Error concealment for JPEG2000 images based on orthogonal edge directed filters

We propose an error concealment algorithm for JPEG2000 image transmissions over unreliable channels. Firstly, the local principal edge information in the damaged area is detected in the spatial domain. Then the proposed orthogonal edge directed filters (OEDFs) are applied to remove the ring or ripple artifact errors due to the loss of some wavelet transform (WT) bitplane data. Two kinds of constraints in WT domain are used for rectifying the recovered WT coefficients obtained from OEDFs, namely the WT known-value constraint and the empirical statistical constraint of the WT coefficients. Finally, this filtering-and-rectifying procedure is iterated until convergent. Simulation results have shown that both objective and subjective image quality have been improved by our proposed algorithm.