Article
Computer Science, Information Systems
Sarayut Gonwirat, Olarik Surinta
Summary: This research focuses on the noise problem in handwritten character images and proposes the DeblurGAN-CNN architecture to synthesize clean character images and enhance recognition performance. Experimental results show that the DeblurGAN-CNN outperforms existing methods on different datasets.
Article
Computer Science, Software Engineering
Deena Hijam, Sarat Saharia
Summary: This paper proposes a multilevel approach for recognizing Meitei Mayek handwritten characters using a fusion of features strategy. The features are designed to maximize inter-class variations and minimize intra-class variations. The methodology achieves superior performance in identifying structurally similar character classes, with an overall recognition accuracy of 97.08%, setting a benchmark on the Meitei Mayek handwritten character dataset.
Review
Computer Science, Artificial Intelligence
Hossam Magdy Balaha, Hesham Arafat Ali, Mahmoud Badawy
Summary: The comprehensive review focuses on the current research trends in the area of Arabic language, especially state-of-the-art approaches to facilitate the adaption of previous systems into new applications. The article highlights the deep, widespread, and unexplored scope of research in the Arabic language field, calling for modern methods with fewer errors. Specifically, the paper critically analyzes the challenges faced by previous researchers and provides recommendations for future advances.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Computer Science, Artificial Intelligence
Danilo Avola, Manoochehr Joodi Bigdello, Luigi Cinque, Alessio Fagioli, Marco Raoul Marini
Summary: Handwritten signatures are widely used for person identification, but face challenges due to possible skilled forgeries; solutions based on convolutional neural networks can extract features for improved performance and reduced training times.
PATTERN RECOGNITION LETTERS
(2021)
Article
Computer Science, Artificial Intelligence
Mohammed Salemdeeb, Sarp Erturk
Summary: This study introduces a novel full depth stacked CNN architecture for recognition of handwritten alphanumeric characters and license plate characters, providing low complexity, high accuracy, and full feature extraction. Test results show very low error rates across multiple datasets.
PEERJ COMPUTER SCIENCE
(2021)
Article
Computer Science, Information Systems
Aqsa Rasheed, Nouman Ali, Bushra Zafar, Amsa Shabbir, Muhammad Sajid, Muhammad Tariq Mahmood
Summary: This study presents a classification framework for automatically recognizing handwritten Urdu characters and digits with higher accuracy. By utilizing transfer learning and pre-trained convolutional neural networks, the proposed method achieves impressive results in recognizing Urdu characters and digits.
Article
Computer Science, Software Engineering
Deena Hijam, Sarat Saharia
Summary: This paper introduces a large-scale Meitei Mayek handwritten character database and focuses on collecting natural handwriting through two phases of sample collection. The experimental results show that using the CNN model proposed in the paper can achieve a test accuracy of 95.56% on the database.
Article
Computer Science, Information Systems
Naseem Alrobah, Saleh Albahli
Summary: Handwriting recognition for computer systems has been researched extensively, with most experiments conducted in English. However, other languages such as Arabic also require research. This study focuses on recognizing and developing Arabic handwritten characters, achieving a recognition rate of up to 96.3% for 29 classes.
Article
Computer Science, Artificial Intelligence
Abhishek Hazra, Prakash Choudhary, Sanasam Inunganbi, Mainak Adhikari
Summary: The paper introduces a unique CNN architecture for recognizing handwritten characters in Bangla and Meitei Mayek scripts. The model is validated on different datasets and shows excellent performance.
APPLIED INTELLIGENCE
(2021)
Article
Computer Science, Artificial Intelligence
Hossam Magdy Balaha, Hesham Arafat Ali, Mohamed Saraya, Mahmoud Badawy
Summary: The article discusses the research on handwritten character recognition in Arabic text, presenting a large dataset and introducing a deep learning system with convolutional neural network structures. Sixteen experiments were conducted and various performance metrics were compared, highlighting the importance of data augmentation in improving accuracy.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Computer Science, Artificial Intelligence
Teressa Longjam, Dakshina Ranjan Kisku, Phalguni Gupta
Summary: This paper proposes a novel approach towards offline signature verification using a hybrid deep learning network, and evaluates its performance on different datasets. The experimental results show that the system performs well on multi-scripted signatures.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Majid A. Khan, Nazeeruddin Mohammad, Ghassen Ben Brahim, Abul Bashar, Ghazanfar Latif
Summary: This paper proposes an offline text-independent writer verification approach for handwritten Arabic text based on individual character shapes. The approach is capable of author verification even with partially damaged documents and identifies the effectiveness of Arabic characters during the verification process. The experimental results on a dataset with 82 different users demonstrate high accuracy of the proposed approach in various scenarios.
PEERJ COMPUTER SCIENCE
(2022)
Review
Computer Science, Information Systems
Sanasam Inunganbi
Summary: This paper presents an overview of the research progress and application demand in the field of character recognition. It discusses the history, steps, and techniques of character recognition, and highlights the importance of offline handwritten recognition systems.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Faisel Mushtaq, Muzafar Mehraj Misgar, Munish Kumar, Surinder Singh Khurana
Summary: The study proposed a handwritten Urdu character dataset and recognition system using a convolutional neural network architecture, achieving a recognition rate of 98.82% for 133 classes, outperforming all existing systems for the Urdu language.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Computer Science, Information Systems
C. Vinotheni, S. Lakshmana Pandian
Summary: This paper proposes an end-to-end deep learning model for Tamil handwritten document recognition. The model utilizes segmentation at the word and line levels, deep convolutional neural network, and water strider optimization algorithm to achieve real-time recognition with high accuracy and precision.
Article
Acoustics
Shuai Nie, Shan Liang, Wenju Liu, Xueliang Zhang, Jianhua Tao
IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING
(2018)
Article
Engineering, Electrical & Electronic
Xifeng Liu, Shan Liang, Wenju Liu, Ping Sun
Summary: This paper presents a voltage reference using MOSFET with high-order curvature compensation and ZTCMOS bias to achieve low temperature coefficient. Test results show an output voltage of 628mV and a temperature coefficient of 2.5 ppm/°C.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II-EXPRESS BRIEFS
(2021)
Article
Computer Science, Artificial Intelligence
Guanjun Li, Shan Liang, Shuai Nie, Wenju Liu, Zhanlei Yang
Summary: The nnGSC is a novel dual-channel DNN-based structure optimized for maximizing ASR performance. It outperforms traditional approaches and has the capability to automatically track target DOA, offering improved robustness against array geometry mismatches.
Article
Computer Science, Artificial Intelligence
Yaping Zhang, Shuai Nie, Shan Liang, Wenju Liu
Summary: Robust text reading is a challenging task due to the changing distribution of text images in real-world scenarios. Existing domain adaptation methods may struggle with sequence-like text images as they fail to handle variable-length fine-grained character information. To address this, a novel Adversarial Sequence-to-Sequence Domain Adaptation (ASSDA) method is proposed to align local regions containing characters across domains efficiently.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)