Character proposal network for robust text extraction
Authors:
- 1 March 2016
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2633-2637
Abstract
Maximally stable extremal regions (MSER), which is a popular method to generate character proposals/candidates, has shown superior performance in scene text detection. However, the pixel-level operation limits its capability for handling some challenging cases (e.g., multiple connected characters, separated parts of one character and non-uniform illumination). To better tackle these cases, we design a character proposal network (CPN) by taking advantage of the high capacity and fast computing of fully convolutional network (FCN). Specifically, the network simultaneously predicts character-ness scores and refines the corresponding locations. The character-ness scores can be used for proposal ranking to reject non-character proposals and the refining process aims to obtain the more accurate locations. Furthermore, considering the situation that different characters have different aspect ratios, we propose a multi-template strategy, designing a refiner for each aspect ratio. The extensive experiments indicate our method achieves recall rates of 93.88%, 93.60% and 96.46% on ICDAR 2013, SVT and Chinese 2k datasets respectively using less than 1000 proposals, demonstrating promising performance of our character proposal network.Keywords
All Related Versions
This publication has 16 references indexed in Scilit:
- A robust approach for text detection from natural scene imagesPattern Recognition, 2015
- Caffe2014
- Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation2014
- Deep Features for Text Spotting2014
- Robust Scene Text Detection with Convolution Neural Network Induced MSER Trees2014
- Robust Text Detection in Natural Scene ImagesIEEE Transactions on Pattern Analysis and Machine Intelligence, 2013
- ICDAR 2013 Robust Reading Competition2013
- Selective Search for Object RecognitionInternational Journal of Computer Vision, 2013
- Scene Text Detection via Connected Component Clustering and Nontext FilteringIEEE Transactions on Image Processing, 2013
- Word Spotting in the Wild2010