Many approaches have been used to the purpose of image data compression. Each of them combines various techniques to exploit the redundancy of the image signal.
In this paper two different approaches will be presented, analysed and compared. They have been developed by the CNR-CSTV and the RAI Research Centre(1) respectively, in the framework of subproject 3.2 of the CNR «Telecomunicazioni» project. The goal is to define a coding scheme suitable for the bit-rate compression of digital HDTV signals in order to cope with the characteristics of two kinds of channels: a synchronous channel at 70 Mbit/s and an asynchronous one with peak bit-rate of 140 Mbit/s. The coded images must have a good quality for distribution networks.
Both algorithms are based on the hybrid transform image coding using the 2D-DCT to reduce the spatial correlation. In order to further decrease die temporal correlation a motion compensation technique is also applied. The aim of this paper is to describe the two algorithms and to investigate the main differences.
[1]
Norman B. Nill,et al.
A Visual Model Weighted Cosine Transform for Image Compression and Quality Assessment
,
1985,
IEEE Trans. Commun..
[2]
ROGER C. WOOD,et al.
On optimum quantization
,
1969,
IEEE Trans. Inf. Theory.
[3]
Thomas R. Fischer,et al.
A pyramid vector quantizer
,
1986,
IEEE Trans. Inf. Theory.
[4]
C. Smith,et al.
Adaptive Coding of Monochrome and Color Images
,
1977,
IEEE Trans. Commun..
[5]
Joel Max,et al.
Quantizing for minimum distortion
,
1960,
IRE Trans. Inf. Theory.
[6]
Adrian Segall.
Bit allocation and encoding for vector sources
,
1976,
IEEE Trans. Inf. Theory.