Article Preview
Top1. Introduction
Accelerated advancement of contemporary transmission technology has already made digital content as a substantial component in today’s life. It also hikes various issues like illegal manipulation and distribution. As a result, intellectual property right protection has become a burning issue for digital content owners. Digital Watermarking is one of the prominent tools for serving copyright protection (Kannan & Gobi, 2015). Digital watermarking works on the concept in which the owner information is planted into the original information or digital data. The terms owner information and watermark are interchangeable. Through the relevant watermarking scheme, the security of digital data can be ensured and checked whether the content has been tampered or not(Laouamer, 2017). Generally, the digital image watermarking system can be implemented either in the transform domain or in the spatial domain (Al-Gindy, 2017). Transform domain based watermarking techniques are more popular than spatial domain since they provide much more robustness and imperceptibility (Kumsawat, 2010; Walia, Singh, & Suneja, 2015).
Over the past years, the distinct watermarking techniques have been evolved in the different transform domain as well as in the spatial domain. Many of existing methods are imperceptible but not robust or vice-versa. Generally, a digital watermarking scheme should be imperceptible and robust (C. Wang, 2016). Both imperceptibility and robustness are mutually strife to each other (Wu, 2016). So, the tradeoff has to be kept in mind during the designing of the digital watermarking algorithm, and it can be seen as an optimization problem. For solving this optimization problem, the genetic algorithms, fuzzy, the neural networks can be employed in image watermarking system (Naseem et al., 2014; Mehta, Rajpal, & Vishwakarma, 2015).
In the present paper, a digital watermarking scheme has been proposed to keep a balance between imperceptibility and robustness with a good efficiency rate by implementing GA in histogram shape method in conjunction with a Butterworth filter. GA is used to locate the best position to conceal the watermark. As a result, the perceptual quality and robustness are maintained. The Concept of Genetic Algorithms (GA) was devised by John Holland in 1970’. The GA method solves an optimization problem by inferring the rules of genetics and natural selection. The five components- Initialization, Evaluation, Selection, Crossover, and Mutation- compose the basic structure of genetic algorithms. The Figure 1 clearly shows these components:
Figure 1.
The general genetic algorithm training procedure
The first step of the Genetic Algorithms initialize the problem domain variables or set of solutions as chromosomes named as population. The solutions from one population are captured and utilized as a parent to generate a new population. This regeneration of the population is conducted with the expectation that the newly generated population will be excelled than the previous population. The selection from the previous population to form a new population is made according to the fitness function. This practice is repeated until the condition is satisfied. Successive population generation leads towards the optimal solution by relying on bio-inspired operators like selection, crossover, and mutation. The general training procedure of genetic algorithms is organized in the form of a block diagram, which is also elaborated in Figure 1.
In the sequel, Histogram is exploited in the proposed digital image watermarking system to embed the watermark. The Image histogram contains the essence of the intensities of an image, but impotent to reveal any specific information related to the connection among the pixels. As some of the attacks shift the position of pixels; as a result, robustness is affected. So, the histogram shape method is applied in the designing of the digital watermarking algorithm. Fundamentally, the histogram is a notable tool that illustrates the distribution of pixels of an image. The histogram of a gray-scale image can be represented in the form of Equation 1: