Abstract
Recently, the application for transmission of the video stream over the network has been widely discussed. Through applying the latest video coding standards such as H.264 or MPEG4, the video stream has reached high compression rate for real time display. However, in the error-prone channel, the transmission error caused by packets lost or delay brings error in the decoding process, which produce not only the decrease of visual quality but also error propagation problem. Therefore, in order to recover the corrupt data without interfere with the encoder side and the bandwidth; the error concealment techniques which utilize the spatial and temporal relativity are necessary. This paper is going to introduce a temporal and spatial error concealment structure, each side have several concealment strategies. Through game theory, this structure will find out the best concealment strategy to deal with the error circumstance in different image area. In order to reach better edge alignment in temporal error concealment, we proposed a boundary matching error measurement which takes edge information into account in the motion vector searching process, which helps us to choose a better macro block in reference frame. The thesis also proposed a de-blocking scheme to reduce the blocking effect.