Sharif Digital Repository

Capacity is one of the most important parameters in image watermarking. Different works have been done on this subject with different assumptions on image and communication channel. However, there is not a global agreement to estimate watermarking capacity. In this paper, we suggest a method to find the capacity of images based on their complexities. We propose a new method to estimate image complexity based on the concept of Region Of Interest (ROI). Our experiments on 2000 images showed that the proposed measure has the best adoption with watermarking capacity in comparison with other complexity measures. In addition, we propose a new method to calculate capacity using proposed image... 

Some of the most important classes of watermark detection methods in image watermarking are correlation-based algorithms. In these methods usually a pseudorandom noise pattern is embedded in the host image. The receiver can regenerate this pattern by having a key that is the seed of a random number generator. After that if the correlation between this pattern and the image that is assumed to have the watermark is higher than a predefined threshold, it means that the watermark exists and vice versa. Here, we show the advantage of using the Julian set patterns as a watermark, instead of the commonly used pseudorandom noise pattern. Julian set patterns can be regenerated in receiver with few... 

One of the most important parameters in evaluating a watermarking algorithm is its capacity. In fact, capacity has a paradoxical relation with other two important parameters: image quality and robustness. Some works have been done on watermarking capacity and a few on image complexities. Most works on watermarking capacity is based on information theory and the capacity values which are calculated based on these methods are very tolerate. In this paper we propose a new method for calculating image complexity based on Region Of Interest (ROI) concept. After that we analyze three complexity measures named: Image compositional complexity (ICC), Quad tree and ROI method, with three different... 

One of the most important parameters in evaluating a watermarking algorithm is its capacity. Of course capacity has a paradoxical relation with two other important parameters: image quality and robustness. Generally, watermarking capacity is expressed by bit per pixel (bpp) unit measure. This measure does not show what the side effects would be on image quality and capacity. Therefore, we proposed a two dimensional measure named capacity curve which shows the effects of capacity on visual quality. In this measure we can find the amount of degradation by a numerical value for each modification in pixels. Our experimental results are compatible with previous works in this field. © 2008... 

Some of the most important classes of watermark detection methods in image watermarking are correlation based algorithms. In these methods usually a pseudorandom pattern is embedded in host image. Receiver can regenerate this pattern by having a key that is the seed of a random number generator. After that if the correlation between this pattern and the image that is assumed to have the watermark is higher than a predefined threshold, this means that the watermark exists and vise versa. In this paper we showed the advantage of using the Julian set patterns as a watermark, instead of commonly used pseudorandom noisy pattern. Julian set patterns can be regenerate in receiver with few... 

Determining the capacity of watermark for images is a complex task. This capacity may be influenced by many factors such as the content of image and also the strength of watermark. In some recent works the image content has been considered for calculation of watermark capacity. One such approach is to use the Noise Visibility Function (NVF) introduced in [3]. In this paper we introduced a new method for finding NVF according to the quad tree representation of image instead of using pixel neighborhood data. Our experimental results showed that the depth of a pixel in image quad tree represents much information about the pixel and its maximum allowable distortion compared to pixel neighborhood...