Correction of slanted text pictures based on modified opencv

作     者：Lu, Ran Xu, Yongjin Wu, Zhanhe Zheng, Yijie Ren, Lingxin Ding, Feifei Li, Jianjun Chang, Chin-Chen 

作者机构：State Grid Zhejiang Marketing Service Center Hangzhou311100 China School of Coomputer Science and Technology Hangzhou Dianzi University No.1 of the 2nd street XiaSha Hangzhou310018 China Department of Information and Engineering Computer Science Feng Chia University No. 100 Wenhwa Rd. Seatwen Taichung40724 Taiwan 

出 版 物：《Journal of Network Intelligence》 (J. Network Intell.)

年 卷 期：2021年第6卷第2期

页      面：238-246页

核心收录：

基　　金：Acknowledgements. This work was supported in part by National Science Fund of China no.61871170 Key Research and Development Plan of Zhejiang: No .2021C03131 The Basic Research Program of KY2017210A001 Key Laboratory of Brain Machine Collaborative Intelligence of Zhejiang Province 

主　　题：Optical character recognition 

摘      要：Correction of slanted or tilted text is very important in many applications, such as optical character recognition (OCR), text detection and recognition. It is not difficult to find that the recognition accuracy of OCR has a great relationship with the tilt angle of the recognized picture, resulting in reduced OCR adaptability and great limitations. As for the problem, we proposed an innovative optimization to estimate the angle of tilting texts in a picture based on Hough transform and modified OpenCV, which reduces the negative effect of background noises. Our algorithm outperforms the state-of-the-art methods in both performances and robustness. © 2021 Global Research Online. All rights reserved.