Resilient Video Coding via Improved Motion Compensated Prediction

Background

Error Resilient (ER) video coding approaches can be classified into 3 broad categories: encoder, decoder and encoder/decoder co-operation techniques, which are summarized below,

• 1.

  Encoder: These methods involve adding redundancy at source coder, channel coder or both. Bitstream prioritization (Turletti & Huitema, 1996; Zhu, 1997; Sun & Zdepsky, 1994), and Forward Error Correction (FEC) (Chou, Mohr, Wang, & Mehrotra, 2001; Tan & Zakhor, 2001; R. Zhang, Regunathan, & Rose, 2001) fall into this category. Also some interesting work has been done on error resilient techniques that exploit channel characteristics. The main technologies are based on path diversity (Apostolopoulos, Wong, Tan, & Wee, 2002; Nguyen & Zakhor, 2002; Padmanabhan, Wang, & Chou, 2003), network coding (Chou, Wu, & Jain, 2003; Wu, Chou, & Kung, 2005) and cross-layer design/optimization (Wu, Chou, Zhang, et al., 2005; Setton, Yoo, Zhu, Goldsmith, & Girod, 2005).

• 2.

  Decoder: These methods are also known as error concealment.Despite the use of encoder techniques to protect the video bitstream, some errors or losses may escape/penetrate encoder protections and cause a perceptual degradation of received video quality. Hence it is necessary for the decoder to perform error concealment. Most of these techniques exploit either spatial (Aign & Fazel, 1995; Wang, Zhu, & Shaw, 1993; Sun & Kwok, 1995) or temporal correlations (Aravind, Civanlar, & Reibman, 1996; Wang & Zhu, 1998; Wang, Wenger, Wen, & Katsaggelos, 2000; Lam, Reibman, & Liu, 1993; Lu, Lieu, Letaief, & Chuang, 1998) in order to predict the erroneous pixels.

• 3.

  Encoder/Decoder: ARQ methods fall into this category (Podolsky, McCanne, & Vetterli, 2001; Chou & Miao, 2006; Miao & Ortega, 2002). A better approach adjusts the encoder prediction upon receiving channel feedback to send a correcting signal that is able to update the decoder prediction to match that in the encoder (Girod & Farber, 1999; Chang & Lee, 2000). These methods may not be suitable for low delay applications such as video telephony.