A novel location and matching algorithm for rapid logo recognition in video advertisements

Author

Yuan Zhang ; Shuwu Zhang ; Wei Liang ; Jinchun Liang

Author_Institution

High-Tech Innovation Center, Inst. of Autom., Beijing, China

fYear

2012

fDate

9-11 Nov. 2012

Firstpage

1

Lastpage

5

Abstract

This paper introduces a novel location and matching algorithm for rapid logo recognition in video advertisements. This work is distinguished by two major contributions. The first is a new pre-location algorithm to obtain logo prediction regions. The algorithm uses HSV color quantization method to get connected components of the video pictures, and then gains the logo prediction regions based on some aggregative rules. The second is a new index structure for logo matching, using content-based multi-feature fusion, such as HSV block color histogram, CEGCH and representative shape context. Experimental results for video advertisements containing various logos illustrate that our algorithm speeds up logo recognition and shows precise location on our large logo database.

Keywords

advertising; gradient methods; image colour analysis; image fusion; image matching; image recognition; image representation; statistical analysis; video signal processing; visual databases; CEGCH; HSV block color histogram; HSV color quantization method; aggregative rule; color edge gradient co-occurrence histogram; content-based multifeature fusion; logo database; logo prediction region; matching algorithm; prelocation algorithm; rapid logo recognition; representative shape context; video advertisement; Context; Histograms; Image color analysis; Image recognition; Indexes; Shape; CEGCH; HSV block color histogram; HSV color quantization; connected; general shape context;

fLanguage

English

Publisher

ieee

Conference_Titel

Image Analysis and Signal Processing (IASP), 2012 International Conference on

Conference_Location

Hangzhou

Print_ISBN

978-1-4673-2547-9

Type

conf

DOI

10.1109/IASP.2012.6424989

Filename

6424989