Error Concealment for Video

The main purpose of error concealment schemes for video is to limit the impact of loss over the image quality. These schemes take advantages of the redundancy present in the image in the frequency, spatial, and temporal domains. Under these schemes part of the lost information can be recovered by regenerating most of it by means of interpolation in each of these domains [34]. From the literature [52,158,34], error concealment techniques for video codecs are as given below.

$\bullet$

(i)

Block Replacement - The simplest method to interpolate the lost areas is to replace them by the corresponding areas of the previous frame or field. This temporal extrapolation works quite well in still parts of the pictures (e.g., backgrounds) but fails in areas where there is a lot of motion.

(ii)

Linear Interpolation - It is based on the Intra-frame linear interpolation that replaces the lost areas with the linearly interpolated values, calculated from the neighboring areas of the same frame. This method works quite well in a uniform surface with smooth gray-level changes inside the block. This is provided that the surrounding areas are correctly received. If this is not true, which is the case in networks, this methods does not give good results.

(iii)

Motion Vector - The lost areas are replaced with pixel blocks of the previous frame shifted by the average motion vector of the neighboring blocks. The picture quality is greatly improved, but the performance drops when the surrounding blocks have different motion vectors.

(iv)

Hybrid Technique - Concealing by simple motion vector estimation and simply averaging the top and bottom correctly received MBs method gives better results. In addition, concealing by more than one motion vector and averaging all the neighboring MBs can do an even better job but of course at the cost of more computational complexity and delay. The hybrid techniques use both spatial redundancy and more than one motion vectors information to predict the lost MB.