Robust Character Recognition For Optical And Natural Images Using Deep Learning

被引:0
作者
Abdali, Al Maamoon Rasool [1 ]
Ghani, Rana Fareed [1 ]
机构
[1] Univ Technol Baghdad, Minist Higher Educ, Comp Sci, Baghdad, Iraq
来源
2019 17TH IEEE STUDENT CONFERENCE ON RESEARCH AND DEVELOPMENT (SCORED) | 2019年
关键词
EMINST; convolutional neural network; IC-DAR2003; Char74k; OCR;
D O I
10.1109/scored.2019.8896354
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Character recognition is one of the most critical parts of many computer vision system. And many studies have explored character recognition as two subcategories: character recognition in optical images, character recognition natural images while this separation was to achieve high accuracy in each field separated but it needs more hardware resources to operate two models one for each area. In addition to that most researches divided each field into (digits recognition, character recognition) that add extra cost in both time and hardware resources also we found that both areas still have room for accuracy improvement. This paper tackles the problem of the two subclasses by building one robust, accurate classifier using a convolutional neural network that can recognize (characters and digits) accurately in both optical and natural scene images, the proposed model has been trained on a combination of EMNIST and Char74k data sets with a random data augmentation. The proposed model achieved 92% accuracy in EMINST compared to previous works shows that the proposed model has the highest accuracy among all the previous works based on EMNIST data set. We also tested the model on none-seen data sets (ICDAR203) and the obtained results indicate the high generality and the robustness of the classifier.
引用
收藏
页码:152 / 156
页数:5
相关论文
共 50 条
  • [21] Denoising of Video Frames Resulting From Video Interface Leakage Using Deep Learning for Efficient Optical Character Recognition
    Galvis, J.
    Morales, S.
    Kasmi, C.
    Vega, F.
    IEEE LETTERS ON ELECTROMAGNETIC COMPATIBILITY PRACTICE AND APPLICATIONS, 2021, 3 (02): : 82 - 86
  • [22] Hybrid Handwriting Character Recognition with Transfer Deep Learning
    Can, Ferit
    Yilmaz, Atinc
    2019 27TH SIGNAL PROCESSING AND COMMUNICATIONS APPLICATIONS CONFERENCE (SIU), 2019,
  • [23] High accuracy Optical Character Recognition algorithms using learning array of ANN
    Vani, B.
    Beaulah, M. Shyni
    Deepalakshmi, R.
    2014 IEEE INTERNATIONAL CONFERENCE ON CIRCUIT, POWER AND COMPUTING TECHNOLOGIES (ICCPCT-2014), 2014, : 1474 - 1479
  • [24] Improving Optical Character Recognition performance for low quality images
    Brisinello, Matteo
    Grbic, Ratko
    Pul, Matija
    Andelic, Tihomir
    PROCEEDINGS OF 2017 INTERNATIONAL SYMPOSIUM ELMAR, 2017, : 167 - 171
  • [25] Using sparse pixel character vectorisation for optical character recognition
    Salih, Qussay A.
    Raman, VikramAdith
    PROCEEDINGS OF THE IASTED INTERNATIONAL CONFERENCE ON ADVANCES IN COMPUTER SCIENCE AND TECHNOLOGY, 2006, : 174 - 179
  • [26] Understanding Natural Disaster Scenes from Mobile Images Using Deep Learning
    Tang, Shimin
    Chen, Zhiqiang
    APPLIED SCIENCES-BASEL, 2021, 11 (09):
  • [27] A Method to Identify the Cause of Misrecognition for Offline Handwritten Japanese Character Recognition using Deep Learning
    Gyohten, Keiji
    Ohki, Hidehiro
    Takami, Toshiya
    ICPRAM: PROCEEDINGS OF THE 9TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION APPLICATIONS AND METHODS, 2020, : 446 - 452
  • [28] Robust Character Recognition Using Adaptive Feature Extraction Method
    Mori, Minoru
    Sawaki, Minako
    Yamato, Junji
    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS, 2010, E93D (01): : 125 - 133
  • [29] The Application of Deep Convolutional Denoising Autoencoder for Optical Character Recognition Preprocessing
    Wiraatmaja, Christopher
    Gunadi, Kartika
    Sandjaja, Iwan Njoto
    2017 INTERNATIONAL CONFERENCE ON SOFT COMPUTING, INTELLIGENT SYSTEM AND INFORMATION TECHNOLOGY (ICSIIT), 2017, : 72 - 77
  • [30] Optical Character Recognition for Test Automation Using LabVIEW
    Perala, Srinivas
    Roy, Ajay
    Ranjan, Sandeep
    SUSTAINABLE COMMUNICATION NETWORKS AND APPLICATION, ICSCN 2021, 2022, 93 : 489 - 496