What is Tesseract?

Tesseract OCR (Optical Character Recognition) is an open-source software library that enables the recognition and extraction of text from various sources, such as images, scanned documents, and PDF files. It was initially developed at Hewlett-Packard Laboratories in the 1980s, and later maintained and enhanced by different organizations.

Tesseract OCR utilizes machine learning and pattern recognition techniques to analyse the visual patterns of characters and convert them into machine-readable text. As of today it supports over 100 languages, including major scripts like Latin, Cyrillic, Arabic, and Chinese. The library has been widely adopted in various applications and industries. It provides APIs and command-line interfaces, making it accessible for developers to integrate into their software systems. Tesseract OCR can be used for tasks such as document digitization, text extraction from images for indexing and searching, automatic data entry, and more.

How it works?

Tesseract's pattern recognition technology involves several key steps. First, it performs image preprocessing (adaptive thresholding), which includes tasks like binarization, noise removal, and deskewing to enhance the quality of the input image. Next, Tesseract divides the resulting binary image into smaller chunks and identifies individual characters or text regions through a process known as segmentation. Then, using a combination of statistical models and neural networks, Tesseract compares these segmented patterns with its vast knowledge base of character and language models to recognize and interpret the text accurately.

Tesseract OCR assigns a confidence score to each recognized character or word as part of its output. The confidence score represents the level of certainty or reliability of the recognition result. It indicates how confident Tesseract is in its recognition accuracy for a particular character or word. The confidence score typically ranges from 0 to 100, with higher scores indicating higher confidence. In the scope of this project, we will be using the confidence score to determine how well the developed font disguises itself from the Tesseract OCR engine.

Limitations of used implementation

While tesseract.js offers powerful OCR capabilities, it also has some noteworthy limitations:

1. Accuracy: While Tesseract OCR is known for its high accuracy, the accuracy of tesseract.js may be slightly lower compared to the original Tesseract engine. This can be attributed to factors such as the limitations of running OCR in a browser environment, variations in image quality, and the inherent complexities of character recognition.
2. Performance: Tesseract.js heavily relies on WebAssembly for running the OCR engine in the browser. While WebAssembly provides near-native performance, it may still be slower compared to running Tesseract OCR on a dedicated server or desktop application. Large images or complex documents can lead to longer processing times.
3. Language Support: Tesseract OCR supports a wide range of languages; however, not all languages have the same level of accuracy and recognition quality. In the scope of this project, we only implement English and German text recognition.
4. Continuous Development: Tesseract.js is an open-source project that relies on community contributions and updates. While it has seen active development and improvements, it's essential to keep track of the latest versions and updates to benefit from bug fixes, performance enhancements, and new features. This project’s testing environment runs on tesseract.js v4.