A new image segmentation approach based on linked pyramid

Author

Ping, Yin ; Runsheng, Wang ; Diannong, Liang

Author_Institution

Dept. of Electron. Tech., Nat. Univ. of Defense Technol., Hunan, China

Volume

2

fYear

1996

fDate

14-18 Oct 1996

Firstpage

1118

Abstract

Pyramid-based image segmentation is a kind of very precise, universal and efficient method. A new image segmentation approach based on this idea is presented. A new way to build a pyramid of an image and a new strategy of identifying the root nodes are developed. In this approach: (1) the Gaussian filter function with changeable filter scales is used to build a pyramid, so as to decrease the loss of the original image information; (2) the relationship among the pixels which are in both level-in and level-between is used to determine the root nodes. The new approach need not know the image information in advance. A satisfactory result can be obtained even in a low signal-to-noise ratio image

Keywords

Gaussian processes; filtering theory; image segmentation; tree data structures; Gaussian filter function; changeable filter scales; image segmentation; linked pyramid; low signal-to-noise ratio; root nodes; Image processing; Image recognition; Image resolution; Image sampling; Image segmentation; Information filtering; Information filters; Pixel; Signal resolution; Signal to noise ratio;

fLanguage

English

Publisher

ieee

Conference_Titel

Signal Processing, 1996., 3rd International Conference on

Conference_Location

Beijing

Print_ISBN

0-7803-2912-0

Type

conf

DOI

10.1109/ICSIGP.1996.566290

Filename

566290