Digital Image Steganography With Error Correction on Extracted Data

• Published in: IEEE access Vol. 11; pp. 80945 - 80957
• Main Authors: Kosuru, S. N. V. J. Devi, Pradhan, Anita, Basith, K. Abdul, Sonar, Reshma, Swain, Gandharba
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Correlation; Data hiding; Data mining; Digital images; Digital imaging; Distortion; Error correction; Error correction & detection; Error detection; Hamming codes; Hamming distances; modified Hamming code; Pixels; Quotients; QVD; Receivers; Redundancy; Security; Sonar; Steganography; Substitutes
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2023.3300918

The hiding capacity (HC), imperceptibility, and security are the 3 important quality measures for a steganography technique. While the stego-image is on transit on the internet, the hidden data may be changed because of various reasons. The existing techniques does neither focus on detecting the errors in the data nor to correct the errors in data. Therefore, this article brings forward a steganography technique, wherein error detection and correction can be performed at recipient side. The original image is logically sliced into <inline-formula> <tex-math notation="LaTeX">2\times 2 </tex-math></inline-formula> disjoint blocks. From these 4 pixels, 4 quotients and 4 least significant bits (LSBs) are generated. Each quotient is the decimal value of 7 most significant bits (MSBs) of a pixel. In every block 8 data bits can be camouflaged. From the 8 data bits, 4 redundant bits are computed using modified Hamming code. The 8 data bits and one redundant bit are camouflaged in the quotients by either quotient value differencing (QVD) or bit substitution. If camouflaging is performed in quotients using QVD, then indicator bit is set to1. Otherwise, if camouflaging in quotients is performed using bit substitution, then indicator bit is set to 0. The 3 remaining redundant bits along with the indicator bit are stored in the LSBs of the 4 pixels. At the receiver side, data could be extracted, and error correction procedure could be applied to correct 1-bit error over the 8 bits of data extracted from a block. From the experimental reports it could be concluded that the errors in the retrieved data at the recipient can be detected and corrected without reducing the HC and without increasing the distortion.