Text line image splitting with different font sizes

1. A method comprising:

(a) receiving a text line image associated with a line of text contained within a document image;

(b) identifying that the text line image comprises a plurality of zones, wherein each zone contains text whose font differs from the text of each adjacent zone by having a different vertical position or font size than each adjacent zone;

(c) selecting at least one splitting position between multiple zones of the text line image;

(d) splitting the text line image at the splitting position into a plurality of image segments, wherein each image segment contains at least one zone of the text line image; and

(e) performing optical character recognition (OCR) on each image segment to recognize a corresponding text segment of each image segment.

2. The method of claim 1 , further comprising:

combining the text segments to create a text of the text line image.

3. The method of claim 1 , wherein steps (b) and (c) further comprise:

performing OCR on the text line image;

generating an OCR confidence measurement comprising a predicted OCR accuracy of the text line image for a plurality of positions of the text line image; and

selecting a splitting position within the text line image based on the OCR confidence measurement.

4. The method of claim 3 , wherein the plurality of positions of the text line image include positions corresponding to one or more words contained in the text line image.

5. The method of claim 1 , further comprising repeating steps (a) to (e) on a plurality of text line images associated with the document image.

6. A method comprising:

(a) receiving a text line image associated with a line of text contained within a document image;

(b) identifying that the text line image comprises a plurality of zones, wherein each zone contains text whose font differs from the text of each adjacent zone;

(c) selecting at least one splitting position between multiple zones of the text line image;

(d) splitting the text line image at the splitting position into a plurality of image segments, wherein each image segment contains at least one zone of the text line image;

(e) performing optical character recognition (OCR) on each image segment to recognize a corresponding text segment of each image segment; and

(f) combining the text segments to create a text of the text line image;

wherein steps (b) and (c) further comprise:

performing OCR on the text line image;

generating an OCR confidence measurement comprising a predicted OCR accuracy of the text line image for a plurality of positions of the text line image; and

selecting a splitting position within the text line image based a gradient of the OCR confidence measurement.

7. The method of claim 6 , wherein identifying that the text line image comprises a plurality of zones, wherein each zone contains text whose font differs from the text of each adjacent zone, further comprises, at least one selected from the group consisting of:

identifying that the text whose font differs from the text of each of the adjacent zones has a different size than the text in the adjacent zones,

identifying that the text whose font differs from the text of each of the adjacent zones has a different typeface than the text in the adjacent zones, and

identifying that the text whose font differs from the text of each of the adjacent zones has a different vertical position within the text line image than the text in the adjacent zones.

8. A method comprising:

(a) receiving a text line image associated with a line of text contained within a document image;

(b) identifying that the text line image comprises a plurality of zones, wherein each zone contains text whose font differs from the text of each adjacent zone;

(c) selecting at least one splitting position between multiple zones of the text line image;

(d) splitting the text line image at the splitting position into a plurality of image segments, wherein each image segment contains at least one zone of the text line image; and

(e) performing optical character recognition (OCR) on each image segment to recognize a corresponding text segment of each image segment,

wherein steps (b) and (c) further comprise:

estimating a text height of the text line image;

generating a text height confidence measurement comprising a predicted accuracy of the estimated text height for a plurality of positions of the text line image; and

selecting a splitting position within the text line image based on the text height confidence measurement.

9. The method of claim 8 , wherein the splitting position is selected based on a gradient of the text height confidence measurement.

10. The method of claim 8 , wherein the plurality of positions of the text line image include positions corresponding to one or more words contained in the text line image.

11. A method comprising:

(a) receiving a text line image associated with a line of text contained within a document image;

(b) identifying that the text line image comprises a plurality of zones, wherein each zone contains text whose font differs from the text of each adjacent zone;

(c) selecting at least one splitting position between multiple zones of the text line image;

(d) splitting the text line image at the splitting position into a plurality of image segments, wherein each image segment contains at least one zone of the text line image; and

(e) performing optical character recognition (OCR) on each image segment to recognize a corresponding text segment of each image segment,

wherein steps (b) to (d) are repeated on at least one of the image segments to select additional splitting positions of the text line image and to split the text line image into additional image segments.

12. A system comprising:

a processor; and

a memory storing instructions which, when executed by the processor, cause the processor to:

(a) receive a text line image associated with a line of text contained within a document image;

(b) identify that the text line image comprises a plurality of zones, wherein each zone contains text whose font differs from the text of each adjacent zone by having a different vertical position or font size than each adjacent zone;

(c) select a splitting position between multiple zones of the text line image;

(d) split the text line image at the splitting position into a plurality of image segments, wherein each image segment contains at least one zone of the text line image; and

(e) perform optical character recognition (OCR) on each image segment to recognize a corresponding text segment of each image segment.

13. The system of claim 12 , wherein the memory contains further instructions which, when executed by the processor, cause the processor to:

combine the text segments to create a text of the text line image.

14. The system of claim 12 , wherein the memory contains further instructions which, when executed by the processor at steps (b) and (c), cause the processor to:

perform OCR on the text line image;

generate an OCR confidence measurement comprising a predicted OCR accuracy of the text line image for a plurality of positions of the text line image; and

select a splitting position within the text line image based on the OCR confidence measurement.

15. The system of claim 12 , wherein the memory contains further instructions which, when executed by the processor at steps (b) and (c), cause the processor to:

estimate a text height of the text line image;

receive a text height confidence measurement comprising a predicted accuracy of the text height for a plurality of positions of the text line image; and

select a splitting position within the text line image based on the text height confidence measurement.

16. The system of claim 15 , wherein the memory contains further instructions which, when executed by the processor at step (c), cause the processor to select the splitting position based on a gradient of the text height confidence measurement.

17. The system of claim 12 , wherein the memory contains further instructions which, when executed by the processor, cause the processor to repeat steps (b) to (d) on at least one of the image segments to select additional splitting positions of the text line image and to split the text line image into additional image segments.

18. The system of claim 12 , wherein the system is further configured, when executed by the processor, to repeat steps (a) to (e) on a plurality of text line images associated with the document image.

19. A system, comprising:

a processor; and

a memory storing instructions which, when executed by the processor, cause the processor to:

(a) receive a text line image associated with a line of text contained within a document image;

(b) perform OCR on the text line image;

(c) generate an OCR confidence measurement comprising a predicted OCR accuracy of the text line image for a plurality of positions of the text line image;

(d) identify that the text line image comprises a plurality of zones, wherein each zone contains text whose font differs from the text of each adjacent zone;

(e) select a splitting position between multiple zones of the text line image based on a gradient of the OCR confidence measurement;

(f) split the text line image at the splitting position into a plurality of image segments, wherein each image segment contains at least one zone of the text line image; and

(g) perform optical character recognition (OCR) on each image segment to recognize a corresponding text segment of each image segment.

20. A non-transitory computer-readable medium containing instructions which, when executed by one or more processors, cause the one or more processors to:

(a) receive a text line image associated with a line of text contained within a document image, the text line image comprising a plurality of zones, wherein each zone contains text whose font differs from the text of each adjacent zone;

(b) identify that the text line image comprises a plurality of zones, wherein each zone contains text whose font differs from the text of each adjacent zone by having a different vertical position or font size than each adjacent zone;

(c) select at least one splitting position between multiple zones of the text line image;

(d) split the text line image at the splitting position into a plurality of image segments, wherein each image segment contains at least one zone of the text line image; and

(e) perform optical character recognition (OCR) on each image segment to recognize a corresponding text segment of each image segment.