Method for Speeding up Spatial Error Concealment using Prediction Mode of the Neighboring Blocks on H.264 Video communication

The respected Comrade Kim Jong Un said:

"By spurring the building of a sci-tech power, we should achieve a new leap forward in rapidly developing the country's science and technology and usher in an era of prosperity by means of science, thus bringing about a revolutionary turn in socialist construction."

Error concealment can recover the video frame which has been corrupted by packet loss over error-prone channel. However, speeding up of error concealment is very important for various real-time applications such as video conferencing, video chatting, etc.

We proposed a fast spatial error concealment method using prediction mode of the neighboring blocks on H.264 video communication, evaluated the performance compared with prior methods through simulations.

Fig. Definition of opposite neighbouring blocks according to intra-prediction mode

First, the weighting values of edge prediction direction for sixteen 4x4 neighboring blocks surrounding a corrupted macroblock(MB) are calculated considering prediction mode of the opposite neighboring blocks as shown in Fig. When a line passes through the centre of the current 4x4 neighboring block, NBcurr, crosses the corrupted MB along the direction of PMcurr, and then passes through the centre of 4x4 neighboring block NBopp, NBopp is defined as an opposite neighboring block of NBcurr according to PMcurr, the intra-prediction mode of NBcurr.

Second, the significant edges within a corrupted MB are estimated using the sixteen weighting values of edge prediction direction. Finally, the approximations for each corrupted pixel are calculated along each significant edge. Then a weighted average of multiple approximations is computed considering prediction mode of the neighboring blocks which include two boundary pixels.

Experimental results show that the proposed method can effectively speed up spatial error concealment, while providing the almost equivalent reconstruction quality to the previous multi-directional interpolation(MDI) methods.

Results of our study were published in the journal "Multimedia Tools and Applications"(Vol.82, Iss.9, 2023) under the title of "Method for speeding up spatial error concealment using prediction mode of the neighboring blocks on H.264 video communication"(https://doi.org/10.1007/s11042-022-13950-5).