6

I've just found the term "Intelligent Character Recognition" (ICR) on Wikipedia and other pages. According to Wikipedia:

In computer science, intelligent character recognition (ICR) is an advanced optical character recognition (OCR) or — rather more specific — handwriting recognition system that allows fonts and different styles of handwriting to be learned by a computer during processing to improve accuracy and recognition levels.

Is this just a marketing stunt or are there actually techniques which are specified as OCR and other techniques which are ICR? Are there Conferences / Journals about OCR which distinguish between OCR and ICR? (If yes, please add a reference)

Martin Thoma
  • 19,540
  • 36
  • 98
  • 170

1 Answers1

4

Clarifying the Distinction between OCR and ICR

TL;DR: In think the term "Intelligent Character Recognition" (ICR) mostly serves as a marketing term rather than representing a distinct, academically recognised subfield separate from Optical Character Recognition (OCR).

Validity and Terminology

Optical Character Recognition (OCR) encompasses technologies that convert images of printed and handwritten text into machine-readable formats. Within this domain, Intelligent Character Recognition (ICR) often refers to advanced OCR techniques, particularly those involving handwriting recognition. The term "ICR" is frequently used in commercial contexts to highlight systems capable of interpreting diverse handwriting styles. For instance, Carbune et al (2020) describe ICR as the task of deciphering digitised handwritten text, noting its increased complexity compared to OCR due to the variability in handwriting.

Academic Perspective

In academic discourse, OCR methodologies are typically categorised based on input modalities rather than marketing terminology. The primary distinctions are:

  • Online Handwriting Recognition: This involves the real-time interpretation of handwriting as it is written, typically using a stylus or touchscreen. This method captures dynamic data streams that include not only the spatial coordinates of the pen but also temporal features such as stroke sequence, direction, speed, and sometimes pressure or tilt. These temporal features provide additional discriminative power, enabling higher accuracy, especially for cursive or ambiguous handwriting styles.

  • Offline Handwriting Recognition: This refers to the analysis of handwriting captured as static images, typically scanned from paper documents or photographs. Since the input is a two-dimensional bitmap without temporal information, the recognition process relies solely on the spatial features of the handwriting, such as character shapes, contours, pixel density, and contextual elements. This task is generally considered more challenging due to the absence of information about the writing process (eg., stroke order or direction).

References

Carbune, V., Gonnet, P., Deselaers, T., Rowley, H. A., Daryin, A., Calvo, M., ... & Keysers, D. (2020). Fully convolutional networks for handwriting recognition. arXiv preprint arXiv:1907.04888. https://arxiv.org/abs/1907.04888

IEEE. (n.d.). IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI). Retrieved March 24, 2025, from https://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=34

International Conference on Document Analysis and Recognition. (n.d.). ICDAR 2024. Retrieved March 24, 2025, from https://icdar2024.net/

International Journal on Document Analysis and Recognition (IJDAR). (n.d.). Springer. Retrieved March 24, 2025, from https://www.springer.com/journal/10032

Pattern Recognition Journal. (n.d.). Elsevier. Retrieved March 24, 2025, from https://www.sciencedirect.com/journal/pattern-recognition

Robert Long
  • 3,518
  • 12
  • 30