Fig 1.
Flowchart of all steps required for the data embedding in images.
Fig 2.
The index map, capacity and peak are computed for different values of δ using the Generate Index Map algorithm given in Algorithm 1 for an 8 × 8 gray-scale pattern image.
Table 1.
Image transformation calculated by switching the columns according to the index map having highest embedding capacity.
Table 2.
Difference image, histogram shifting and embedding data in the difference image.
Table 3.
Transformed image with embedded data and the production of the marked image using inverse-transform through index map.
Fig 3.
Flowchart explaining all steps in data extraction from marked image.
Table 4.
Converting marked image to transformed image.
Table 5.
Difference image with data, data extraction by peak values and histogram shifting in reverse.
Table 6.
Transformed image to original input image.
Fig 4.
A 512 × 512 checkerboard.
Fig 5.
Embedding of data in checkerboard image.
Table 7.
Comparison of maximum capacity and PSNR in different images achieved by proposed embedding and other state-of-the-art techniques.
(Cap. means embedding capacity in bits).
Fig 6.
Embedding capacity for different values of δ for checkerboard.
As noted we have high capacity when we have δ in multiple of 64.
Fig 7.
Image with multiple repeated large and small patterns.
Fig 8.
Performance comparisons between the proposed method and difference method for images given in Figs 5b, 7c and 7f respectively.
Fig 9.
Data embedding in three publicly available images.
Fig 10.
Performance comparison of embedding techniques with difference method of three publicly available images.
Fig 11.
Input images, transformed images and marked images.
Fig 12.
Comparisons of embedding capacity and PSNR in proposed and difference methods.
Fig 13.
Pixel Difference Histogram (PDH) for input images and marked images (part I).
Fig 14.
Pixel Difference Histogram (PDH) for input images and marked images (part II).