Automatic logo detection and recognition continues to be of great interest to the document retrieval community as it enables effective identification of the source of a document. In this paper, we propose a new approach to logo detection and extraction in document images that robustly classifies and precisely localizes logos using a boosting strategy across multiple image scales. At a coarse scale, a trained Fisher classifier performs initial classification using features from document context and connected components. Each logo candidate region is further classified at successively finer scales by a cascade of simple classifiers, which allows false alarms to be discarded and the detected region to be refined. Our approach is segmentation free and lay-out independent. We define a meaningful evaluation metric to measure the quality of logo detection using labeled groundtruth. We demonstrate the effectiveness of our approach using a large collection of real-world documents.
%0 Conference Paper
%1 4377038
%A Zhu, Guangyu
%A Doermann, D.
%D 2007
%J Document Analysis and Recognition, 2007. ICDAR 2007. Ninth International Conference on
%K Fisher automatic boosting classifier detection document extraction feature groundtruth image images labeled logo processing recognition retrieval strategy
%P 864-868
%R 10.1109/ICDAR.2007.4377038
%T Automatic Document Logo Detection
%V 2
%X Automatic logo detection and recognition continues to be of great interest to the document retrieval community as it enables effective identification of the source of a document. In this paper, we propose a new approach to logo detection and extraction in document images that robustly classifies and precisely localizes logos using a boosting strategy across multiple image scales. At a coarse scale, a trained Fisher classifier performs initial classification using features from document context and connected components. Each logo candidate region is further classified at successively finer scales by a cascade of simple classifiers, which allows false alarms to be discarded and the detected region to be refined. Our approach is segmentation free and lay-out independent. We define a meaningful evaluation metric to measure the quality of logo detection using labeled groundtruth. We demonstrate the effectiveness of our approach using a large collection of real-world documents.
@inproceedings{4377038,
abstract = {Automatic logo detection and recognition continues to be of great interest to the document retrieval community as it enables effective identification of the source of a document. In this paper, we propose a new approach to logo detection and extraction in document images that robustly classifies and precisely localizes logos using a boosting strategy across multiple image scales. At a coarse scale, a trained Fisher classifier performs initial classification using features from document context and connected components. Each logo candidate region is further classified at successively finer scales by a cascade of simple classifiers, which allows false alarms to be discarded and the detected region to be refined. Our approach is segmentation free and lay-out independent. We define a meaningful evaluation metric to measure the quality of logo detection using labeled groundtruth. We demonstrate the effectiveness of our approach using a large collection of real-world documents.},
added-at = {2009-04-23T17:17:00.000+0200},
author = {Zhu, Guangyu and Doermann, D.},
biburl = {https://www.bibsonomy.org/bibtex/268dfe18983916932a5b94683cbbe0326/dieudonnew},
doi = {10.1109/ICDAR.2007.4377038},
interhash = {50018496427a657287df168ba2fbf53b},
intrahash = {68dfe18983916932a5b94683cbbe0326},
issn = {1520-5363},
journal = {Document Analysis and Recognition, 2007. ICDAR 2007. Ninth International Conference on},
keywords = {Fisher automatic boosting classifier detection document extraction feature groundtruth image images labeled logo processing recognition retrieval strategy},
month = {Sept.},
pages = {864-868},
timestamp = {2009-04-23T17:17:00.000+0200},
title = {Automatic Document Logo Detection},
volume = 2,
year = 2007
}