Abstract
In this thesis, we propose new methods to reduce the side effect of predictive coding both in facilitating VCR functionality and error recovery while maintain coding efficiency. In particular, two methods are introduced. First, for facilitating VCR functionality, a new I frame reduction method is demonstrated in which an I-P frame conversion is made to preserve coding efficiency while a small frame set is maintained to help provide the random accessing functionality. Second, for error recovery, we propose a new error concealment method for use with a layered video coding. In this method, the relationship between the previous base and enhancement layer is used to help recover lost or damaged pixels.