Error concealment using data hiding for resolution scalable transmission

摘要

In this paper, a novel error concealment scheme using data hiding and a modified version of the HS-SPIHT algorithm, called MHS-SPIHT, is proposed. At the transmitter side, the coefficients of wavelet decomposed image are partitioned into perfect trees. The MHS-SPIHT algorithm is applied to encode each perfect tree independently. The output bitstream is then reordered to generate a properly tailored code as reference code. The reference code consists of different code-parts related to a certain spatial resolution level. This reference code is embedded into the coefficients of another perfect tree which is located in different place. Each part of the reference code that belongs to a certain spatial resolution is embedded into the coefficients of the host perfect tree at the same spatial resolution. At the decoder side, if a part of the requested image by the receiver is lost, the decoder extracts the appropriate reference code. This code is concealed the perfect trees related to lost area at resolution level requested by the decoder. Simulation results show that the proposed method can achieve a considerable improvement for resolution scalable image transmission over unreliable channels such as wireless channels.