Embedding Data In Personal Image By Using Mosaic Image

Many ideas have been dealt with in recent years about hiding data and different types and sizes of images. Different ways of hiding leads to unifying the desirable goals of this process. The main goal of a great bits of information with the smallest size or at least maintaining the original one of the hidden images or the target ones. One way to achieve the two goals is the hiding method by using mosaic image. Mosaic Images that are used image processing are those made by combining small tiles. The tiles are the source images which aim at restructuring the original visual information which is rendered into a new mosaic-like style. The Creation of mosaic images, made by a sequence of partial views, is a powerful means for obtaining a larger view than available within a single view, and it has been used in a wide range of applications especially in the field of steganography. Using this way, it is possible to hide information with the same size of the original image without increasing the size or the compression of data to reduce the size. In this research, twice the data size used has been hidden by hiding the data of the ‘secret’ image embedded in a ‘target’ personal image, noising the image itself. Thus the features of personal image disappear, making a hidden image and a noisy one. Different types and extensions have been chosen in this research including personal, natural, or satellite images. The personal image is used as a target image to hide other data inside it while other images are secret ones embedded in the personal image. Matlab is used to execute mosaic image algorithm.


Introduction
Steganography is the science of hiding secret messages into cover media so that no one can realize the existence of the secret data [1] [2]; it is derived from the Greek, since it was used for covered writing; it is essentially means "to hide in plain sight". According to Cachin [3], finding the message in steganography, which is the art and science of communication, is unattainable. Simple steganographic techniques have been in use for hundreds of years, but with the increasing use of files in an electronic format, new techniques for information hiding have become possible [4].
Existing steganography techniques can be classified into three categories: image, video, and text steganographies; image steganography aims at embedding a secret message into a cover image with the yielded stego-image looking like the original cover image [5].
Many image steganography techniques have been proposed [6] [7] [8] [9], and some of them try to hide secret images behind other images [8], [9]. The main issue in these techniques is the difficulty to hide a huge amount of image data into the cover image without causing intolerable distortions in the stego-image. [10] One of these techniques is called Mosaic Image, which means the collection of small images arranged in such a way that when they are seen together from a distance they suggest a larger image. Arranging the small images to form a large picture as much as possible to helps in giving the larger form in a visual way, and then adjusting their colors to suggest the overall form in a better way [4]. Mosaic images can be classified into four types: crystallization mosaic, ancient mosaic, photo mosaic, and puzzle mosaic image [11] [12]. The first two types of mosaics decompose a source image into tiles (with different color, size and rotation), reconstructing the image by properly painting the tiles. So they can be grouped together under the denomination of tile mosaics. The last two kinds are obtained by fitting images from a database to cover an assigned source image. Hence they may be grouped together under the denomination of multi-picture mosaics [12]. This taxonomy should not be intended as a rigid one. Many mosaic techniques may fit in more than a single class and it is likely that other new types of mosaics will appear in the future [13]. In this research, steganography and the use of the theory of mosaic as a method of hiding have been tackled. In section 2 there is a reference to some previous researches in the same field. In sections 3,4, and 5 the used images in this research have been formed and chosen to be primarily processed in order to be used in section 6. The results and conclusions of the next sections are to be discussed.

Related works
In 1996 Silvers R. created photomosaic based on the same property [14]; he wrote a program that starting from a data base of small photos called tiles images. and Finkelstein et al described methods for arranging a set of small images that we call tile images and adjusting their colors so that together they suggest a larger form [4]. Carlo B. and Clemente considered a particular class of pseudorandom images and the image mosaics description. [15]. Li Y. and Tsai W. [10] described the algorithm for mosaic image creation and secret image recovery. But Lai I. and Tsai W. defined mosaic image as a result of random rearrangement of the fragments of image [16] [17]. In 2013 Kede Ma finds algorithm for data hiding into an image by reserving room before image encryption; this research shows that first enough space is reserved in the image after which it is converted into encrypted form [18].
Accordingly, we propose in this paper a method that creates secret-fragment visible mosaic images; any image may be selected as the target image for a given secret image. Selecting a target image randomly, the provided secret image is divided firstly into fragment that are rectangular, which then become similar blocks in the target image according to a similarity criterion based on color variations. Then, the color characteristic of each tile image is transformed to be that of the corresponding block in a target image, resulting in a mosaic image which looks like the target image. Such a type of camouflage image can be used for securely keeping of a secret image in disguise of any pre-selected target image. Relevant schemes are also proposed to conduct nearly-lossless recovery of the original secret image. There are various examples of these study as shown in figure 1 [4] and figure 2 [15].
In 2016 Mukund R. Joshi and Renuka A. Karkada [19] shows a method that is based on secret fragment visible mosaic image which includes two phases: mosaic image creation and secret image recovery, this method is shown in the following block diagram:   2. Images: The target image contains personal image of a boy figure (4) and images that Stand for secret images which consist of two satellite images and pepper image. The first satellite image represents one of the region in Nenawa Governorate (Elan mountain) north west Mosul taken by ETM which was borne by landsat7-figure(5) and the second is the region of red sea taken by TM which was borne by landsat5-figure (6). the last image represents pepper in figure (7).
Elan mountain Red Sea pepper 3. Mean squared error (MSE) of an estimator (of a method for estimating a seen quantity) computes the average of the error squares or deviations that shows the difference between the estimator and what is estimated. The MSE is considered a non-negative, and the better value is the one close to zero [20]. If Y is a vector of n predictions, and Y is the vector of the observed values of the variable being predictor is computed as …………… (3) i.e., the MSE is the mean of the squares of the errors . This is an easily computable quantity for a particular sample.
4. Signal-to-noise ratio (with the abbreviations SNR or S/N) is used to measure the comparison of the ratio of a desired signal to the ratio of background noise. It is mainly used in science and engineering. S/N ratio is defined as the ratio of signal power to the noise power, often expressed in decibels. If the ratio is higher than 1:1 then it indicates more signal than noise. Signalto-noise ratio is sometimes used metaphorically to refer to the ratio of useful information to false or irrelevant data in a conversation or exchange. For instance, in forums of online discussions and other online websits, off-topic posts and spam are considered as "noise" that interferes with the "signal" of appropriate discussion.
Signal-to-noise ratio is defined as the ratio of the power of a signal (meaningful information) and the power of background noise (unwanted signal): ………………….. (4) where P is average power. Both signal and noise power must be measured at the same or equivalent points in a system, and within the same system bandwidth [21].

Primary processing:
These images in section (3)

Algorithm of creating mosaic image
To create image mosaic, two images must be identified: first tile image, second: target image applying the following steps: 1. Choosing target image. 2. Choosing tile image. 3. Finding a method to organized tile image depending on its structure or texture. 4. Correcting tile image to be match with the target image.
Based on the above steps, detailed algorithms for mosaic image creation and secret image recovery [10]. Input: a secret image S with n tile images of size NT; a pre-selected target image T of the same size of S; and a secret key K. Output: a secret-fragment-visible mosaic image F. 4. Do similarly to the last step to compute the means ( μr′, μg′, μb′), the standard deviations ( σr′, σg′, σb′), and the average standard deviation σBj = ( σr′ + σg′ +σb′)/3 for each Bj in Starget where j = 1 through n. 5. Sort the blocks in Stile and Starget according to the average standard deviation values of the blocks; map in order the blocks in the sorted Stile to those in the sorted Starget in a 1-to-1 manner; and reorder the mappings according to the indices of the tile images into a mapping sequence L of the form of T1 → Bj1, T2 → Bj2, etc. 6. Create a mosaic image F by fitting the tile images of secret image S to the corresponding target blocks of target image T according to mapping sequence L. figure (9) shows the flowchart of algorithm for mosaic image creation which hides the data in the target image.

Results
After the previous algorithm has been executed, the results represented by a personal image is obtained. The image includes satellite data as shown in figure (11) Observing the results, it is obvious that the image whose tiles are segmented by 8*8 gives a higher resolution than the one which is segmented 16*16. Mere observation is not enough to measure clarity and fussiness of each image and finding the optimal image in the embedding. Two measures have been used in this research, the first to measure the ratio of error in the image(MSE). The second measure is used to measure the noise (SNR) as follows: The results obtained in the two measures reveal a ratio of error and fussiness in both images with a slight difference. This fussiness is considered natural because the image is a number of tiles representing the original image. However, there is a slight difference among the resulting ratios as the tiles are taken from images that differ in their textures and grey levels. Concerning the images of different grey levels (peppers), they have higher degree of error and fussiness than the image which has a little grey levels since the tiles do not include all grey levels of the original image (image of Elan mountain). Besides, an image of small tiles (8*8) provides less error ratio than that of the bigger tiles (16*16).

Conclusions
In this paper we have presented a novel approach to image embedding in images. We have identified a class of images that are suitable for this task, these images are mosaics, which can be used, not only for secure keeping of secret images but also can be used as a new option to transport the secret images safely behind cover images as a personal photo. A novel algorithm has also been proposed for searching the tile images in a secret image for the most similar ones to fit the target blocks of a selected target image.