Information hidden in color images (2)

The basic principle of 3 bitmap method

3.1 Information Embedding

Information embedding adds information to be hidden to the cover media. The general process is shown in Figure 1:

Input Generates the information to be hidden M----- Hide algorithm FUNC------ Disguised media B of hidden information

↓
Coverage Media A

Figure 1 General process of information embedding

The above process can be expressed by the formula =Func(,A) The screening media A can be an image, a sound, etc. As long as the difference between A and the perception of people is small, Func can be considered as a good hiding algorithm. In this article, the cover media A is a 24-bit BMP image that can be any file. In order to make A and the difference is not large, choose the method to modify the lowest bit (LS) of the BMP image, because doing so can minimize the difference between A and A and will not cause too much distortion.

A 24-bit BMP image consists of a 54-byte header and image data. The header cannot hide information. After the 55th byte is the image data part, information can be hidden. The image data part is composed of a series of 8-bit binary numbers. Since the number of "1"s in each 8-bit binary number is either odd or even, the convention is: if the number of "1" in a byte is For odd numbers, the byte is said to be an odd byte, denoted by "1"; if the number of "1" in a byte is an even number, the byte is said to be an even byte, denoted by "0". We use the parity of each byte to represent the hidden information.

Example: Set the data of a 24-bit BMP file as, 00111100, 10001111,00011010,00000000, 10101011, 00111110, 10110000, then the parity of its bytes is: 0,0,1,1,0,1,1,1 . Now need to hide the information 79, because 79 is converted to 8-bit binary is 01001111, the two sequences are compared and found that the second, third, fourth, fifth inconsistency, so some of the bytes of this 24-bit BMP file data The parity is modulated to match the 79 converted 8-bit binary:

Bit 2: When 00111100 is changed to 00111101, the byte changes from even to odd;

Bit 3: change 10001111 to 10001110, then the byte changes from odd to even;

Bit 4: Change 00011010 to 00011011, then change the byte from odd to even;

Bit 5: Change 00000000 to 00000001, then the byte changes from even to odd;

After such modulation, the parity of the data segment byte of this 24-bit BMP file is exactly the same as that of the 79-transformed 8-bit binary number, so that 8 bytes conceal one byte of information.

To sum up, the steps to embed information into a BMP file are:

I. Convert the information to be hidden into a binary data stream;

II. Compare the parity of each byte of the image data portion of the BMP file with the above binary digital stream;

III. By adjusting the "0" or "1" of the least significant bit of the byte, the parity of the byte is changed so that it is consistent with the above binary data stream, that is, the information is embedded in the 24-bit BMP image.

3.2 Information Extraction

The information extraction is to read the hidden information from the disguised media, and the process and steps are exactly the opposite of the information embedding:

I. Determines the parity of each byte of the image data part of the BMP file. If the number of "1"s in the byte is an even number, it outputs "0"; if the number of "1"s in the byte is an odd number, it outputs "1". ";

II. Every 8 bytes are judged, the output 8 digits form a binary number (the first output is high);

III. After the above processing, a series of 8-bit binary numbers is obtained, which is a code for hiding information and converting the code into text, or an image, or sound, which is hidden information.

4 Conclusion

1 Since the original 24-bit BMP image file hides information, its byte value changes by at most 1 (because the lowest bit of the byte is added "1" or minus "1"), the color concentration represented by this byte only changes at most 1/256, so BMP images with hidden information and BMP images with no hidden information are not visible to the naked eye;

2 The advantage of embedding information directly into the pixel RG value is that the capacity of the embedded information is proportional to the size of the selected screening image, and is no longer limited to the size of the palette.

3 With this method, a 24-bit BMP image file with a size of 32 k can hide information about 32 k/8 = 4 k (ignore 54 bytes where the file header cannot hide data). This method has a higher value. Information hiding rate.




Liu Cai, Du Shipei

(College of Computer Science and Information Technology, Guizhou University of Technology;