Method for improving quality of recognition of a single frame

1. A method, comprising:

receiving, by a processing device, an original image of a document;

identifying location of a first symbol sequence in the original image;

applying a first set of conversion operations to a first portion of the original image comprising the location of the first symbol sequence to generate a first converted portion of the original image, wherein the first set of conversion operations comprises a first conversion operation that changes a first image setting, and wherein the first image setting comprises at least one of a resolution setting, a binarization method, a correction of distortion settings, glare correction settings, fuzziness correction settings, or noise reduction;

performing optical character recognition (OCR) of the first converted portion of the original image to produce a first recognized first symbol sequence;

applying a second set of conversion operations to the first portion of the original image comprising the location of the first symbol sequence to generate a second converted portion of the original image, wherein the second set of conversion operations is different than the first set of conversion operations, wherein the second set of conversion operations comprises a second conversion operation that changes a second image setting, wherein the second image setting comprises at least one of the resolution setting, the binarization method, the correction of distortion settings, the glare correction settings, the fuzziness correction settings, or the noise reduction, and wherein the second image setting is different from the first image setting;

performing optical character recognition (OCR) of the second converted portion of the original image to produce a second recognized first symbol sequence; and

combining the first recognized first symbol sequence and the second recognized first symbol sequence to produce a resulting first symbol sequence.

2. The method of claim 1 further comprising

applying a third set of conversion operations to the first portion of the original image comprising the location of the first symbol sequence to generate a third converted portion of the original image;

performing optical character recognition (OCR) of the third converted portion of the original image to produce a third recognized first symbol sequence; and

combining the first recognized first symbol sequence, the second recognized first symbol sequence, and the third recognized first symbol sequence to produce the resulting first symbol sequence.

3. The method of claim 2 wherein the set of conversion operations comprises a first conversion operation with a first set of conversion parameters the second conversion operation with a second set of conversion parameters.

4. The method of claim 2 further comprising performing the steps of

applying next set of conversion operations to the first portion of the original image comprising the location of the first symbol sequence to generate another converted portion of the original image;

performing optical character recognition (OCR) of the other converted portion of the original image to produce another recognized first symbol sequence;

predetermined number of times.

5. The method of claim 2 further comprising

detecting quality of the resulting first symbol sequence;

comparing the detected quality of the resulting first symbol sequence to a predetermined quality threshold;

in response to determination that the detected quality of the resulting first does not comply with the predetermined quality threshold, applying a next set of conversion operations to the first symbol sequence to generate another converted portion of the original image;

performing optical character recognition (OCR) of the other converted first symbol sequence to produce another recognized first symbol sequence;

combining the other recognized first symbol sequence with previous recognized first symbol sequence to produce an updated resulting first symbol sequence.

6. The method of claim 1 wherein the combining of the first recognized first symbol sequence and the second recognized first symbol sequence comprises

identifying a median string representing the first symbol sequence.

7. The method of claim 6 wherein the identifying a median string representing the first symbol sequence comprises

identifying edit distance between the first recognized first line of text symbol sequence and the second recognized first symbol sequence.

8. The method of claim 6 wherein the identifying a median string representing the first symbol sequence comprises

identifying Levenshtein distance between the first recognized first symbol sequence and the second recognized first symbol sequence.

9. The method of claim 1 , wherein the first symbol sequence is provided in a first natural language, the method further comprising:

translating the resulting first symbol sequence into a second natural language.

10. A system, comprising:

a memory;

a processing device, coupled to the memory, the processing device configured to:

receive, by a processing device, an original image of a document;

identify location of a first symbol sequence in the original image;

apply a first set of conversion operations to a first portion of the original image comprising the location of the first symbol sequence to generate a first converted portion of the original image, wherein the first set of conversion operations comprises a first conversion operation that changes a first image setting, and wherein the first image setting comprises at least one of a resolution setting, a binarization method, a correction of distortion settings, glare correction settings, fuzziness correction settings, or noise reduction;

perform optical character recognition (OCR) of the first converted portion of the original image to produce a first recognized first symbol sequence;

apply a second set of conversion operations to the first portion of the original image comprising the location of the first symbol sequence to generate a second converted portion of the original image, wherein the second set of conversion operations is different than the first set of conversion operations, wherein the second set of conversion operations comprises a second conversion operation that changes a second image setting, wherein the second image setting comprises at least one of the resolution setting, the binarization method, the correction of distortion settings, the glare correction settings, the fuzziness correction settings, or the noise reduction, and wherein the second image setting is different from the first image setting;

perform optical character recognition (OCR) of the second converted portion of the original image to produce a second recognized first symbol sequence; and

combine the first recognized first symbol sequence and the second recognized first symbol sequence to produce a resulting first symbol sequence.

11. The system of claim 10 further configured to

apply a third set of conversion operations to the first portion of the original image comprising the location of the first symbol sequence to generate a third converted portion of the original image;

perform optical character recognition (OCR) of the third converted portion of the original image to produce a third recognized first symbol sequence; and

combine the first recognized first symbol sequence, the second recognized first symbol sequence, and the third recognized first symbol sequence to produce the resulting first symbol sequence.

12. The system of claim 11 wherein the set of conversion operations comprises a first conversion operation with a first set of conversion parameters the second conversion operation with a second set of conversion parameters.

13. The system of claim 11 further configured to

apply next set of conversion operations to the first portion of the original image comprising the location of the first symbol sequence to generate another converted portion of the original image;

perform optical character recognition (OCR) of the other converted portion of the original image to produce another recognized first symbol sequence;

predetermined number of times.

14. The system of claim 11 further configured to

detect quality of the resulting first symbol sequence;

compare the detected quality of the resulting first symbol sequence to a predetermined quality threshold;

in response to determination that the detected quality of the resulting first symbol sequence does not comply with the predetermined quality threshold, apply next set of conversion operations to the first symbol sequence to generate another converted portion of the original image;

perform optical character recognition (OCR) of the other converted first symbol sequence to produce another recognized first symbol sequence;

combine the other recognized first symbol sequence with previous recognized first symbol sequence to produce an updated resulting first symbol sequence.

15. The system of claim 10 , wherein the first symbol sequence is provided in a first natural language, the system further configured to translate the resulting first symbol sequence into a second natural language.

16. A computer-readable non-transitory storage medium comprising executable instructions that, when executed by a processing device, cause the processing device to:

receive an original image of a document;

identify location of a first symbol sequence in the original image;

apply a first set of conversion operations to a first portion of the original image comprising the location of the first symbol sequence to generate a first converted portion of the original image, wherein the first set of conversion operations comprises a first conversion operation that changes a first image setting, and wherein the first image setting comprises at least one of a resolution setting, a binarization method, a correction of distortion settings, glare correction settings, fuzziness correction settings, or noise reduction;

perform optical character recognition (OCR) of the first converted portion of the original image to produce a first recognized first symbol sequence;

apply a second set of conversion operations to the first portion of the original image comprising the location of the first symbol sequence to generate a second converted portion of the original image, wherein the second set of conversion operations is different than the first set of conversion operations, wherein the second set of conversion operations comprises a second conversion operation that changes a second image setting, wherein the second image setting comprises at least one of the resolution setting, the binarization method, the correction of distortion settings, the glare correction settings, the fuzziness correction settings, or the noise reduction, and wherein the second image setting is different from the first image setting;

perform optical character recognition (OCR) of the second converted portion of the original image to produce a second recognized first symbol sequence; and

combine the first recognized first symbol sequence and the second recognized first symbol sequence to produce a resulting first symbol sequence.

17. The computer-readable non-transitory storage medium of claim 16 further comprising executable instructions that, when executed by a processing device, cause the processing device to

apply a third set of conversion operations to the first portion of the original image comprising the location of the first symbol sequence to generate a third converted portion of the original image;

perform optical character recognition (OCR) of the third converted portion of the original image to produce a third recognized first symbol sequence; and

combine the first recognized first symbol sequence, the second recognized first symbol sequence, and the third recognized first symbol sequence to produce the resulting first symbol sequence.