Paper Title
Error Concealment for Stereoscopic View Plus Depth Based on K Nearest Neighbor Method

Abstract
This paper proposes a novel hybrid concealment scheme to recover packet loss on a slice during stereoscopic color plus depth or view plus depth transmission over error prone channel. The proposed scheme aims at minimizing quality degradation on both color and depth maps sequences. The scheme uses two resilience tools in video coding H264/AVC, which are slice groups also known as Flexible Macro-block Ordering (FMO) and Redundant slices (RS). FMO adds flexibility to control errors that are spread spatially. This paper also exploits the K nearest neighbors (KNN) regression method to conceal missing packets. Initial results have shown that the proposed scheme significantly minimizes quality degradation for 3D video transmission over error prone channels. Index Terms— View plus depth, concealment scheme, KNN, slice groups (FMO), redundant slices (RS) , H264/AVC, subjective quality.