Intelligent detection of text on a page

• US 6,289,122 B1
• Filed: 04/15/1999
• Issued: 09/11/2001
• Est. Priority Date: 04/15/1999
• Status: Active Grant

• Current Assignee: Electronics For Imaging Incorporated
• Original Assignee: Electronics For Imaging Incorporated
• Inventors: Karidi, Ron J.
• Primary Examiner(s): Johnson, Timothy M.
• Application Number:
  US09/293,637
• Time in Patent Office: 880 Days
• Field of Search: 382/173, 382/176, 382/180, 382/205, 382/224, 382/228, 382/165, 358/1.9, 707/531
• US Class Current: 382/176
• CPC Class Codes: G06V 30/413 Classification of content, e.g. text, photographs or tables

Abstract

A technique for segmenting an image into text areas and non-text areas in which an image is stored with the following information per pixel: gray scale intensity (4 bits) and an indication of whether the pixel is neutral or color (1 bit). The image, e.g. a scanned RGB image, is converted to 0-15 levels of intensity and has a neutral/color indication bit assigned to each pixel. The technique proceeds in three phases as follows: Tile the image by square blocks, e.g. 6×6 or 8×8 for 600 dpi images, and store information about each block in a buffer; sweep the buffer left to right three tile rows at a time and make a preliminary decision for every tile-block in the middle row; examine the decision made in the previous step in a context block, e.g. a 3×3 block, and make revisions if necessary.

Claims

1. A method for segmenting an image into text areas and non-text areas, comprising the computer aided steps of:

tiling said image by blocks;

wherein each block is typed as to text content, ranked for intensity, and rated as neutral or color;

storing information about each block in a buffer;

sweeping said buffer a predetermined number of tile rows at a time;

making a preliminary decision whether a portion of the image being swept is a text area or a non-text area of said image for every tile-block in a middle row of said sweep;

examining said preliminary decision in a context block;

making revisions to said preliminary determination as necessary based upon said examination step; and

outputting image information which segments said image into text areas and non-text areas.

2. The method of claim 1, wherein said blocks are either of 6×

6 or 8×

8 square blocks for 600 dpi images.

3. The method of claim 1, wherein said sweeping step sweeps said buffer three rows at a time.

4. The method of claim 1, wherein said context block is a 3×

3 block.

5. The method of claim 1, wherein said image is stored with the following information per pixel:

gray scale intensity; and

an indication of whether said pixel is neutral or color.

6. The method of claim 1, wherein said image is a scanned RGB image.

7. The method if claim 1, further comprising the step of:

converting said image to M levels of intensity; and

assigning a neutral/color indication bit to each pixel.

8. An apparatus for segmenting images into text and non-text areas, comprising:

a buffer for storing input images;

means for tiling said image in tile-blocks;

means for determining an intensity value for each pixel in said image relative to a threshold level;

means for classifying each tile-block as to text content type and intensity; and

means for rating each tile-block as neutral or color.

9. The apparatus of claim 8, wherein said types comprise any of undetermined;

not text;

tile-block core;

small text;

outline of text; and

white tile-block.

10. The apparatus of claim 8, further comprising:

means for sweeping groups of tile-blocks.

11. The apparatus of claim 10, wherein sets of three consecutive tile-rows are sweeped from left to right;

and

wherein each three-row-stripe is cut into segments and a decision is made whether a portion of the image being swept is a text area or a non-text area of said image for each segment.

12. The apparatus of claim 10, said sweeping means comprising:

a left pointer that marks a beginning of a segment;

a right pointer that marks said segment'"'"'s end; and

means for advancing both pointers to a point just beyond a segment each time a segment is cut out;

wherein said right pointer is advanced one tile-block at a time so a decision can be made whether to cut out a segment;

wherein said right pointer is advanced if a cut-out is not supported; and

wherein a decision is made as to the type of tile tile-block corresponding to said segment once the segment is cut out.

13. The apparatus of claim 10, wherein said means for sweeping implements a break process that sets one of four attributes:

NON-TEXT, CORE, SPACE, and SIZE; and

wherein said process breaks as NON-TEXT if the number of non-text tile-blocks in a current column is three, the number of neutral tile-blocks is three, the number of non-text blocks in a current segment exceeds four, or the number of color tile-blocks exceeds three;

the process breaks as CORE if the number of tile-blocks with type BLACK-CORE exceeds three less than the segment area and the segment width exceeds the threshold T_core_limit;

the process breaks as SPACE if the number of blocks in a current column with type;

WHITE equals three, the segment width exceeds the threshold T_segment_width and the number of tile-block in a current column with a white₁₃ rows_cols indicator equals three, or the end-of-line is reached; and

the process breaks as SIZE if the segment width exceeds a threshold T_size.

14. The apparatus of claim 12, wherein an attribute for the type of tile-block is attached in a cut segment.

15. The apparatus of claim 12, further comprising:

means for making a preliminary decision as to whether tile-blocks corresponding to a segment are text or non-text.

16. The apparatus of claim 15, wherein if said preliminary decision is that a current tile-block is non-text but said current tile-block is surrounded by text, then said current tile-block is identified as also being text.

17. The apparatus of claim 15, wherein if said preliminary decision is that a current tile-block is text but at least half of the tile-blocks surrounding said current tile-block are non-text, then said current tile-block is identified as also being non-text.

18. The apparatus of claim 8, wherein said means for classifying identifies a type:

WHITE if substantially all the pixels in a tile-block are at or near maximum intensity;

a type;

SMALL-TEXT if a majority of pixels contain ink and a range of lightest to darkest exceeds a predefined range;

a type;

BLACK-CORE if, on average, there is significant dark gray in substantially all the pixels;

a type;

OUTLINE if there is significant contrast and there are large relative differences in the light areas to the dark areas; and

a type;

NOT-TEXT if said tile-block includes pixels containing significant grayscale information.

19. The apparatus of claim 8, wherein said means for determining intensity attributes a one-of-four level code for intensity of each tile-block, wherein intensity is initially set to zero;

the intensity is set;

1 if as few as one pixel has an intensity greater than threshold T1 or if the total amount of ink in pixels with intensity darker than threshold T1 is greater than a predetermined amount;

the intensity is set;

2 if the average pixel intensity is in a middle range; and

the intensity is set;

3 if there is no ink in any pixel that exceeds threshold T1.

20. A method for segmenting an image into text and non-text areas, comprising the steps of:

tiling said image by blocks to classify pixel information therein;

typing each tile-block for text content;

ranking each tile-block for intensity;

rating each tile block as neutral or color;

sweeping said image to determine the identity of text areas and non-text areas therein; and

revising said determination as necessary in accordance with a predetermined scheme.

21. The method of claim 20, wherein said tiling step overlaps a tile-block with a previously scanned tile-block to the left or top by one row.