Character recognition in natural scene images using local description

Authors:
Boyu Zhang;Wei Zhao;JiaFeng Liu;Rui Wu;XiangLong Tang
Affiliations:
School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China;School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China;School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China;School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China;School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China
Venue:
IScIDE'11 Proceedings of the Second Sino-foreign-interchange conference on Intelligent Science and Intelligent Data Engineering
Year:
2011

Citing 5
Cited 0

Character recognition—a review

Pattern Recognition
A Bayesian Approach to Unsupervised One-Shot Learning of Object Categories

ICCV '03 Proceedings of the Ninth IEEE International Conference on Computer Vision - Volume 2
One-Shot Learning of Object Categories

IEEE Transactions on Pattern Analysis and Machine Intelligence
An embedded application for degraded text recognition

EURASIP Journal on Applied Signal Processing
Text String Detection From Natural Scenes by Structure-Based Partition and Grouping

IEEE Transactions on Image Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Text information extracted from scene images is often the key clue for better performance of scene understanding and image retrieval. However, the clutter background and variations, which are intrinsic in scene images, make the natural scene character recognition task rather complicated. To overcome these disadvantages, we propose a novel approach for character recognition task in natural scene images. In the method, character classes are described by groups of local features using a probabilistic model. Structures of characters are represented by mutual positions of local features. For model learning, parameter estimating is done through expectation-maximization in a weak-supervised manner. Experiment results over datasets which includes both synthetic and authentic data demonstrate the validity of the approach.