Unsupervised color image segmentation

Author

Rujie, Liu ; Baozong, Yuan

Author_Institution

Fujitsu R&D Center Co. Ltd., Beijing, China

Volume

1

fYear

2002

fDate

26-30 Aug. 2002

Firstpage

744

Abstract

Color image segmentation is highly useful in many applications including image enhancement, target recognition, and image indexing for content-based retrieval. A novel method for unsupervised color image segmentation is proposed in this paper, which consists of two steps. In the first step, the high frequency wavelet coefficients are used to divide the whole image into perceptually meaningful object-of-interest and background. Then, Deng´s (2001) ´good segmentation´ criterion is applied to the extracted object areas to get the contour of the objects. This method is automatic in that it does not need any interaction. The efficiency is shown through some experimental results.

Keywords

edge detection; feature extraction; image colour analysis; image segmentation; wavelet transforms; Deng good segmentation criterion; background; extracted object areas; high frequency wavelet coefficients; object contour; perceptually meaningful object; unsupervised color image segmentation; Clustering algorithms; Color; Convolution; Discrete wavelet transforms; Focusing; Frequency; Image segmentation; Low pass filters; Object detection; Wavelet coefficients;

fLanguage

English

Publisher

ieee

Conference_Titel

Signal Processing, 2002 6th International Conference on

Print_ISBN

0-7803-7488-6

Type

conf

DOI

10.1109/ICOSP.2002.1181163

Filename

1181163