Iterative Method
First initial the corrupted MB with neighboring MB information then use iterative techniques to conceal pixels.
Weighted Average Method
Simple way to conceal corrupted MB with the help of location and neighboring pixel’s value.
Directional Interpolation(Quadrilinear Border Interpolation)
With the help of edge detection, this kind of method is good for the picture full of edge information.
Hybrid Method
Use previous information and iterative.
BNM
Best neighboring matching. Get corrupted and good part , then set a conceal region in ref frame and search in the region, then get best matching by MSE. This method uses pixel information.
ECDF(error concealment deblocking filter)
Edge detection, Replacement, Smoothing, De-blocking filter
After edge detection, no using interpolation, just using neighboring pixels to conceal corrupted blocks. And as for deblocking, taking conceal mode into consideration.
Especially for smoothing, we can think of intra prediction.
Edge-Preserving Interpolation
Both using edge-preserving interpolation and texture analysis and synthesis, then choose between them.
MB Copy
Get corrupted MB’s neighbors and conceal the corrupted MB.
MB Mean Value Interpolation
All value of corrupted MB are concealed by the mean of neighbors.
Maximally Smooth Interpolation
Processing in frequency and deal corrupted MB with DCT coefficient.
