Title :
Compact inverse discrete cosine transform circuit for MPEG video decoding
Author :
Hung, Ghing-Yu ; Landman, Paul
Author_Institution :
DSPS R&D Center, Texas Instrum. Inc., Dallas, TX, USA
Abstract :
The discrete cosine transform (DCT) and its inverse transform (IDCT) are essential computations in many signal processing applications. IDCT, specifically, is used to decompress MPEG video bitstreams and, therefore, needs to be computed inside cost-sensitive end-user units. This paper presents a very compact IDCT design based on digit-serial arithmetic techniques and the even/odd decomposition algorithm. This IDCT design has been synthesized in a 0.35 μm gate array technology, and requires 4,600 gates plus 1,024 bits of memory. It processes 26 million pixels/sec, and has been simulated to meet the precision and dynamic range requirements of MPEG. Because of its digit-serial construction, the design can be scaled to produce efficient circuits to meet different data width and processing rate requirements
Keywords :
decoding; digital arithmetic; digital signal processing chips; discrete cosine transforms; video coding; MPEG video decoding; compact inverse discrete cosine transform circuit; digit-serial arithmetic techniques; digit-serial construction; even/odd decomposition algorithm; gate array technology; processing rate requirements; signal processing; Algorithm design and analysis; Arithmetic; Circuits; Decoding; Discrete cosine transforms; Discrete transforms; Signal processing algorithms; Transform coding; Video compression; Video signal processing;
Conference_Titel :
Signal Processing Systems, 1997. SIPS 97 - Design and Implementation., 1997 IEEE Workshop on
Conference_Location :
Leicester
Print_ISBN :
0-7803-3806-5
DOI :
10.1109/SIPS.1997.626269