![abbyy cli ocr for linux abbyy cli ocr for linux](https://www.siteprice.org/SiteThumbs/o/ocrsdk.com.jpeg)
The table is arranged from highest ratio to lowest according to the Levenshtein ratio when using the language recognition key. In almost all pages the OCR quality was better when using language recognition. The following table shows the Levenshtein ratios for each page of the libretto when using and not using the language recognition key. So a page with a high Levenshtein ratio means that ABBYY’s output for that page is very similar to the ground truth, which means a high OCR quality for that page and a low ratio means a low OCR quality. In this case, the first sequence was conformed by all the text contained in the ground truth file for a single page in the libretto, and the second sequence was conformed by all the text in ABBYY’s output file corresponding to that same page. This ratio gives a measure of similarity between two sequences of characters.
![abbyy cli ocr for linux abbyy cli ocr for linux](https://1.bp.blogspot.com/-emtFlgMk3Ks/XyM9wgXbyYI/AAAAAAAANXo/PxWIlZ3giJY2TNeLfL9cb8nZWFg-67YTwCLcBGAsYHQ/w640-h360/ABBYY%2BFineReader%2BPDF%2B15%2BBuild%2B15.0.113.3886%2BRepack.png)
These numeric results were obtained by comparing ABBYY’s output file of a single page to a ground truth file of the same page using the Levenshtein ratio. Table 1 shows the quantitative analysis done to evaluate the output. Its performance improved when using the Italian dictionary. When there is no key for language recognition, English is used.Īt first glance, ABBYY’S output was understandable even if it wasn’t perfect. The libretto from the opera Aroldo was used to test the accuracy of ABBYY’s output. We got two outputs for every page by using the default instruction (without any key) and the instruction with the input key for language recognition, in this case the language was Italian (key: -rl Italian). And see how you can take more than one input file, or have just one input image and produce different output files (i.e. Here you will be able to see all the input and output keys available. So only the input keys may be able to improve the OCR. At the moment the only input key tested properly, by this I mean a quantitative testing and not only a qualitative one, is the recognition language key, which actually improves the result. The output keys are just export options, they affect the layout and coding of the output file for a particular format, this is not relevant for our purpose of evaluating the quality of the OCR process.
![abbyy cli ocr for linux abbyy cli ocr for linux](https://www.howtogeek.com/wp-content/uploads/2020/07/13-3.png)
table and barcode analysis, recognition keys). There are many keys (options) for the input file, which involve image processing and analysis (ex. The output file is the export file and it can be in any of these formats: The input file is an image that can be in any of the following formats: The input and output file must be specified. The -f flag indicates the format of the output file, by specifying the format of the file you can now insert output options that have to do with this particular file type. They must be followed by the file name and be preceded by the options So they are supposed to be followed by the (input or output) file name, and also to be preceded by any options to process the file (this are not compulsory). The flags -if and -of stand for input file and output file, respectively. So every statement begins with the word abbyyocr11 The command-line instruction to ask ABBYY to do OCR on an image has this base structure:Ībbyyocr11 -if -f format -of We are using ABBYY FineReader Engine, version 11, in its Command Line Interface (CLI) for Linux. We have been testing how well ABBYY does OCR on a Italian opera libretto. ABBYY FineReader is an optical character recognition (OCR) software.