Biometric Image Watermarking Using Ripplet Transform And QZ Decomposition: Analysis Of Related Works
In Modern age, with the boundless use of internet and wireless, the digital content sharing like image, audio, video has been increased like a nuclear reaction. But these contents can be attacked by an intruder so that the owner of the content will loss his/her copyright. One can use encryption of his/her data to get rescue from the attack. But, if somehow it can be decrypted into its original form, his/her data will be in danger again. So, it creates a much concern of how to protect digital contents from various kind of attacks. Proceeding all above things in mind, it has been found by the researchers that digital watermarking can solve the security problem at a certain degree.
The idea of digital watermarking is information hiding into a host signal such as image, audio, video etc. In Todays era, watermarking is largely used to secure the biometric data of an user. It can be used in two ways. Either embedding of personal information in biometric data or embedding biometric data into other digital content to protect the biometric data from illegitimate claiming. To give protection of biometric data various watermarking technique has been proposed in the literature. Four major issues are considered in watermarking system. They are robustness, imperceptibility, security and capacity. Robustness means the difficulty in eliminating the watermark from the host whereas imperceptibility refers to the invisibilty of watermark. The capacity dictates the length of the embedded message and the security means how secured the watermark is.
Nowadays, many researchers have been proposed many watermarking technique for biometric image. Fragile, semi-fragile and robust are the three techniques for watermarking. Robust watermarking mainly used for protecting the data from various attacks while fragile watermarking mainly used for authentication and protection. Watermarking can be done in spatial and tranform domain of an image. In frequency domain, the computational complexity is higher. The coefficients are slightly changed in transform domain. It provides changes in the image that makes it robust to attack compared to spatial domain. Spatial domain is pretty much straightforward. Compared to transform domain, watermarking in spatial domain is more fragile to common attacks.
On the other hand, watermarking in transform domain provides more robustness in various attacks. Watermarking is a novel technique which provides security of biometric images. Various methods in watermarking are used to ensure the protection of biometric image. Understanding the necessity of providing security in biometric images, researchers have given their consent to use watermarking in this field. Biometric image watermarking get popular day by day because using various kinds of template matching algorithm attacker can steal the biometric image. In this chapter, we will see various types of method used by various researchers to watermark the biometric image into a host image. Some researchers have done watermarking in spatial domain, but most of them in transform domain as it gives more robustness.
The transform which are used most in watermarking are discrete cosine transform (DCT), discrete wavelet transform (DWT), contourlet transform (CT). Discrete curvelet transform (DCT), hermite transform and radon transform are regarded as less known method. Researchers use blind biometric image watermarking based on contourlet transform. Here, contourlet transformation is used to decompose the host image into a multiscale, directional subimages. High frequency directional subband which contains highest energy is manipulated for watermark embedding. Watermark image is encrypted using Arnold transform.
In other work authors use fragile watermarking using discrete curvelet transform. Here, multiple biometric images are used as watermark. Watermark is embedded here in the mid band frequency curvelet coefficients of host image.
In one more research authors use sparse watermarking technique which is based on compressive sensing theory. After applying the 2 level DWT in the LL2 subband is used for watermark embedding. Before embedding the coefficients of host image is modified according to the sparse measurement which is in watermark the biometric image. Then apply the Cox equation for embedding the watermark.
