Optical character recognition

About: Optical character recognition is a research topic. Over the lifetime, 7342 publications have been published within this topic receiving 158193 citations. The topic is also known as: OCR & optical character reader.

Approximate string matching algorithms for limited-vocabulary OCR output correction

Abstract: Five methods for matching words mistranslated by optical character recognition to their most likely match in a reference dictionary were tested on data from the archives of the National Library of Medicine. The methods, including an adaptation of the cross correlation algorithm, the generic edit distance algorithm, the edit distance algorithm with a probabilistic substitution matrix, Bayesian analysis, and Bayesian analysis on an actively thinned reference dictionary were implemented and their accuracy rates compared. Of the five, the Bayesian algorithm produced the most correct matches (87%), and had the advantage of producing scores that have a useful and practical interpretation.

Towards a real-time system for finding and reading signs for visually impaired users

Abstract: Printed text is a ubiquitous form of information that is inaccessible to many blind and visually impaired people unless it is represented in a non-visual form such as Braille. OCR (optical character recognition) systems have been used by blind and visually impaired persons for some time to read documents such as books and bills; recently this technology has been packaged in a portable device, such as the smartphone-based kReader Mobile (from K---NFB Reading Technology, Inc.), which allows the user to photograph a document such as a restaurant menu and hear the text read aloud. However, while this kind of OCR system is useful for reading documents at close range (which may still require the user to take a few photographs, waiting a few seconds each time to hear the results, to take one that is correctly centered), it is not intended for signs. (Indeed, the KNFB manual, see knfbreader.com/upgrades_mobile.php , lists "posted signs such as signs on transit vehicles and signs in shop windows" in the "What the Reader Cannot Do" subsection.) Signs provide valuable location-specific information that is useful for wayfinding, but are usually viewed from a distance and are difficult or impossible to find without adequate vision and rapid feedback. We describe a prototype smartphone system that finds printed text in cluttered scenes, segments out the text from video images acquired by the smartphone for processing by OCR, and reads aloud the text read by OCR using TTS (text-to-speech). Our system detects and reads aloud text from video images, and thereby provides real-time feedback (in contrast with systems such as the kReader Mobile) that helps the user find text with minimal prior knowledge about its location. We have designed a novel audio-tactile user interface that helps the user hold the smartphone level and assists him/her with locating any text of interest and approaching it, if necessary, for a clearer image. Preliminary experiments with two blind users demonstrate the feasibility of the approach, which represents the first real-time sign reading system we are aware of that has been expressly designed for blind and visually impaired users.

Optical character recognition (OCR)

Abstract: Optical character recognition (OCR) is performed by optical character readers which are automated electronic systems. OCR may be defined as the process of converting images of machine printed or handwritten numerals, letters, and symbols into a computer- processable format. The long history of research in this area, commercial success, and the continuing need and ability to handle less restricted forms of text make OCR the most important application area in machine perception to date.

Word-wise Script Identification from Bilingual Documents Based on Morphological Reconstruction

Abstract: In a multi-lingual country like India, english has proven to be the binding language. So, a line of a bilingual document page may contain text words in regional language and numerals in English (printed or handwritten). For optical character recognition (OCR) of such a document page, it is necessary to identify different script forms before running an individual OCR of the scripts. In this paper an automatic technique for script identification at word level based on morphological reconstruction is proposed for two printed bilingual documents of Kannada and Devnagari containing English numerals (printed and handwritten). The technique developed includes a feature extractor and a classifier. The feature extractor consists of two stages. In the first stage, shape (eccentricity, aspect ratio) and directional stroke features (horizontal and vertical) are extracted based on morphological erosion and opening by reconstruction using the line structuring element. The average height of all the connected components of an image is used to threshold the length of the structuring element. In the second stage, average pixel distribution is obtained from these resulting images. The k-nearest neighbour algorithm is used to classify the new word images. The proposed algorithm is tested on 2250 sample words with various font styles and sizes. The results obtained are quite encouraging.

Writer adaptation of online handwriting models

Abstract: Writer adaptation is the process of converting a writer-independent handwriting recognition system, which models the characteristics of a large group of writers, into a writer-dependent system, which is tuned for a particular writer. Adaptation has the potential of increasing recognition accuracies, provided adequate models can be constructed for a particular writer. The limited amount of data that a writer typically provides makes the role of writer-independent models crucial in the adaptation process. Our approach to writer-adaptation makes use of writer-independent writing style models (called lexemes), to identify the styles present in a particular writer's training data. These models are then retrained using the writer's data. We demonstrate the feasibility of this approach using hidden Markov models trained on a combination of discretely and cursively written lower case characters. Our results show an average reduction in error rate of 16.3% for lower case characters as compared against representing each of the writer's character classes with a single model.