Dresden China Figurines

The most basic way to hide information in images is by using the method LSB (Least Significant Bit English, English bit less significant). Unfortunately not know the reference of the first work which mentions the use of this method, if any of you know please help me to know.

This method involves replacing the least significant bits each pixel with the bits of information to hide.

As the best way to learn is by example, then we see this method in action. Take

carrier medium Lena image, the message will hide 2048 bits in length and will follow the following pattern:

Message = 11111111 00000000 11111111 00000000 ...

Lena image, used to hide the message

As mentioned in the entry some images, the images are formed by a depth of color, is the case Lena image is an image of 8 bits. That is, each pixel is represented with 8 bits, which causes the image is 256 colors (in this case is in gray scale) ranging from black (0 in decimal or in binary 00000000 ) to white (255 in decimal or in binary 11111111 ).

The least significant bit (lsb in lower case or to refer to the least significant bit and not the method for hiding information) determines the parity of a number. The lsb is also known as the rightmost bit . Against part or most significant bit msb represents the most bit left. For example in the following image shows the binary representation of decimal number 211 , the lsb is marked with orange and purple color lsb. The msb represents a decimal value of 128 lsb and a decimal value of 1 .

Binary representation of decimal number 211 , the lsb is marked with orange

The plane lsb is 1-bit binary image formed by all the lsb of the image. The following image depicts the LSB plane of the image of Lena.

lsb of Plano Lena image

Now for the other planes binary image of Lena.

Plans binary image of Lena

LSB method is to modify the LSB plane of the image carrier with the brand to hide. Thus, if we replace lsb bits of the plane row by row with the brand mentioned above (the pattern of eight zeros and eight ones) resulting lsb level will appear as shown in the picture below.

lsb Plano marked

The watermarked image, ie the image with the modified lsb plane is shown below.

marked Lena image using the LSB method

is important to note that if we see the two images can not perceive Jutas which is marked and what is the original, since it only changed the value of the pixels in a unit and this happens at the worst. The following image shows the original Lena image, the watermarked image and finally the absolute difference of the two images. In this last image the white pixels represent the pixels that changed in magnitude to be marked.

difference between the original Lena image and marked

These changes made using the LSB method are imperceptible to the human eye, however if the level is plotted lsb can easily notice that "something strange in the picture" and not for the fact that we have used a clearly visible mark above. You might think that the plane has a random nature lsb, ie not have a pattern like the other planes (ie the bit planes 4,5,6,7 and 8) but the reality is different, the plane if you have lsb a pattern that depends on the image, but at first glance not feel it. For example the following figure shows a flat image and lsb. Here you may notice a pattern to the naked eye at the lsb.

Pattern en el plano lsb de la imagen Bird

Para evitar que la marca sea detectada a simple vista (esto se considera un ataque visual, como se mencionó en la entrada clasificaciones de las marcas de agua ), muchos diseñadores no la esconden linealmente como lo hicimos en el ejemplo descrito, sino la esconden en posiciones dadas por un generador de números pseudoaleatorios, así la esparcen en toda la imagen incrementando la indetectibilidad.

En la siguiente entrada analizaremos más a fondo este método LSB y de igual forma mostraremos unas modificaciones del mismo. Por ejemplo, la modificación para ocultar una imagen dentro de otra, la modificación para detectar cambios hechos en marked image (integrity) and can be modified to recover the original image of the watermarked image (reversible method).

Be well and we are in contact ...

PD. The source code (in Matlab) of the method explained in this entry is listed below:
1. LSB Method hiding a message in an image bmp newspaper.


LSB Steganography Method lsb Plano Images