An Open Source Tesseract Based Optical Character Recognizer for Bangla Script

Hasnat, Md. Abul; Chowdhury, Muttakinur Rahman; Khan, Mumit

doi:10.1109/icdar.2009.62

Cited by 25 publications

(12 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…17,18 Although the accuracy of the character recognition depends on the image conditions, some studies using Tesseract OCR have reported 70% or higher accuracy for grayscale images. 19,20 Because the text recorded in this study contains only numeric characters, some confusing alphabetical characters and symbols (i.e., o, l, I, and B) were automatically replaced by numeric characters to avoid the recognition failure. The values were collected every 200 ms and the median of five values was collected every second to eliminate any errors due to image lag.…”

Section: C Beam Linearity and Consistencymentioning

confidence: 99%

Characteristics of flattening filter free beams at low monitor unit settings

et al. 2013

View full text Add to dashboard Cite

show abstract

Section: C Beam Linearity and Consistencymentioning

confidence: 99%

Characteristics of flattening filter free beams at low monitor unit settings

et al. 2013

View full text Add to dashboard Cite

show abstract

“…Future works could also analyze the impact of an automatic correction method based on machine learning. As proposed in [Hasnat et al 2009], some correction methods can be implemented to correct spelling mistakes based on information that has a specific format and predefined rules, such as, date, hour, total amount and etc.…”

Section: Discussionmentioning

confidence: 99%

“…Hasnat et al [Hasnat et al 2009] designed an OCR process software for the Bengali language in combination with the Tesseract library, which was called BanglaOCR. This paper focused mainly on Tesseract training and post-processing techniques.…”

Section: Related Workmentioning

confidence: 99%

“…However, depending on image quality and character font used to write the receipt, these two techniques may not be enough. As a way to circumvent this fact, previous works [Hasnat et al 2009] [Bassil and Alwani 2012] proposed a step called post-processing, known as a way to enhance information obtained by OCR through the use of online or offline spelling checkers. Since internet access is not available all the time for the users, any application that performs digital image processing, OCR and post-processing shall let it into account.…”

Section: Motivations and Main Objectivesmentioning

confidence: 99%

“…Some work has been done to improve Digital Image Processing, Optical Character Recognition and Post-Processing techniques such as the proposed by Hang [Heng 2013], Hasnat et. al.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Smart Control of Expenses Using Mobile Devices

Giovani¹,

Amorim²,

Gomes³

et al. 2016

Anais Do XLIII Seminário Integrado De Software E Hardware (SEMISH 2016)

View full text Add to dashboard Cite

It is estimated that only 3.1% of the Brazilian population controls their expenses through digital applications. While 8.9% does not use a digital platform due to a lack of knowledge, 8.1% do not have time. Considering the current usage levels, applications providing a more automated control of expenditures would simplify this task for an average user, making mobile applications a more attractive option. Using digital image processing techniques, optical character recognition (OCR), and post-processing a novel software was developed in Android for receipts automatic recognition allowing mobile users to monitor expenses using photos. Information recognized by the application replicated a real receipt with an acceptable level of similarity.

show abstract