# US 2020/0394397 A1 · Method to improve performance in document processing

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2020/0394397 A1
- Document ID: US20200394397A1
- Status: Published application
- Kind: A1
- Application number: 16/441,601
- Filed: 2019-06-14
- Published: 2020-12-17
- Earliest claimed: 2019-06-14
- Assignees: Xerox Corporation
- Applicants: Xerox Corporation
- Inventors: Xing Li; DHEVENDRA ALAGAN PALANIVEL; Sainarayanan Gopalakrishnan; RAJASEKAR KANAGASABAI
- Web page: https://patentgrep.com/patent/US20200394397A1

## CPC

- G06T 7/11: Region-based segmentation
- G06T 7/13: Edge detection
- G06T 7/194: involving foreground-background segmentation
- G06V 10/56: relating to colour
- G06V 10/60: relating to illumination properties, e.g. using a reflectance or lighting model
- G06V 30/414: Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text
- G06V 30/416: Extracting the logical structure, e.g. chapters, sections or page numbers; Identifying elements of the document, e.g. authors
- H04N 1/4072: {dependent on the contents of the original}
- G06T 2207/30176: Document
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION

## Abstract

A document processing system produces a digital representation of an original image by determining a background level for white space of the original image. Using the background level, the system identifies pixels not belonging to the white space. The system generates a binary image of the original image based from the identified pixels. The system then identifies image content surrounded by the white space in the original image using the binary image. Next, the system generates a block for each image content and extracts each block from the original image. After processing each block, the system reapplies the processed blocks to a blank canvas to generate the digital representation.

## Claims

### Claim 1

A method of processing image data for a document processing system, the method comprising:
acquiring an input document at the document processing system;
determining a background level for white space of the input document;
using the background level, identifying pixels not belonging to the white space;
generating a binary image of the input document based from the identified pixels;
identifying image content surrounded by the white space in the input document using the binary image;
generating a block for each image content;
extracting the each block from the input document;
performing processing on the each block; and
reapplying the processed blocks to a blank canvas to generate an output document.

### Claim 2 (depends on 1)

The method of claim 1 further comprising:
generating the block for the each image content by creating a bounding box around the each image content;
determining coordinates of the each bounding box;
reapplying the processed blocks to the blank canvas using the coordinates.

### Claim 3 (depends on 1)

The method of claim 1 further comprising:
performing the processing on the each block in parallel while foregoing the processing on pixels belonging to the white space.

### Claim 4 (depends on 1)

The method of claim 1, wherein the block generated for the each image content is surrounded by white space; wherein the processed blocks reapplied to the blank canvas are separated by white space; wherein the white space between the blocks on the canvas avoid stitching artifacts.

### Claim 5 (depends on 1)

The method of claim 1, wherein the generating the binary image of the input document includes:
performing background detection on a given pixel to determine whether the given pixel belongs to the white space or to image content;
performing a neutral pixel detection on the given pixel to generate a tag for the each pixel; and,
feeding the tags into an X-Y cut algorithm to determine elements of the input document.

### Claim 6 (depends on 1)

The method of claim 1 further comprising:
performing background detection on the input document to determine the background level of a pixel;
in response to the background level of the pixel meeting a predetermined threshold, performing a neutral pixel detection on the pixel to determine a chrominance value of the pixel;
in response to the chrominance value meeting a second predetermined threshold, generating a tag for the pixel; and,
feeding the tag into an X-Y cut algorithm to decompose the image into a set of blocks.

### Claim 7 (depends on 6)

The method of claim 6, wherein the tag is a binary tag associating the pixel as belonging to one of white pixel and nonwhite pixel.

### Claim 8 (depends on 5)

The method of claim 5, wherein the performing the background detection includes:
determining an intensity of a given pixel in the input document;
classifying the given pixel as belonging to white space if an intensity of the given pixel falls within a predetermined range, wherein the predetermined range corresponds with the background level; and
classifying the given pixel as belonging to image content if the intensity falls outside the predetermined range.

### Claim 9 (depends on 5)

The method of claim 5, wherein the determining the neutral pixel detection tag for each pixel includes:
determining an amount of luminance variation between a current pixel and a neighborhood of pixels around the current pixel;
determining a new chrominance value for the current pixel using the amount of luminance variation; and
determining the neutral pixel detection tag using the new chrominance value.

### Claim 10 (depends on 1)

The method of claim 1, wherein the performing the processing on the each blocks includes performing an operation selected from the group consisting of:
edge detection;
two-dimensional filtering;
background suppression;
color adjustment;
adjusting lightness and/or contract;
resizing;
color space conversion;
neutral replacement;
error diffusion; and
a combination of the above.

### Claim 11

A document processing system comprising:
a document processing device for producing a digital representation of an original image; and
a processor for:
determining a background level for white space of the original image;
using the background level, identifying pixels not belonging to the white space;
generating a binary image of the original image based from the identified pixels;
identifying image content surrounded by the white space in the original image using the binary image;
generating a block for each image content;
extracting the each block from the original image;
performing processing on the each block; and
reapplying the processed blocks to a blank canvas to generate the digital representation.

### Claim 12 (depends on 11)

The system of claim 11, wherein the processor is further operative to:
generate the block for the each image content by creating a bounding box around the each image content;
determine coordinates of the each bounding box;
reapply the processed blocks to the blank canvas using the coordinates.

### Claim 13 (depends on 11)

The system of claim 11, wherein the processor is further operative to:
perform the processing on the each block in parallel while foregoing the processing on pixels belonging to the white space.

### Claim 14 (depends on 11)

The system of claim 11, wherein the block generated for the each image content is surrounded by white space; wherein the processed blocks reapplied to the blank canvas are separated by white space; wherein the white space between the blocks on the canvas avoid stitching artifacts.

### Claim 15 (depends on 11)

The system of claim 11, wherein the processor is further operative to:
perform background detection on a given pixel to determine whether the given pixel belongs to the white space or to image content;
perform a neutral pixel detection on the each pixel to generate a tag for the each pixel; and,
feed the tags into an X-Y cut algorithm to generate the binary image.

### Claim 16 (depends on 11)

The system of claim 11, wherein the processor is further operative to:
perform background detection on the input document to determine the background level of a pixel;
in response to the background level of the pixel meeting a predetermined threshold, perform a neutral pixel detection on the pixel to determine a chrominance value of the pixel;
in response to the chrominance value meeting a second predetermined threshold, generate a tag for the pixel; and,
feed the tag into an X-Y cut algorithm to decompose the image into a set of blocks.

### Claim 17 (depends on 16)

The system of claim 16, wherein the tag is a binary tag associating the pixel as belonging to one of white pixel and nonwhite pixel.

### Claim 18 (depends on 15)

The system of claim 15, wherein the processor is further operative to:
determine an intensity of a given pixel in the input document;
classify the given pixel as belonging to white space if an intensity of the given pixel falls within a predetermined range, wherein the predetermined range corresponds with the background level; and
classify the given pixel as belonging to image content if the intensity falls outside the predetermined range.

### Claim 19 (depends on 15)

The system of claim 15, wherein the processor is further operative to:
determine an amount of luminance variation between a current pixel and a neighborhood of pixels around the current pixel;
determine a new chrominance value for the current pixel using the amount of luminance variation; and
determine the neutral pixel detection tag using the new chrominance value.

### Claim 20 (depends on 11)

The system of claim 11, wherein the performing the processing on the each blocks includes performing an operation selected from the group consisting of:
edge detection;
two-dimensional filtering;
background suppression;
color adjustment;
adjusting lightness and/or contract;
resizing;
color space conversion;
neutral replacement;
error diffusion; and
a combination of the above.
