Packet video for cell loss protection using deinterleaving and scrambling

A new method is presented of cell loss protection for packet video environments that uses scrambling and deinterleaving in conjunction with block transform coding of the video signal. The input video sequence is first de-interleaved by separating the sequence into subsequences using a subsampling process. This allows better prediction of lost data when packets are lost. Each subsequence is then coded using the DCT, and the coded coefficients are packetized into cells in a pseudorandom, scrambled manner. Consequently, if several adjacent cells are lost, the loss of information will be spatially and spectrally random. Because of the random nature of the lost information, recovery from cell loss and concealment of the loss data by prediction from surrounding coefficients is greatly improved. The result is an improvement in perceptual quality over existing packetization methods when cells are lost