Title :
A low-power encoder architecture for pyramid vector quantization of 2D subband coefficients
Author :
Namgoong, Won ; Devenport, Mark ; Meng, Teresa
Author_Institution :
Comput. Syst. Lab., Stanford Univ., CA, USA
Abstract :
This paper describes a VLSI architecture designed to pyramid vector quantize (PVQ) 2-D subband image for low-power portable applications. Using the 0.8 μ CMOS process, the chip is estimated to dissipate less than 300 μW at a 1.5 V supply, encoding at 1.27 Mpixels/sec, for display of 240 pixels wide, 176 lines, and 30 frames per second, while introducing no noticeable degradation in the decompressed video quality. Low power consumption is achieved through efficient algorithm-to-hardware mapping requiring no multipliers, dividers, or external memory accesses, innovative computational methods, and reduction in supply voltage
Keywords :
CMOS digital integrated circuits; VLSI; image coding; vector quantisation; 0.8 micron; 1.5 V; 240 pixel; 2D subband coefficients; 2D subband image; 300 muW; CMOS process; VLSI architecture; algorithm-to-hardware mapping; decompressed video quality; low-power encoder architecture; low-power portable applications; pyramid vector quantization; Application software; Computer architecture; Energy consumption; Entropy; Image coding; Laboratories; Portable computers; Rate-distortion; Vector quantization; Very large scale integration;
Conference_Titel :
VLSI Signal Processing, VIII, 1995. IEEE Signal Processing Society [Workshop on]
Conference_Location :
Sakai
Print_ISBN :
0-7803-2612-1
DOI :
10.1109/VLSISP.1995.527510