A Binary Wavelet Transform Based Authentication Algorithm for Binary Images

We propose a novel block-wise authentication algorithm for binary images. We partition an image into tree structured, mutually exclusive, and collectively exhaustive blocks. The size of each block depends on the number of flippable pixels in the block. Flippable pixels are found in the binary wavelet transform domain. A random bit sequence generated by the block identifying key is embedded into the flippable pixels in the block. Block-wise localization of tampered region is accomplished by verifying the random bits. While previous research results show partial imperfect localization for special cases, our results do not show those problems.