comparative analysis based on image quality measures for watermark embedded in HL and LH quadrants.
In extraction process the watermarked image is taken and transformed using DWT Haar equations mentioned previously and the coefficients for HH, HL, LH and LL are obtained. Using the pixel values of quadrant LH, the lower 4 bits of each pixel is extracted and used as higher 4 bits of pixel values of watermark image for the formation of extracted watermark image. The extraction process used in this technique is blind, that is during extraction the original image is not used to obtain the watermark from the watermarked image.
- Area overage is low
- Power consumption is low
- Xilinx ISE