Interpolation with Sigmoid Functions for Spatial Error Concealment

Error concealment methods for Intra frames in block-based video systems reconstruct the missing macroblock often by computing weighted average of the boundary pixels of the neighboring blocks. However, the simple averaging of pixel values leads to blurring and degrades the picture quality severely. Directional interpolation, in which the interpolation is performed along the possible edge direction, has been proved to alleviate the problem in some cases. However, still directional approach results in degraded performance when edges are not clear. First, we propose weighted interpolation with the sigmoid function which gives more weight to closer values, so overall blurring is reduced considerably. Secondly, an adaptive approach is presented in which directional interpolation is chosen by a threshold computed from the distribution of local directions around the lost block. Experiments show improvement of picture quality of about 0.5-2.0 dB compared to existing methods.