DocumentCode :
1748846
Title :
A self-organizing map with dynamic architecture for efficient color quantization
Author :
Kirk, James S. ; Chang, Dar-Jen ; Zurada, Jacek M.
Author_Institution :
Comput. Sci. & Eng. Program, Louisville Univ., KY, USA
Volume :
3
fYear :
2001
fDate :
2001
Firstpage :
2128
Abstract :
Color quantization is often used to convert 24-bit RGB images to 8-bit palette-table images. However, in some cases, the imposed 8 bits per pixel may be too stringent to adequately represent the image. For other images, 8 bits per pixel are unnecessarily generous. For image storage and transmission, it is important to compress an image as much as possible without exceeding an allowable level of degradation. The paper describes the use of a dynamically-growing self-organizing map (SOM) to determine the palette-table required to adequately represent the colors of an RGB image, given an allowable degree of quantization error
Keywords :
data compression; image coding; image colour analysis; learning (artificial intelligence); neural net architecture; self-organising feature maps; RGB image; color quantization; dynamic architecture; image compression; image storage; image transmission; quantization error; self-organizing map; Color; Computer architecture; Computer science; Image converters; Image storage; Kirk field collapse effect; Neurons; Pixel; Prototypes; Quantization;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Neural Networks, 2001. Proceedings. IJCNN '01. International Joint Conference on
Conference_Location :
Washington, DC
ISSN :
1098-7576
Print_ISBN :
0-7803-7044-9
Type :
conf
DOI :
10.1109/IJCNN.2001.938495
Filename :
938495
Link To Document :
بازگشت