Frequency scalable video coding using the MDCT
Alan W. Johnson, John Princen, M.H. Chan · 2002
Efficient video coding incorporating motion compensation and overlapped transforms requires different coding architectures and methods of motion estimation to those used in traditional DCT block based coding schemes. In this paper a transform domain motion estimation procedure is proposed for the MDCT. This method of motion estimation, while performed in the spatial domain, is equivalent to minimising the sum square error of transform coefficients. The application of the MDCT to frequency scalable video coding is investigated. A performance advantage is achieved over an equivalent DCT based coder for both the reconstructed low resolution and high resolution sequences of an open loop frequency scalable video coder.>