Character Recognition Accuracy Research Articles

V. Hrinkov, G. Hrinkova, S. Hrinkov. Analysis of modern optical character recognition tools for character recognition and text from the image

Optical character recognition is a powerful technology that converts images with text into an editable and searchable format. This ensures efficiency in working with documents, increases the accessibility of information, and helps automate many processes. This technology was first used in the early 90s, when digitizing historical newspapers to create an electronic archive. In recent years, the optical character recognition system has been refined to the "ideal": current optical character recognition systems show almost perfect text recognition accuracy. But this requires the following requirements to be met: equality and contrast of symbols; uniformity of the text background; contrast between the background and letters. The process of recognizing symbols and text from images is much more complicated when the above requirements are not met, and solving such problems is the current requirement for solving practical tasks in the military sphere. Text recognition from images has many important applications, which makes it relevant and necessary. The article analyzes several of the most well-known and popular artificial intelligence models for text recognition from images, such as Tesseract OCR, PyTorch, EasyOCR, Keras OCR5, OpenCV, using specific types of images, and obtains results for recognizing text and symbols from images of varying complexity. To assess the accuracy of character and text recognition, a recognition accuracy assessment method has been developed based on special evaluation metrics, which are based on comparing the recognized text with the reference (correct) text. The most common metrics include the character recognition accuracy (CAR) and the word recognition accuracy (WAR). Using the developed recognition accuracy assessment method, an analysis of the recognition accuracy of the most popular tools of optical text and symbol recognition technology from images of varying complexity has been conducted. The analysis showed that the EasyOcr model demonstrates the greatest efficiency and accuracy of recognition, which, even in conditions of strong "noise" and poor image contrast, demonstrated a stable result and, subject to further customization for the user's needs, can be used to solve a specific task.

Klasifikasi Aksara Sasak Menggunakan Convolutional Neural Networks (CNN)

Sasak script is an important cultural heritage for the people of Lombok, but its use is decreasing along with the development of digital technology. This study aims to develop a classification system for Sasak script handwriting using Convolutional Neural Networks (CNN) to improve the accuracy of character recognition. The dataset used consists of handwritten images of 18 basic Sasak script characters collected from 50 volunteers with various writing styles. The methods applied include data preprocessing, augmentation, and training a CNN model with an architecture consisting of several convolutional and pooling layers. The results showed that the model achieved a validation accuracy of 92%, an average precision of 0.91, a recall of 0.89, and an F1-score of 0.90, indicating excellent performance in recognizing Sasak script characters. The conclusion of this study is that the developed system is not only effective in character recognition, but can also function as an interactive learning tool, supporting efforts to preserve Sasak script in the digital era. This research opens up opportunities for further development in the introduction of other traditional scripts.

Self-distillation with beta label smoothing-based cross-subject transfer learning for P300 classification

Background:The P300 speller is one of the most well-known brain-computer interface (BCI) systems, offering users a novel way to communicate with their environment by decoding brain activity. Problem:However, most P300-based BCI systems require a longer calibration phase to develop a subject-specific model, which can be inconvenient and time-consuming. Additionally, it is challenging to implement cross-subject P300 classification due to significant inter-individual variations. Method:To address these issues, this study proposes a calibration-free approach for P300 signal detection. Specifically, we incorporate self-distillation along with a beta label smoothing method to enhance model generalization and overall system performance, which can not only enable the distillation of informative knowledge from the electroencephalogram (EEG) data of other subjects but effectively reduce individual variability. Experimental results:The results conducted on the publicly available OpenBMI dataset demonstrate that the proposed method achieves statistically significantly higher performance compared to state-of-the-art approaches. Notably, the average character recognition accuracy of our method reaches up to 97.37% without the need for calibration. And information transfer rate and visualization further confirm its effectiveness. Significance:This method holds great promise for future developments in BCI applications.

Handwritten Text Recognition: A Survey of OCR Techniques

Optical Character Recognition of handwritten texts has witnessed remarkable advancements with the integration of deep learning and machine learning techniques. Recognizing handwritten characters poses unique challenges due to script variability, linguistic diversity, and the complexities of historical documents. This survey explores recent developments in OCR for various languages, emphasizing innovative approaches such as Convolutional Neural Networks, attention mechanisms, and transfer learning. We analyze methodologies that enhance character recognition accuracy across languages, including Amharic, Arabic, Uchen Tibetan, Devanagari, and Tamil, while addressing resource efficiency in scene text recognition. Furthermore, the paper discusses advanced techniques for multilingual numeral recognition and writer identification in Indic scripts, highlighting cutting-edge strategies that push the boundaries of OCR technology. By synthesizing findings from the latest literature, this review provides valuable insights into ongoing challenges and future research directions in handwritten text recognition.

Stretchable, conductive Hydrogel-Based triboelectric nanogenerator integrated with deep learning algorithm for character recognition

With the rapid development of flexible electronics, combining wearable devices together with deep learning has been a hot research topic. Conventional TENGs are constrained by their lack of flexibility, hindering their applications on wearable devices. In this work, a highly stretchable and conductive double network PAM/agar/KCl-4 hydrogel was fabricated, with an elastic modulus of 506.67 kPa, and a conductivity of 1.33 S m−1, the hydrogel showed great potential in strain sensing and information transmission. In addition, we combined deep learning algorithm together with the triboelectric nanogenerator which employed the hydrogel to work as its electrode. The deep learning-assisted character recognition accuracy reached 99.11 %. This work is expected to provide new ideas for flexible TENG devices.

Deep learning-based digitization of Kurdish text handwritten in the e-government system

Many government institutions in developing countries such as the Kurdistan Region of Iraq (KRI) keep a variety of paper-based records that are available in printed or handwritten format. The need for technology that turns handwritten writing into digital text is therefore highly demanded. E-government in developed and developing countries is a crucial facilitator for the provision of such services. This paper aims to develop a deep learning model based on the mask region convolutional neural network (mask-RCNN) to effectively digitize kurdish handwritten text recognition (KHTR). In this research, typical datasets, which includes the isolated handwritten Central Kurdish character images, an extensive database of 40,410 images, and 390 native writers have been produced to determine the developed approach’s performance in terms of identification rates. This approach achieves adequate outcomes in terms of training time and accuracy. The proposed model gives higher performance for detection, localization, and recognition when using a dataset containing many challenges, the results were 80%, 96%, and 87.6 for precision, recall, and F-score respectively. The findings revealed that the proposed model obtained better results compared to other similar works. The accuracy of optical character recognition (OCR) is more than 99%.

Hetero-associative Memory Based New Iraqi License Plate Recognition

نتيجة للتطورات الأخيرة في أبحاث الطرق السريعة بالإضافة إلى زيادة استخدام المركبات، كان هناك اهتمام كبير بنظام النقل الذكي الأكثر حداثة وفعالية ودقة (ITS) في مجال رؤية الكمبيوتر أو معالجة الصور الرقمية، يلعب تحديد كائنات معينة في صورة دورًا مهمًا في إنشاء صورة شاملة. هناك تحدٍ مرتبط بالتعرف على لوحة ترخيص السيارة (VLPR) بسبب الاختلاف في وجهة النظر، والتنسيقات المتعددة، وظروف الإضاءة غير الموحدة في وقت الحصول على الصورة والشكل واللون، بالإضافة إلى الصعوبات مثل ضعف دقة الصورة ، الصورة الباهتة ، الإضاءة السيئة، التباين المنخفض، يجب التغلب عليها. اقترحت هذه الورقة نموذجًا باستخدام تعديل الذاكرة الترابطية ثنائية الاتجاه (MBAM)، وهي نوع واحد من الذاكرة الترابطية غير المتجانسة، وتعمل MBAM على مرحلتين)مرحلتي التعلم والتقارب) للتعرف على اللوحة، ويمكن لهذا النموذج المقترح التغلب على تلك الصعوبات بسبب قدرة الذاكرة الترابطية لـ MBAM على قبول الضوضاء وتمييز الصور المشوهة، وكذلك سرعة عملية الحساب نظرًا لصغر حجم الشبكة. نتيجة دقة تحديد منطقة اللوحة هي 99.6٪، ودقة تجزئة الأحرف 98٪، والدقة المحققة للتعرف على الأحرف هي100 ٪ في ظروف مختلفة.

FINet: Handwriting trajectory reconstruction of Chinese characters based on the font imitate network

Recovering online handwriting trajectory from Chinese handwritten images is still a difficult problem because of abundant strokes, complex structures and different writing styles of Chinese characters. To address this problem, this paper proposes a handwriting trajectory reconstruction method for Chinese characters called the Font Imitate Network. The Font Imitate Network integrates two networks, the spatial encoder uses HRNet to encode the spatial association information of each position on the offline image; the temporal decoder consists of a multilayer perceptron and a recurrent neural network. The stroke feature network is designed to help predict the next handwriting point by focusing on the stroke path in the form of heatmap. Experiments are conducted on the CASIA-OLHWDB1.1 with DTW, LDTW, RMSE and character recognition accuracy as the evaluation metrics. The experimental results show our FINet outperforms the state-of-the-art methods and the recognition accuracy of offline characters is improved by the handwriting reconstruction.

Vehicle License Plate Detection and Recognition using OpenCV and Tesseract OCR

License plate recognition (LPR) is essential as the number of vehicles increases and the human ability to accomplish this task is limited. If human labor is used to manage these, it will take a lot of time and energy and cause a discrepancy. License Plate Recognition (LPR) is an advanced technology that leverages optical character recognition (OCR) and various image processing methods to read vehicle license plates automatically. Typically, an LPR system comprises two primary components: detecting vehicles and their license plates and recognizing the alphanumeric characters displayed on those plates. This study explores the use of OpenCV for license plate detection and Tesseract OCR for character recognition. In this research, the dataset for training and testing the system included 100 license plates evenly split between plates featuring English and Lao characters. The Lao license plates presented unique complexities due to their specific characteristics. The experimental setup involved processing images of license plates taken from multiple angles. The system's performance was evaluated based on the speed and accuracy of line and character recognition. For English character plates, the recognition process took 0.12 seconds with an accuracy of 98.8%. In contrast, the Lao character plates required 0.24 seconds, achieving an accuracy rate of 89.42%.

Research on efficient braille recognition based on the pair-U-shaped micro-nano optical fiber fingerprint-like tactile sensors

Tactile sensors are essential for capturing touch information, yet existing sensors struggle to accurately mimic human pressure and motion perception. This study introduces a novel design resembling a fingerprint, featuring a pair-U-shaped flexible optical fiber tactile sensor, and assesses its efficacy in braille input interactions. Constructed from 3D-printed elastomeric resin, the sensor incorporates a bilayer structure that mirrors a fingerprint pattern, with interlocking microstructures and encapsulated pair-U-shaped micro-nano-optical fibers (MNF) within a polydimethylsiloxane (PDMS) layer. These design elements allow these two MNF sensing channels to achieve synchronization, enhancing durability and sensitivity (66.08 % N−1 and 65.59 %N−1), quickening response (84.17/17.49 ms) and recovery time (58.31/10.02 ms) for detecting braille surfaces. Utilizing the Long Short-Term Memory (LSTM) algorithm significantly boosts the accuracy of braille character recognition. Post-training, the sensor achieved a 97.10 % recognition rate for 26 English letters under varying speeds and pressures. Moreover, the recognition rate for letters with similar features (B/K, F/M, G/X, H/U) improved to 98.58 %, underscoring the sensor’s effectiveness in braille recognition. This innovative approach holds potential for applications in braille reading, surface sensing, and related areas.

Fine Segmentation of Chinese Character Strokes Based on Coordinate Awareness and Enhanced BiFPN.

Considering the complex structure of Chinese characters, particularly the connections and intersections between strokes, there are challenges in low accuracy of Chinese character stroke extraction and recognition, as well as unclear segmentation. This study builds upon the YOLOv8n-seg model to propose the YOLOv8n-seg-CAA-BiFPN Chinese character stroke fine segmentation model. The proposed Coordinate-Aware Attention mechanism (CAA) divides the backbone network input feature map into four parts, applying different weights for horizontal, vertical, and channel attention to compute and fuse key information, thus capturing the contextual regularity of closely arranged stroke positions. The network's neck integrates an enhanced weighted bi-directional feature pyramid network (BiFPN), enhancing the fusion effect for features of strokes of various sizes. The Shape-IoU loss function is adopted in place of the traditional CIoU loss function, focusing on the shape and scale of stroke bounding boxes to optimize the bounding box regression process. Finally, the Grad-CAM++ technique is used to generate heatmaps of segmentation predictions, facilitating the visualization of effective features and a deeper understanding of the model's focus areas. Trained and tested on the public Chinese character stroke datasets CCSE-Kai and CCSE-HW, the model achieves an average accuracy of 84.71%, an average recall rate of 83.65%, and a mean average precision of 80.11%. Compared to the original YOLOv8n-seg and existing mainstream segmentation models like SegFormer, BiSeNetV2, and Mask R-CNN, the average accuracy improved by 3.50%, 4.35%, 10.56%, and 22.05%, respectively; the average recall rates improved by 4.42%, 9.32%, 15.64%, and 24.92%, respectively; and the mean average precision improved by 3.11%, 4.15%, 8.02%, and 19.33%, respectively. The results demonstrate that the YOLOv8n-seg-CAA-BiFPN network can accurately achieve Chinese character stroke segmentation.

Terminal strip detection and recognition based on improved YOLOv7-tiny and MAH-CRNN+CTC models

For substation secondary circuit terminal strip wiring, low efficiency, less easy fault detection and inspection, and a variety of other issues, this study proposes a text detection and identification model based on improved YOLOv7-tiny and MAH-CRNN+CTC terminal lines. First, the YOLOv7-tiny target detection model is improved by the introduction of the spatially invariant multi-attention mechanism (SimAM) and the weighted bidirectional feature pyramid network (BiFPN). This also improves the feature enhancements and feature fusion ability of the model, balances various scales of characteristic information, and increases the positioning accuracy of the text test box. Then, a multi-head attention hybrid (MAH) mechanism is implemented to optimize the convolutional recurrent neural network with connectionist temporal classification (CRNN+CTC) so that the model could learn data features with larger weights and increase the recognition accuracy of the model. The findings indicate that the enhanced YOLOv7-tiny model achieves 97.39%, 98.62%, and 95.07% of precision, recall, and mean average precision (mAP), respectively, on the detection dataset. The improved MAH-CRNN+CTC model achieves 91.2% character recognition accuracy in the recognition dataset.

Study On Advanced Vehicle Number Plate Detection Systems-A Review

In response to the rising volume of vehicles on roadways, there's an urgent demand for advanced technologies to bolster security and streamline traffic-related operations. This project introduces an innovative solution: an Advanced Automatic Number Plate Detection System. Leveraging Python, Easy OCR, and Open CV, this system aims to develop a sophisticated Automatic Number Plate Recognition (ANPR) mechanism capable of accurately detecting license plates, extracting alphanumeric characters, and providing actionable insights for various applications. Python is chosen for its versatility and ease of use, serving as the primary programming language. OpenCV is employed for its robust image processing tools and license plate detection capabilities, while Easy OCR simplifies character extraction from license plates. Key features include license plate detection through OpenCV algorithms, character recognition via Easy OCR, and real-time processing suitable for tasks like traffic monitoring and surveillance. The project follows a structured approach encompassing data collection, image pre-processing, machine learning model training, integration, and thorough testing. Anticipated outcomes include a fully operational ANPR system with high accuracy in license plate detection and character recognition, surpassing conventional methods. Its potential integration into real-world applications such as traffic management underscores its significance in addressing challenges posed by increasing vehicle numbers, with YOLO v8 utilized for precise outcomes. Keywords- Python, Open CV, YOLO, tensorflow, Machine learning, ANN, CNN, and Easy OCR

Advanced approach for Moroccan administrative documents digitization using pre-trained models CNN-based: character recognition

In the digital age, efficient digitization of administrative documents is a real challenge, particularly for languages with complex scripts such as those used in Moroccan documents. The subject matter of this article is the digitization of Moroccan administrative documents using pre-trained convolutional neural networks (CNNs) for advanced character recognition. This research aims to address the unique challenges of accurately digitizing various Moroccan scripts and layouts, which are crucial in the digital transformation of administrative processes. Our goal was to develop an efficient and highly accurate character recognition system specifically tailored for Moroccan administrative texts. The tasks involved comprehensive analysis and customization of pre-trained CNN models and rigorous performance testing against a diverse dataset of Moroccan administrative documents. The methodology entailed a detailed evaluation of different CNN architectures trained on a dataset representative of various types of characters used in Moroccan administrative documents. This ensured the adaptability of the models to real-world scenarios, with a focus on accuracy and efficiency in character recognition. The results were remarkable. DenseNet121 achieved a 95.78% accuracy rate on the Alphabet dataset, whereas VGG16 recorded a 99.24% accuracy on the Digits dataset. DenseNet169 demonstrated 94.00% accuracy on the Arabic dataset, 99.9% accuracy on the Tifinagh dataset, and 96.24% accuracy on the French Special Characters dataset. Furthermore, DenseNet169 attained 99.14% accuracy on the Symbols dataset. In addition, ResNet50 achieved 99.90% accuracy on the Character Type dataset, enabling accurate determination of the dataset to which a character belongs. In conclusion, this study signifies a substantial advancement in the field of Moroccan administrative document digitization. The CNN-based approach showcased in this study significantly outperforms traditional character recognition methods. These findings not only contribute to the digital processing and management of documents but also open new avenues for future research in adapting this technology to other languages and document types.

The Chinese lexicon of deaf readers: A database of character decisions and a comparison between deaf and hearing readers.

We present a psycholinguistic study investigating lexical effects on simplified Chinese character recognition by deaf readers. Prior research suggests that deaf readers exhibit efficient orthographic processing and decreased reliance on speech-based phonology in word recognition compared to hearing readers. In this large-scale character decision study (25 participants, each evaluating 2500 real characters and 2500 pseudo-characters), we analyzed various factors influencing character recognition accuracy and speed in deaf readers. Deaf participants demonstrated greater accuracy and faster recognition when characters were more frequent, were acquired earlier, had more strokes, displayed higher orthographic complexity, were more imageable in reference, or were less concrete in reference. Comparison with a previous study of hearing readers revealed that the facilitative effect of frequency on character decision accuracy was stronger for deaf readers than hearing readers. The effect of orthographic-phonological regularity differed significantly for the two groups, indicating that deaf readers rely more on orthographic structure and less on phonological information during character recognition. Notably, increased stroke counts (i.e., higher orthographic complexity) hindered hearing readers but facilitated recognition processes in deaf readers, suggesting that deaf readers excel at recognizing characters based on orthographic structure. The database generated from this large-scale character decision study offers a valuable resource for further research and practical applications in deaf education and literacy.

Improving accuracy and explainability of online handwritten character recognition

Improving accuracy and explainability of online handwritten character recognition

Build a Trained Data of Tesseract OCR engine for Tifinagh Script Recognition

This article introduces a methodology for constructing a trained dataset to facilitate Tifinagh script recognition using the Tesseract OCR engine. The Tifinagh script, widely used in North Africa, poses a challenge due to the lack of built-in recognition capabilities in Tesseract. To overcome this limitation, our approach focuses on image generation, box generation, manual editing, charset extraction, and dataset compilation. By leveraging Python scripting, specialized software tools, and Tesseract's training utilities, we systematically create a comprehensive dataset for Tifinagh script recognition. The dataset enables the training and evaluation of machine learning models, leading to accurate character recognition. Experimental results demonstrate high accuracy, precision, recall, and F1 score, affirming the effectiveness of the dataset and its potential for practical applications. The results highlight the robustness of the OCR system, achieving an outstanding accuracy rate of 99,97 %. The discussion underscores its superior performance in Tifinagh character recognition, exceeding the findings in the field. This methodology contributes significantly to enhancing OCR technology capabilities and encourages further research in Tifinagh script recognition, unlocking the wealth of information contained in Tifinagh documents

License plate Chinese character recognition based on ViT model

Transformer applications have been widely used in the computer vision field. Many related literatures show that the advantages of the model such as increased receptive field and globality are gradually emerging in image processing. However, with the popularity of the transformer, whether it can compete with the convolutional neural network (CNN) in terms of performance is still questionable and remains to be further studied. This paper will use the most basic structural model in the visual transformer (ViT) to classify and identify Chinese characters that are frequently used in the field of transportation and logistics and compare them with two classical CNN models. The results demonstrate that the performance of the transformer is obviously better than that of the traditional CNN structure, and the final accuracy of character recognition is higher than that of CNN, up to 98.66 %. It fully shows the infinite potential and excellent performance of the transformer in the area of computer vision and has high reliability and generalization ability.

Gradual OCR: An Effective OCR Approach Based on Gradual Detection of Texts

In this paper, we present a novel approach to optical character recognition that incorporates various supplementary techniques, including the gradual detection of texts and gradual filtering of inaccurately recognized texts. To minimize false negatives, we attempt to detect all text by incrementally lowering the relevant thresholds. To mitigate false positives, we implement a novel filtering method that dynamically adjusts based on the confidence levels of recognized texts and their corresponding detection thresholds. Additionally, we use straightforward yet effective strategies to enhance the optical character recognition accuracy and speed, such as upscaling, link refinement, perspective transformation, the merging of cropped images, and simple autoregression. Given our focus on Korean chart data, we compile a mix of real-world and artificial Korean chart datasets for experimentation. Our experimental results show that our approach outperforms Tesseract by approximately 7 to 15 times and EasyOCR by 3 to 5 times in accuracy, as measured using a Jaccard similarity-based error rate on our datasets.

Pixel-Level Degradation for Text Image Super-Resolution and Recognition

In the realm of image reconstruction, deep learning-based super-resolution (SR) has established itself as a prevalent technique, particularly in the domain of text image restoration. This study aims to address notable deficiencies in existing research, including constraints imposed by restricted datasets and challenges related to model generalization. Specifically, the goal is to enhance the super-resolution network’s reconstruction of scene text image super-resolution and utilize the generated degenerate dataset to alleviate issues associated with poor generalization due to the sparse scene text image super-resolution dataset. The methodology employed begins with the degradation of images from the MJSynth dataset, using a stochastic degradation process to create eight distinct degraded versions. Subsequently, a blank image is constructed, preserving identical dimensions to the low-resolution image, with each pixel sourced randomly from the corresponding points across the eight degraded images. Following several iterations of training via Finetune, the LR-HR method is applied to the TextZoom dataset. The pivotal metric for assessment is optical character recognition (OCR) accuracy, recognized for its fundamental role in gauging the pragmatic effectiveness of this approach. The experimental findings reveal a notable enhancement in OCR accuracy when compared to the TBSRN model, yielding improvements of 2.4%, 2.3%, and 4.8% on the TextZoom dataset. This innovative approach, founded on pixel-level degradation, not only exhibits commendable generalization capabilities but also demonstrates resilience in confronting the intricate challenges inherent to text image super-resolution.