The Optical Character Reader has traditionally been well-known in the area of scanning of handwritten documents (preprinted such as utility bills filled in with meter readings by human readers) and process the numbers or text from the scanning process into computer readable formats through software. The OCR is one of the best methods to use when there is a need for the capture of neat handwritten documents. SAT tests, electronic bill calculation and MCQ quizzes are part of the applications of the OCR. This paper will however, research the OCR comparing it with other available methods/devices for data capture and evaluate the usefulness of the OCR against them.
THE RESEARCH
GOCR: Historically, GOCR software has not been one of the toppers in this field. With high error rates in character recognition (98% for version 0.4), it is just worth giving a test try at most. Although the subsequent versions had those bugs fixed, the efficiency of GOCR has always been lower than the other OCR software. GOCR works in two modes: reading off black text off white backgrounds as well as reading off white text off black backgrounds. The latter was however, more difficult to program for the developers and still has high margin of errors. Its ability to recognize handwritten characters with a lot of deviation is poor.
Although much work has been done in the later versions to improve this, optical character recognition accuracy is still one of the biggest issues for GOCR. The GOCR has the highest number of characters recognized incorrectly. Therefore, “I”’s are recognized wrongly as “l”’s and “v”’s are recognized as “u”’s. GOCR is useful in situations where the handwriting is exceptionally neat or the document error rate is not a matter of concern (which of course will be a rare case).
Also it should be noted that GOCR is open-source software. This means that GOCR code is readily available free of cost. Therefore, different versions floating around are actually revisions by different programmers on the basis of their knowledge. Thus, GOCR offers a few features that are unique: the ability to work with a different variety of formats of images (which is also found in others, but with one or two omissions).
Tesseract OCR: One of the reviews of this software went like this: “It sounds like it Tesseract OCR I unusable at the current moment, but the developments made by Google in the subsequent versions leave a promising note for the future.”
In short, Tesseract is one of those open source optical character reading software that is not considered to be one of the most efficient software suites. In fact, one of the drawbacks of Tesseract is the command line interface with the user. This seems most absurd for software that deals with pictures and graphics. However, the software is configured to accept picture or graphics from hardware and then automatically read it and transform it into text. This OCR software has yet to come to a more user friendly state for it to be more popular as well as efficient. The error rate of 25% is not very low. However, considering the interface, this is quite an achievement for developers who could not or simply did not want to make an interactive software to achieve 75% success in recognizing handwritten characters.
The main aim of an OCR is to reduce the need for manual typing in either large volumes or in the case of transfer of data (programmers often use it as disaster recovery plans).
Thus it is necessary that the software for OCR be fool-proof to high degree. It should provide accuracy, reliability as well as be able to work with deviations from normal. Both GOCR and Tesseract OCR are able to recognize characters in printed images and files, but the main problem arises in handwriting. Apart from that too, there are evident problems described above. Thus, the third software for comparison, Microsoft Document Imaging, coupled with Microsoft Word OCR capabilities overcomes these problems and provides a cost effective solution. It should be noted that this is not single software. Instead there are two complementary software working towards reaching the desired result.
Microsoft Document Imaging with Microsoft Word OCR capability: This is without any doubt the strongest software suit available for the purpose of accurate and effective optical character recognition. Microsoft document Imaging provides the first process service: scanning the image and making it ready for the software by converting the physical input to a machine readable format. In the next step the scanned file is given to Microsoft Word. This has in-built OCR capabilities that are strong, efficient as well as user-friendly. The OCR recognizes the characters with up to a 98% accuracy level, allowing a very small room for errors. Apart from that, this software also provides the capability of recognizing handwritten characters to stay at par with the standard OCR software.
It is an effective software solution with a powerful interface, speedy solution as well as a cost-effective solution. It has all the features that make it overcome the problems discussed in GOCR and Tesseract OCR. Microsoft Document Imaging and Word complement each other perfectly and thus they are the choice of any OCR requiring scenario. The capability of this suite to handle images with accuracy and speed drives up its performance greatly.
It is clear from the above comparisons done that the third option has the strongest and most cost effective benefits over the other software. It is quickest and the cheapest form of input of handwritten characters into the computer so that they can also be edited. It also provides the best recognition capability of deviating characters.