A vector coding method for natural images

Author

Osada, Reiko ; Aoki, Terumasa ; Yasuda, Hiroshi

Author_Institution

Graduate Sch. of Eng., Tokyo Univ., Japan

fYear

2000

fDate

2000

Firstpage

146

Lastpage

150

Abstract

A new vector coding for natural image, which has a compressive algorithm, is presented and the image features showed in this method are described. The pixel-value-boundary contours on RGB or YCrCb planes of natural images are expressed by function approximation and then their contour shapes are predicted from similar contours in the decoder side. The proposed vector coding method has two main merits: one is to meet the demands of expansion and reduction of images, the other is to suit content distribution because the image quality can be kept even under the repetition of editing due to representation of function approximation. To put vector coding for natural images to practical use, we compress the data by using the characteristics in three-dimensional representation of images

Keywords

data compression; function approximation; image coding; image representation; vector quantisation; function approximation; image coding; image compression; natural image; pixel value boundary contours; vector coding; Decoding; Digital images; Frequency; Function approximation; Image coding; Image segmentation; Pixel; Predictive models; Shape;

fLanguage

English

Publisher

ieee

Conference_Titel

Applied Imagery Pattern Recognition Workshop, 2000. Proceedings. 29th

Conference_Location

Washington, DC

Print_ISBN

0-7695-0978-9

Type

conf

DOI

10.1109/AIPRW.2000.953617

Filename

953617