Int J Performability Eng ›› 2019, Vol. 15 ›› Issue (1): 281-287.doi: 10.23940/ijpe.19.01.p28.281287

Previous Articles     Next Articles

Can Machine Automatically Discover Text Image from Overall Perspective

Wei Jianga, Jiayi Wua, and Chao Yaob*()   

  1. a School of Software, North China University of Water Resources and Electric Power,Zhengzhou, 450045,China
    b School of Automation,Northwestern Polytechnic University,Xi’an,710071, China
  • Revised on ; Accepted on
  • Contact: Yao Chao E-mail:yaochao@nwpu.edu.cn
  • About author:Wei Jiang received the PH.D. degree from Xidan University, Xi’an, China, in 2014. He now works as a senior lecturer in School of Software, North China University of Water Resources and Electric Power in Zhengzhou, China. His interest is scene text detection and recognition.|Jiayi Wu received her master degree from the University of Warwick, England, in 2010. She now works for School of Software, North China University of Water Resources and Electric Power in Zhengzhou, China.|Chao Yao received the PH.D. degree from Xidan University, Xi’an, China, in 2014. He has visited Concordia University in Montreal, Canada as joint PH.D. Student from 2011 to 2012. He has finished post-doc in 2017 and works as assistant professor in School of Automation, Northwestern Polytechnic University, in Xi’an, China. His interest is text recognition and dimension reduction.

Abstract:

Recently, more and more researchers have focused on the problem about how to automatically distinguish text images from non-text ones. Most of previous works have originated from local features, which are computational expensive, and usually employ GPU in their procedure. To address this problem, we propose a new and simple but effective scheme from an overall perspective. In the proposed scheme, a sort of holistic feature is first extracted from Fourier spectrum, which describes the characteristic of the image or the sub-image as a whole without local feature extraction; then, random forests are utilized to classify images into text and non-text ones. Experimental results in several public datasets demonstrate that this scheme is efficient and effective.

Key words: natural images, holistic feature, text/non-text image classification, random forests