Characterness

Characterness: An Indicator of Text in the Wild

Yao Li, Wenjing Jia, Chunhua Shen, Anton van den Hengel

Australian Centre for Visual Technologies, Univerisity of Adelaide, Australia
Research Centre for Innovation in IT Services and Applications, University of Technology Sydney, Australia

 

characterness_pipeline2

Abstract: Text in an image provides vital information for interpreting its contents, and text in a scene can aid a variety of tasks from navigation to obstacle avoidance and odometry. Despite its value, however, detecting general text in images remains a challenging research problem. Motivated by the need to consider the widely varying forms of natural text, we propose a bottom-up approach to the problem which reflects the ‘characterness’ of an image region. In this sense our approach mirrors the move from saliency detection methods to measures of ‘objectness’ . In order to measure the characterness we develop three novel cues that are tailored for character detection, and a Bayesian method for their integration. Because text is made up of sets of characters, we the design a Markov random field (MRF) model so as to exploit the inherent dependencies between characters. We experimentally demonstrate the effectiveness of our characterness cues as well as the advantage of Bayesian multi-cue integration. The proposed text detector outperforms state-of-the-art methods on a few benchmark scene text detection datasets. We also show that our measurement of ‘characterness’ is superior than state-of-the-art saliency detection models when applied to the same task.

Citation: Yao Li, Wenjing Jia, Chunhua Shen and Anton van den Hengel “Characterness: An Indicator of Text in the Wild,” IEEE Transcations on Image Processing, 2014. [PDF]

Code is avaiable from our GitHub repository. You can also download the text saliency maps of 100 testing images of ICDAR 2013 dataset from here.

Quantitative evaluation

Quantitative precision and recall curves (left, middle) and F-measure (right): 

saliencyEvaluation

Results on benchmark datasets:

precision recall f-measure
ICDAR2003 Dataset 0.79 0.64 0.71
ICDAR2011 Dataset 0.80 0.62 0.70
Oriented Scene Text Database 0.72 0.60 0.61

Qualitative evaluation
Examples of characterness maps (original image, ground truth, characterness map):

saliencyExample

Examples of scene text detection result:

saliencyEvaluation