Image Recognition System That Uses Visual Word

Author

Min-Uk Kim ; Kyoungro Yoon

Author_Institution

Sch. of Comput. Sci. & Eng., Konkuk Univ., Seoul, South Korea

fYear

2014

fDate

6-9 May 2014

Firstpage

1

Lastpage

2

Abstract

To deal with a large-scale image database, we design and implement an image recognition system using visual word. First, SIFT (Scale-invariant Feature Transform) features are extracted from the images. Subset of these features is then selected during the feature selection process. Finally selected features are quantized to the visual words. These visual words play an important role at the first search phase and to the overall precision. At the second search, original SIFT features are used to rearrange the result. Experimental results show 93% precision and 2 seconds retrieval time.

Keywords

document image processing; feature extraction; feature selection; image recognition; image retrieval; quantisation (signal); transforms; visual databases; SIFT features; feature extraction; feature selection process; image recognition system; large-scale image database; quantization; scale-invariant feature transform; search phase; visual word; Feature extraction; Image recognition; Indexing; Libraries; Quantization (signal); Three-dimensional displays; Visualization;

fLanguage

English

Publisher

ieee

Conference_Titel

Information Science and Applications (ICISA), 2014 International Conference on

Conference_Location

Seoul

Print_ISBN

978-1-4799-4443-9

Type

conf

DOI

10.1109/ICISA.2014.6847410

Filename

6847410