A Data Hiding Method Based on Partition Variable Block Size with Exclusive-or Operation on Binary Image

Abstract

In this paper, we propose a high capacity data hiding method applying in binary images. Since a binary image has only two colors, black or white, it is hard to hide data imperceptible. The capacities and imperception are always in a trade-off problem. Before embedding we shuffle the secret data by a pseudo-random number generator to keep more secure. We divide the host image into several non-overlapping (2n+1) by (2n+1) sub-blocks in an M by N host image as many as possible, where n can equal 1, 2, 3 , …, or min(M,N). Then we partition each sub-block into four overlapping (n+1) by (n+1) sub-blocks. We skip the all blacks or all whites in each (2n+1) by (2n+1) sub-blocks. We consider all four (n+1) by (n+1) sub-blocks to check the XOR between the non overlapping parts and center pixel of the (2n+1) by (2n+1) sub-block, it embed n 2 bits in each (n+1) by (n+1) sub-block, totally are 4*n 2 . The entire host image can be embedded 4×n 2×M/(2n+1)×N/(2n+1) bits. The extraction way is simply to test the XOR between center pixel with their non-overlapping part of each sub-block. All embedding bits are collected and shuffled back to the original order. The adaptive means the partitioning sub-block may affect the capacities and imperception that we want to select. The experimental results show that the method provides the large embedding capacity and keeps imperceptible and reveal the host image lossless.