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.

Steganography is the art and science of hiding and has its place in security. The word Steganography has its orgin from the Greek word "Stego" which means covered and graphia means "writing". A steganographic system thus embeds hidden content in an innocuous cover media, so as not to arise an eavesdropper's suspicion. This paper proposes an innovative method to enhance the available algorithms and produce a new secure and robust steganography method. In this method edge detection is combined with steganography to improve the imperceptibility of the hidden data. Edges of images offers high embedding capacity compared to other portions of the image. The edges are the pixels where the gradient magnitude is greater than the neighbouring pixels taken in both sides. This is because embedding in the central portion of the image is susceptible to greater distortion. Hence message is hidden in the pixels representing the edges of images. This outcome however depends upon the properties of the cover image being used. This paper deals with color images and compares the performance of application of edge detection with normal steganographic techniques. The results confirm that a high quality stego image is guaranteed.


Internet has become the most effective and fastest media for communication. Albeit, it is susceptible to face many problems such copyright, hacking, eavesdropping etc. Hence the need for secret communication is required. Cryptography and Steganography are the two fields available for data security. Cryptography is a technique in which the data is scrambled in an unintelligent gibberish fashion so that it becomes difficult for any malicious user to extract the original message. Only the desired recipient will be having the code for decryption and will be able to extract messages. Cryptography has helped a great deal in data security but it has some disadvantages. The encrypted data will arouse suspicion to malicious users and there is a possibility of it being decrypted or being suppressed. Hence the intended information might not reach its destination effectively. The disadvantages of Cryptography have lead to the development of Steganography. Steganography deals with data hiding. That is the secret message is hidden in a cover media (images, audio, video) in such a way that any malicious user won't know that a secret message has been transmitted. The blending of data in the stego-media can be considered as secret communication but this is not similar to cryptography and hence steganography has an advantage over cryptography since the stego-image appears to be normal and it conceals the fact that a secret communication or secret sharing is going on.

Steganography field is young and has its place in security but the concept of data hiding is not new. Our history reveals a plethora of data hiding methods. According to the Greek historian Herodotus who lived around 471 B.C, Histiaeus of Miletus used to tonsure the head of his most trusted slave and tattooed a secret message on his scalp. Within months the slave's hair would have grown and then he was sent to the Greeks who shaved his head and read the message. During the World War II invisible inks and microdots where used to embed secret information in books and newspapers. Another example of non technical steganography is during the Vietnam War the captured US armed forces used hand gestures during the photo sessions to convey some military secrets.

In terms of "Digital Steganography" an electronic file can be embedded in another electronic file. For this a suitable cover media is required which can be text, image, audio, video or any other electronic file. Digital watermarking is used to establish the ownership for a particular data. Steganography can be used for preventing privacy and can differentiate the difference between the rightful owner and the copyright.


This paper deals with steganography using edge detection algorithm and color images are used as the cover media. The edge detection algorithm is used to detect and highlight the discontinuities in an image. This was first developed for processing satellites pictures. Edge pixels are detected in order to hide data. A pixel is regarded as an edge if the gradient magnitude of the pixel is greater than the neighboring pixels taken in both the directions. The reason behind selecting edge pixels is that central main part will be smooth and hence the bits of the neighboring pixels will be nearly same. By embedding in the smooth region there will be greater distortion and hence it will be perceptible. The proposed methodology is every much efficient and has higher robustness to steg- analysis. Color images are used here and the edges of these images are detected and represented as bit 1. Then using any steganographic method example LSB substitution the messages are embedded in the edge pixels. To improve randomness the background pixels can also be used for embedding with different technique. For example in the edge pixels 4 bit embedding can be done and in the background pixels 2 bit embedding can be done. This presentation has been divided into three modules. They are as follow- 1) Detection of edge pixels in the cover image 2) K-bit LSB Substitution Method


Edge detection algorithm is used in order to detect the edges of the image. An edge is regarded as the pixel whose gradient intensity is greater when compared with the neighboring pixels. The edge detection algorithms use six different edge detectors. They are as follows- 1).Sobel Edge Detector 2).Robert Edge Detector 3).Prewitt Edge Detector 4).Laplacian of Gaussian Edge Detector 5).Canny Edge Detector 6).Zero Crossing Edge Detector

In an image an edge is characterized by the significant discontinuity in the gray levels in order to distinguish between the boundaries between two regions in an image fragment. The edge detection is a significant image processing and machine vision due to the fact that edges are considered as the most important aspect for analyzing the information of the cover image. There are many ways to find the edge detection. They are by using gradient and Laplacian. The Gradient method detects the edges by looking for the maximum and minimum in the first derivative of the image. The Laplacian method searches for zero crossings in the second derivative of the image to find edges. The function of the different edge detectors are illustrated as follows:

Sobel Edge Detector:

Edge computes an approximation of the gradient of the intensity function. This edge detector detects edge in the horizontal and vertical directions. Based on the one dimensional analysis, the theory is carried over to two dimensional methods. The Sobel method is used for 2D spatial gradient measurement on an image.

Prewitt Edge Detector:

The Prewitt edge detector calculates the maximum response of a set of 8 convolution kernel to find the local edge orientation for each of the pixels. This process is known as edge template edging.

Robert Edge Detector:

Robert edge detector performs a simple, quick, 2D spatial gradient measurement on an image by convolving with 2x2 kernels. These kernels respond maximally to the edges running at 45 degrees to the pixel grid.

Laplacian of Gaussian Edge Detector:

The image is first blurred by convolving it with the image with the Gaussian. Then Laplacian is performed of the blurred image. Then the zero crossings of Laplacian are performed and then the local variance at this point compared to the threshold. If the threshold is exceeded then it is called an edge. The median filter is used to file the image. This filter is used in order to filter out the spot noise while preserving the edges.

Canny Edge Detector:

In Canny Edge Detection the image is smoothened by Gaussian Convolution. After this the 2 D first derivative is found out and non maximal suppression is performed. This is due to the fact that edges are considered as continuous line and by performing the suppression the discontinuous lines will be removed. The local maxima are considered as the edge. Tracking and hysteresis are performed for determining the edge. The Canny edge detection algorithm depends upon three parameters-1.Size of the Gaussian filter, 2.Upper Threshold and 3.Lower Threshold. The Canny Edge Detector uses 4 filters and detects edges in the horizontal, vertical and diagonal directions.

For the determination of edges we require an optimal edge detector. Canny is found to be the optimal edge detector due to the following factors: 1) Good detection: The algorithm should mark as many real edges in the image as possible. 2) Good location: Edges marked should be as close as possible to the edge in the real image. 3) Minimal response: A given edge in an image should only be marked once. 4) The detector must be able to distinguish between image noise and edges. It should create any false edges.


In the classic LSB Substitution the secret message is considered as a bit stream and is embedded in the LSB's of the 8 bit pixels of the gray images. Suppose the pixels P1, P2 and P3 are present first in the image. Their bit representations are as follows: [1110 0011], [1100 0010] and [1011 0010]. The data to be embedded is [110001]. The data is embedded in the last two LSB's of the image pixels. The resulting pixels will be as follows: [1110 0011], [1100 0000] and [1011 0001]

The LSB Substitution can be done using k=1 to 4. Substituting in the MSB's will lead to disruption to the image and the image becomes easily susceptible to malicious users. This paper deals with color images. Hence LSB Substitution is done in the color images. The color images are represented with 24 bit pixels. There are three channels red, green and blue. Each of these channels is represented by 8 bits. The LSB Substitution can be performed in these three channels. Hence much more data can be embedded in the colored images when compared to the gray images. For example let the color pixel be represented as [1100 1000, 1100 0010, 0100 1100]

The data to be embedded is represented as [111000100]. Now the data is embedded in the last three LSB's of the pixels. [1100 1111, 1100 0000, 0100 1100]

Our methodology deals with detecting the edges of the color image and then performing the LSB Substitution in the edges. This methodology gives higher efficiency and gives better response. The method in which the LSB Substitution is done can be randomized to give much higher efficiency and robustness.