Adaptive error concealment for Multiple Description Video Coding using motion vector analysis

Adaptive error concealment for Multiple Description video coding using error estimation

“…All the experiments are run 100 times with different permutations of lost packets for each PLR (5%, 10%, 15% and 20%) and the results represent the average of the PSNR obtained from these different loss patterns. The performance of our proposed method (represented by "Proposed") is compared with two other methods: the adaptive concealment method based (denoted as "Old-ad") presented in [9] and the non-adaptive method (denoted as "Spatial") presented in [6]. Figure 4 shows the comparison for the "Bridge-Close" sequence.…”

“…LetγM l be the estimated average motion of a lost MB M l as computed in [9]. Changing the notation slightly to accommodate FRAMETYPE, letγj,i be the estimated average motion of the ith temporally concealed MB of FRAMETYPE j (jǫ{P, B}).γj,i is used as a measure of MOTM.…”

“…The MDC Architecture We use the partition architecture presented in [6,8,9]. The original video sequence is first temporally subsampled by taking evennumbered and odd-numbered (starting from 0) frames to form two descriptions, which are known as the Even and Odd sequences respectively.…”

“…Both of the above methods require a reliable transmission of side information which also adds to the data rate. In [9], an adaptive method based on This work was supported by the endowment of the Charles William Harrison Distinguished Professorship at Purdue University. Address all correspondence to Edward J. Delp, ace@ecn.purdue.edu motion vector analysis is developed.…”