This technique was proposed by Khodaei & Faez (2012). The embedding procedure is mentioned in the following steps.
Step 1: The image is raster scanned and divided into 1×3 size non-overlapping blocks, Figure 1. The pixels are designated as , , and .
A pixel block
Step 2: In k-bit LSB substitution is applied, where k is chosen from the list {5, 6, 7, 8}. The new value of is . Let L be the decimal value of the k LSBs of and S be the decimal value of k-data bits hidden in . The value can be further optimized using Eq. (1), where d=L-S.
Step 3: Using .the following two differences are calculated as in Eq. (2).
Step 4: The Table 1 is range table for variant-1 and Table 2 is range table for variant 2.
Step 5: The value falls into a range whose embedding length is and lower bound is . Similarly, the value falls into a range whose embedding length is and lower bound is .
Step 6: From the secret binary data stream bits of data is taken and converted to a decimal value . Similarly, from the secret binary data stream bits of data is taken and converted to a decimal value . Two new difference values and are computed as in Eq. (3). + , + (3)
Step 7: Using and two new values for namely, and .are calculated. Similarly, using and two new values for namely, and are calculated using Eq. (4). = - , = + , = - , = + (4)
Step 8: Now the stego value of , say is calculated using and . Similarly, the stego value of, say is calculated using and . This is done using Eq. (5) and Eq. (6).
Hence the stego block is represented in Figure 2.
The stego-block
The extraction of secret embedded data can be done in the following manner. Suppose the stego-block is as shown in Figure 2 from which we have to extract the hidden data. The following steps are used.
Step 1: From the center pixel , k-LSBs are extracted. The value k is one of the values in the list {5, 6, 7, 8} which was chosen during embedding.
Step 2: Using and , the difference value is computed. Similarly, using and , the difference value is computed as in Eq. (7).
Step 3: The difference belongs to a range whose lower bound is and embedding length is . Similarly, the difference belongs to a range whose lower bound is and embedding length is . Refer the range table.
Step 4: As in Eq. (8), using and , calculate the decimal value and convert to binary bits. Similarly using and , calculate the decimal value and convert to binary bits. = , = (8)