Title :
Image coding by using trimming modified vector quantization
Author :
Yu, Ping ; Venetsanopoulos, Anastasios N.
Author_Institution :
Dept. of Electr. & Comput. Eng., Toronto Univ., Ont., Canada
Abstract :
A new coding technique, the trimming vector quantization (TVQ) is introduced in this paper, which provides a low coding bit rate as well as high reconstruction quality to image compression. It is well known that in conventional VQ and many other image coding techniques, if the bit rate is lower than 0.2 bit/pixel, the quality of the encoded image may dramatically decline due to the artificial block effects and/or blurry edges. The TVQ may overcome those disadvantages by trimming a coarsely encoded image, removing the artificial block contours and sharpening the major edges, based on the features of the different image regions. By using the TVQ we can encode an image at a compression ratio higher than 50 and the reconstructed image still has a satisfactory perceptual quality with SNR=31 dB
Keywords :
image coding; image reconstruction; vector quantisation; artificial block contours; blurry edges; coarsely encoded image trimming; compression ratio; edge sharpening; encoded image quality; high reconstruction quality; image coding; image compression; image regions; low coding bit rate; perceptual quality; reconstructed image; trimming vector quantization; Bit rate; Computational complexity; Data compression; Image coding; Image decomposition; Image reconstruction; Vector quantization;
Conference_Titel :
Electrical and Computer Engineering, 1993. Canadian Conference on
Conference_Location :
Vancouver, BC
Print_ISBN :
0-7803-2416-1
DOI :
10.1109/CCECE.1993.332257