A new scheme of image data hiding based on EBCOT of JPEG2000 lossy compression

摘要

In the EBCOT stage of JPEG2000, data losses are incurred twice from an image. Bits are discarded during quantization and from entropy coding. Therefore, a new scheme is proposed for large data images. Data losses are avoided by hiding data in the compressed bitstreams after EBCOT. Layer optimization is used to increase the length of bitstreams by repeatedly readjusting a fake bitrate br to force the reassignment of bits in the last bitstream layer. All bitstreams in the code-blocks are forced to repeat the rate allocation process for reassignments at the new bitrate. Bitstreams are then repacked into new output layers. The rate allocation algorithm calculates every output layer and its estimated length. Bitstreams are extracted from the code-blocks of the output layers and replaced into the related recalculated output layers. Experimental results proved that data hiding capacity could be adjusted as needed with very little computation required. The proposed scheme also works well for complex images.