A neural network approach for learning image similarity in adaptive CBIR

Author

Muneesawang, P. ; Guan, L.

Author_Institution

School. of Electr. and Inf. Eng., University of Sydney, NSW, Australia

fYear

2001

fDate

2001

Firstpage

257

Lastpage

262

Abstract

The adoption of neural network techniques is studied for the purpose of image retrieval. More specifically, we propose an adaptive retrieval system which incorporates learning capability into the image retrieval module where the network weights represent the adaptivity. This system can learn users´ notions of similarity between images through the continual relevance feedback from the users. Accordingly it makes the proper adjustment to improve performance. This retrieval system has demonstrated its effectiveness in performance. It is confirmed by simulations conducted for applications such as texture retrieval and retrieval of DCT compressed images

Keywords

adaptive systems; content-based retrieval; data compression; discrete cosine transforms; image coding; image matching; image retrieval; image texture; information retrieval system evaluation; learning (artificial intelligence); neural nets; relevance feedback; transform coding; CBIR; DCT compressed images; LVQ; SOTM; adaptive retrieval system; content-based image retrieval; image database; image matching; image similarity; learning vector quantization; machine learning; network weights; neural network; radial basis function network; relevance feedback; retrieval system performance; self-organizing tree map; single-pass RBFN; texture retrieval; Adaptive systems; Clustering algorithms; Discrete cosine transforms; Image coding; Image retrieval; Intelligent networks; Neural networks; Neurofeedback; Prototypes; Training data;

fLanguage

English

Publisher

ieee

Conference_Titel

Multimedia Signal Processing, 2001 IEEE Fourth Workshop on

Conference_Location

Cannes

Print_ISBN

0-7803-7025-2

Type

conf

DOI

10.1109/MMSP.2001.962743

Filename

962743