COLOUR CORRECTING FOREGROUND COLOURS FOR VISUAL QUALITY IMPROVEMENT

• US 20100246951A1
• Filed: 03/18/2010
• Published: 09/30/2010
• Est. Priority Date: 03/31/2009
• Status: Active Grant

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: McDonell, Trevor Lee, Chen, Yu-Ling
• Granted Patent: US 8,644,602 B2
• Time in Patent Office: Days
• US Class Current: 382/167
• CPC Class Codes: H04N 1/58 Edge or detail enhancement; Noise or error suppression, e.g. colour misregistration correction H04N1/62 takes precedence
• CPC Class Codes: H04N 1/62 Retouching, i.e. modification of isolated colours only or in isolated picture areas only

Abstract

A method (199) is disclosed for modifying colours of text extracted from an image of a document page to improve visual quality. The document image comprises text and image regions of various colours. The method divides (100) the document image into a plurality of connected components, the connected components comprising pixels of similar colour that are closely spatially located, and selects from the image a plurality of the connected components that represent text. The plurality of text connected components are grouped to form a logical structure grouping of said page comprising one or more text blocks wherein each said text block represents a text line or paragraph. One or more logically consistent colour sets are then formed (120, 699, 399, 499) for each text block based on the colours of the connected components in the text block. The logically consistent colour sets represent all text of visually similar colour of the corresponding text line or paragraph. The method then applies colour enhancement (130) to the logically consistent colour sets to enhance visual quality and correct image defects in the document image.

Claims

1. A method of modifying colours of text extracted from an image of a document page to improve visual quality, said document image comprising text and image regions of various colours, the method comprising:

(a) dividing the document image into a plurality of connected components, the connected components comprising pixels of similar colour that are closely spatially located, and selecting from said image a plurality of the connected components that represent text;

(b) grouping said plurality of text connected components to form a logical structure grouping of said page comprising one or more text blocks wherein each said text block represents a text line or paragraph;

(c) forming one or more logically consistent colour sets for each said text block based on the colours of said connected components in said text block, wherein said logically consistent colour sets represent all text of visually similar colour of the corresponding text line or paragraph; and

(d) applying colour enhancement to said logically consistent colour sets to enhance visual quality and correct image defects in the document image.

2. A method according to claim 1, wherein the said logically consistent colour sets of a said text block are formed by the steps of:

(ca) designating each connected component of said text block to a colour set based on colour similarity of said connected component and of a corresponding enclosing background colour, wherein each colour set has a foreground and background colour; and

(cb) merging pairs of said colour sets of said text block based on their foreground and background colour similarity.

3. A method according to claim 2, wherein step (cb) comprises determining whether to merge a pair of said colour sets of the said text block based on a threshold for a foreground colour distance of said pair, wherein said threshold is determined by a background colour difference of the said pair.

4. A method according to claim 2, wherein step (cb) comprises the steps of:

(cba) compensating each of the foreground colours of said colour sets based on the corresponding background colour; and

(cbb) determining whether to merge based on a predetermined colour threshold for the difference of the foreground colours of the pair.

5. A method according to claim 2, wherein the background colour of a said colour set further includes information of any closely spatially located graphic regions, a pixel count, and a minimum bounding rectangle.

6. A method according to claim 2, wherein the background colour of a said merged colour set pair, where the colour sets have different background colour is determined, in order, by the steps of:

(a) selecting the non-white background colour of the said pair if one of the said pair is non-white;

(b) selecting from said pair the background colour covering the largest area if both of the said colour sets have non-white background colour; and

(c) updating the associated background colour information with;

(i) the sum of background pixel counts of the said pair;

(ii) the sum of the statistics of any closely spatially located graphic regions of the said pair; and

(iii) the minimum bounding rectangle enclosing both of the minimum bounding rectangles of the said pair.

7. A method according to claim 2 further comprising creating a new colour set if no colour set with a similar foreground and background colour to the said connected component can be found.

8. A method according to claim 1, wherein the amount of colour correction depends on the edge ratio of the connected components within each of the said logically consistent colour sets.

9. A method according to claim 1, wherein visually black colours are made blacker, visually white colours are made whiter, and visually non-greyish colours are made more saturated.

10. A method according to claim 9, wherein visually black colour is determined by the foreground and background of a said colour set comprising one of these criteria:

(a) the foreground is very close to black;

or

(b) the foreground has a much lower HSV value component than the background and the hue channel components are similar.

11. A method according to claim 9, wherein visually white colour is determined by the foreground and background of a said colour set comprising one of these criteria:

(a) the foreground is very close to white;

or

(b) the foreground has a much higher HSV value component than the background and the hue channel components are similar.

12. A method according to claim 9, wherein visually non-greyish colour is determined by HSV saturation and YCbCr luminance of the foreground of a said colour set if HSV saturation is below a predetermined threshold and the sum of HSV saturation and YCbCr luminance is below a second predetermined threshold.

13. A method according to claim 1, wherein the colour correction is optionally disabled if a said logically consistent colour set is closely spatially located to a significant graphic region.

14. A method according to claim 1, further comprising merging colour corrected sets across said text blocks with a similar foreground colour.

15. A computer readable storage medium having a computer program recorded thereon, the program being executable by computer apparatus to modifying colours of text extracted from an image of a document page to improve visual quality, said document image comprising text and image regions of various colours, the program comprising:

code for dividing the document image into a plurality of connected components, the connected components comprising pixels of similar colour that are closely spatially located, and selecting from said image a plurality of the connected components that represent text;

code for grouping said plurality of text connected components to form a logical structure grouping of said page comprising one or more text blocks wherein each said text block represents a text line or paragraph;

code for forming one or more logically consistent colour sets for each said text block based on the colours of said connected components in said text block, wherein said logically consistent colour sets represent all text of visually similar colour of the corresponding text line or paragraph; and

code for applying colour enhancement to said logically consistent colour sets to enhance visual quality and correct image defects in the document image.

16. Computer apparatus comprising a processor, a memory, an input for a document image and an output for a modified document images, said memory having a computer program recorded therein, the program being executable by the processor to modify colours of text extracted from the document image, said document image comprising text and image regions of various colours, the program comprising:

code for dividing the document image into a plurality of connected components, the connected components comprising pixels of similar colour that are closely spatially located, and selecting from said image a plurality of the connected components that represent text;

code for grouping said plurality of text connected components to form a logical structure grouping of said page comprising one or more text blocks wherein each said text block represents a text line or paragraph;

code for forming one or more logically consistent colour sets for each said text block based on the colours of said connected components in said text block, wherein said logically consistent colour sets represent all text of visually similar colour of the corresponding text line or paragraph; and

code for applying colour enhancement to said logically consistent colour sets to enhance visual quality and correct image defects in the document image to form the modified document image.

17. A method of modifying colours of text extracted from a document image of a page, the method comprising;

(a) generating a first text connected component and a second text connected component from the document image, said first and second text connected components being a different colour and each having a different enclosing background colour;

(b) determining an output colour based on the colours of the first and second text connected components, the determined output colour being based on a comparison between a colour of at least one text connected component and a threshold, wherein the threshold or colour of the at least one text connected component is adjusted according to the colour of the enclosing background;

(c) setting the colour for the first and second text connected components according to the determined output colour.

18. A method according to claim 17, wherein step (b) comprises determining the output colour by merging the colours of the first and second text connected components based on a threshold for a colour distance between said text connected component colours, wherein said threshold is determined by a colour difference between the enclosing background colours.

19. A method according to claim 17, wherein step (b) comprises:

(ba) compensating at least one of the first and second connected component colours based on the enclosing background colour; and

(bb) determining the output colour by merging the text connected component colours based on a predetermined threshold for a colour distance between said compensated text connected component colours.