Skip to content

Rule-based OCR🔗

Optical Character Recognition (OCR) is the technical term for reading, i.e., identifying symbols. In HALCON, OCR is defined as the task to assign an interpretation to regions of an image. This chapter considers methods where these regions typically represent single characters and therefore we consider this as reading single symbols. With Deep OCR, a different approach is available.

In an offline phase, the characters are trained by presenting several samples for each character. In the online phase, the image is segmented to extract the regions representing the characters and then the OCR reader is applied to get the interpretation for each character.

The advantage of OCR is the flexibility of the training, which allows to select features optimized for an application. Furthermore, you can choose between different classifiers.

As a further advantage, HALCON provides you with a set of pretrained fonts, which are based on a large amount of training data from various application areas. These fonts allow you to read text in documents, on pharmaceutical or industrial products, dot prints, and even handwritten numbers. Furthermore, HALCON includes pretrained fonts for OCR-A and OCR-B, and a generic font based on Convolutional Neural Networks (CNNs).