Watermarking And Digital Signature For Image Authentication Computer Science Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

The advancement of computers and internet has exploded in the last few years. The digital images are distributed, duplicated and tampered easily through www, thus the protection of transmitting data with digital images becomes an important issue. In this paper, we proposed a technique which is robust image authentication. The proposed scheme includes two parts. The first is embedding a message into a image. The second is signature process, which can be used to prove the authenticity integrity of the images. The input of the signature process is the edge properties extracted from the watermarked image.

KEYTERMS: Watermarking, Digital Signature, Image Authentication, Clustering Algorithm


Image authentication is becoming more important because of easy replication and modification of digital images. The study of image authentication falls into two broad categories: Digital watermarking and Digital signature.

Digital watermarking is an image authentication technique that embeds invisible information into an image. For image content authentication, the watermark sequence can be extracted from the image and it is used for authentication purpose [4].Watermarking techniques are

including to various methods such as fragile watermark and semi-fragile watermark.

A fragile watermark is very sensitive and is designed to detect every possible change in a marked image, but in most multimedia applications, minor data modifications are acceptable as long as the content is authentic. A semi-fragile watermark is robust to acceptable content preserving manipulations such as lossy compression while fragile to malicious distortions such as content modification and is used for data authentication.

Digital signature is a set of feature extracting from an image to form a compact representation that can be used for authentication [19].There are two general modes of digital signatures: appendix mode and message recovery mode.

In appendix mode, the signed message and the corresponding signature are transmitted to the verifier separately. The hash value of the signed message is usually involved in the signature-generation process to protect the signed message and prevent the signature from being modified illegally [12]. While the receiver wants to verify the received message, the received message is needed as the input of the hash function. The hash value of the signed message is used in the verification process to prove the source of the signed message. In message recovery mode, the signed message is recovered by the receiver from the received signature. Message redundancy schemes and hash functions are used to ensure the correctness of the recovered message and to resist the forgery attacks.

The paper is organized as follows. Related works in Section 2.Section 3, feature extractions are briefly introduced. Section 4 presents the hash function. Proposed image authentication scheme in section 5. Section 6 provides various experimental results. Finally, Section 7 concludes the paper.


Venkatesan et al, [22], have proposed a robust image hashing scheme for the purpose of an image indexing. Wavelet decomposition of an image is first computed and each subband is then randomly tiled into small rectangles. For each rectangle, some statistics measure such as mean or variance is calculated and quantized using a random quantizer. The results are form to the hash value.

Lin and Chang [20], have proposed an image authentication technique that relies on the invariance relationship of DCT coefficients. Their scheme is sensitive to catch malicious manipulations made in a part of an image at the same time is resilient to JPEG compression.

Image authentication scheme is the structure of an image as the digital signature [1, 14, 20], he image structure is obtained by identifying the parent child pairs located at the multiple scales in the wavelet domain. The parent child relationship has been found to be quite stable for several acceptable manipulations like JPEG compression and blurring and is able to detect parts in an image. Content dependent structural image features and wavelet filter parameterization are incorporated into the traditional crypto signature scheme to enhance the system robustness and security

Der-Chyuan et al [21], have proposed an public key based fault resilient and compression tolerant digital signature. They discussed not only verify the authenticity and the integrity of commodities.

Min wu et al [24], have proposed watermark embedded for image authentication in which the signature is embedded in the image and the marked image can be kept in compressed form. Both [20, 23] can detect and locate where comparing has occurred.

Dekun Zou et al [15], have proposed new watermarking framework for image content authentication such that the original image can be restored is robust to JPEG compression and is signed with cryptographic signature algorithms.

The proposed an image authentication scheme [21 23] are verifying the originality of the received images. The authentication signature can distinguishes content changing manipulations from content preserving manipulations.

Chun-Shien Lu et al [19], have proposed a multipurpose watermarking technique which can be simultaneously achieve both copyright protection and content authentication by hiding multipurpose watermarks at the same time which can be used to recover the host image within distinguishable perceptual degradation. This information is very useful in calculating the detector responses about robust watermarking and fragile watermarking.

Chih-Ming Kung et al [2], have proposed watermarking technique that includes watermarking performed in frequency domain and digital signature process. The input of the signature process is the edge properties extracted from the image. The signature can be correctly verified when the image is incidentally damaged.

Shuiming Ye et al [16], have proposed integrated method for error detection and error concealment based on image content. Error detection method makes use of some features of its neighboring blocks of the damaged image. The damaged image blocks are detected by exploring the contextual information in images.

Ching-Yung Lin et al [20] have proposed an image authentication technique that distinguishes the JPEG lossy baseline compression from other malicious manipulations. But prevent malicious manipulations.

M.Y siyal et al [9], have proposed a secure and robust hashing scheme which used key dependent feature extraction to form the image hash. This method is first divides an image into non overlapping blocks and calculates the hash of each block using secret permutation key.

Sunil Lee and et al [7], proposed a reversible watermarking scheme with high embedding capacity for digital images without using lossless compression. Here watermark is embedded into wavelet co-efficients using either LSB-substitution or bit shifting.

In this digital era, transforming information in the form of textual and digital images is unavoidable through an electronic medium. Nowadays, transforming messages in the communication channel without any disturbances is challengeable. Message authentication and user authentication requires digital certificate and digital signature respectively and an encryption algorithm (two-way function) also needed. These processes for authentication are very complex and time consuming to perform all these mathematical operations. Contented image authentication is an approach, not requiring digital certificate but one of the features of the given images would be required.

Watermarking is a technique to authenticate an image. But it is alone could not authenticate efficiently. We propose a method in addition with a watermarking, extracting feature from the watermarked image and applying one-way and two-way function, which produce unintelligible information. It provides both a strong user and message authentication.


Initialize K-Cluster

Scan the Image

Random Seed Point Selection

Compute the Distance between each Pixel and Seed Point

Group the Pixels to Minimum Distance Cluster

Update Centroids

Compute the New Distance

Clustered Image

Compare Centroids



Fig (1): Flow Diagram of Feature Extraction

Transforming the input data into the set of features is called feature extraction

If the features extracted are carefully chosen it is expected that the features set will extract the relevant information from the input data. This operation is performed to do the desired task using this reduced representation instead of the full size input.

The watermark image is extracted using the K-means algorithm.

When the input data to an algorithm is too large to be processed and it is suspected to be notoriously redundant (much data, but not much information) then the input data will be transformed into a reduced representation set of features (also named features vector).

It can be used in the area of image processing which involves using algorithms to detect and isolate various desired portions or shapes (features) of a digitized image or video stream. The feature extraction is shown in Fig (1).


The basic edge-detection operator is a matrix area gradient operation that determines the level of variance between different pixels. The edge-detection operator is calculated by forming a matrix centered on a pixel chosen as the center of the matrix area. If the value of this matrix area is above a given threshold, then the middle pixel is classified as an edge. Examples of gradient-based edge detectors are Roberts, Prewitt, and Sobel operators.

All the gradient-based algorithms have kernel operators that calculate the strength of the slope in directions which are orthogonal to each other, commonly vertical and horizontal. Later, the contributions of the different components of the slopes are combined to give the total value of the edge strength.

The Prewitt operator measures two components such as vertical and horizontal. The vertical edge component is calculated with kernel Mx and the horizontal edge component is calculated with kernel My. |Mx| + |My| gives an indication of the intensity of the gradient in the current pixel.


Cluster analysis, an important technology in data mining, is an effective method of analyzing and discovering useful information from numerous data. Cluster algorithm groups the data into classes or clusters so that objects within a cluster have high similarity in comparison to one another

K-means algorithm belongs to a popular partition method in cluster analysis. The most widely used clustering error criterion is squared-error criterion, it can be defined as

Where is a chosen distance measure between a data point and the cluster centre, is an indicator of the distance of the nj data points from their respective cluster centre.


First extract the original image after detect the edge. The maximum and minimum value was founded using the image. It is based on number of clusters. Then the centroid value was calculated using the following formula


Where k is a number of clusters. The centroid values are varied for each input image. Then find out the difference between each pixel. The resulted pixels are grouped which form the image feature. The flow diagram of k-means algorithm is shown in Fig (2).

Original Image

Find Max & Min



Image Feature




Fig (2): Flow Diagram of K-means Algorithm


The technique of cryptographic hash functions is utilized to achieve these security services. Hash Functions compress a string of arbitrary length to a string of fixed length. The purpose of a hash function is to produce a "fingerprint" of a file, message, or other block of data. The hash functions are including to the various method such as SHA-1, MD5. Here we are used to the SHA-1 algorithm. Because the MD5 is using small number of bits compare than SHA-1 algorithm.


Watermark Image

Feature Extraction

Hash Function

Hash Value



Cipher Text

Watermarking technique

The hash algorithm basic round, as the specifications of the standard define, has 80 steps. In order to implement in hardware this operation a feedback loop in the basic round output is added.

Each message, after it has been appended by padding bits (PADDING component), is processed in blocks of 512 bits in length. The processing relies on four additive constants, bit rotations (circular left shift) and additions modulo 232. Three different primitive logical functions are used each one for a corresponding step of the processing. Each logical function performs a set of bit-wise logical operations. It takes three 32-bit words as input and produces a 32-bit word as output

An intermediate data register is used for temporary storage of the data after every transformation round execution. The last transformation unit transforms the data in final form. The output of this unit is the final message digest.


The proposed authentication scheme is a kind of sender and receiver side protocol. The main work of the sender is to generate the digital signature for intended image and to create some parameters for the process of verification that performed by the receiver side. In another side, the received image is verified by the receiver.


The original image and message is embedded using LSB algorithm [7], which yields us a watermark image. A feature of watermark image is extracted using k-means algorithm [17]. A hash function is applied to the image feature using SHA-1 algorithm to digest the image feature. A hash value is encrypted using RSA algorithm with sender's private key, which yields us a scrambled data. The scramble data and watermark image is encrypted

Fig (3): Diagram of Signing Procedure

using RSA algorithm with receiver's public key, which yields us a scrambled data. These data are sending to the receiver side. It provides as strong security such as user and message authentication and confidentiality.The signing procedure is shown in Fig (3).


Cipher text


Watermark Image

Cipher text


Hash Value

Hash Function

Feature Extraction

Hash Value

If valid/


Apply the process of removing watermarking from image


Original Image

Receiver's Private Key

Sender's Public Key

If computed & received message digest are same


(b) Extract Image

Fig 7: Gold hill Image


(b) Extract Image

Fig 6: Fruits Image


(b) Extract Image

Fig 5: Lena ImageThe received data is first decrypted using receiver's private key to get the watermark image and cipher text. The cipher text is decrypted using sender's public key, which provides the hash value. The watermark image is used to extract the image features the way how sender performed. The hash function is applied to the extracted image feature using SHA-1 algorithm which provides us a digest the image feature. If the decrypted hash value and newly calculated hash value are same, then the received image is authentic, otherwise it is unauthentic. If computed and received hash value are same and then apply the process of removing watermarking from image. Finally get

Fig (4): Diagram of Verification Procedure

message and original image. The verification procedure is shown in Fig (4).


In our experiments we used four grayscale images of size 512x512 pixels such as Lena, cameraman, goldhill and peppers. These images are first extracted using clustering algorithm after detect the edge using edge detection method.


(b) Extract Image

Fig 8: Cameraman Image


The proposed method of secure data transmission for image authentication, in which the watermark image can be extracted using clustering algorithm and signed with cryptographic signature algorithm. It provides integrity and verifying authenticity of images.