Systems and methods for generating background and foreground images for document compression

• US 7,783,117 B2
• Filed: 03/26/2007
• Issued: 08/24/2010
• Est. Priority Date: 08/12/2005
• Status: Expired due to Fees

• Current Assignee: Seiko Epson Corporation (Seiko Group)
• Original Assignee: Seiko Epson Corporation (Seiko Group)
• Inventors: Bhattacharjya, Anoop K, Liu, Che-Bin
• Primary Examiner(s): Tran; Phuoc
• Application Number:
  US11/691,339
• Publication Number:
  US 20070189615A1
• Time in Patent Office: 1,247 Days
• Field of Search: 382/162, 382/164, 382/165, 382/173, 382/199, 382/232, 382/243, 382/264
• US Class Current: 382/232
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/162 Quantising the image signal
• CPC Class Codes: H04N 1/409 Edge or detail enhancement; Noise or error suppression

Abstract

Disclosed are embodiments of systems and methods to generate background and foreground images for a document, which enables high-quality and high-ratio document compression. In embodiments, high-accuracy layer processing enables text enhancement, paper color removal, and many other advanced image analysis and processing. Embodiments of the systems support several operation modes and its many parameters, such as layer compression ratios, image segmentation, and modulized image processing, may be adjusted to generate optimal compressed files for different purposes.

Claims

1. A method for compressing an image comprising pixels, the method comprising:

identifying edge pixels using gradient pixels identified in a plurality of pixels selected from the image, in a first low-pass filtered set of the plurality of pixels, and in a second low-pass filtered set of the plurality of pixels;

determining an edge pixel type for each of the identified edge pixels;

using at least some of the identified edge pixels to label each pixel of the plurality of pixels;

using the labeled pixels to identify non-background region pixels;

generating a foreground image layer comprising the non-background region pixels;

generating a background image layer comprising pixels with a background label;

generating a mask layer that identifies from which of the foreground and background image layers a pixel should be selected when generating a representation of the plurality of pixels; and

compressing the mask layer differently than at least one of the foreground and background image layers.

2. The method of claim 1 further comprising the step of:

responsive to estimating a background color, relabeling the label associated with at least one pixel.

3. The method of claim 1 wherein the step of using the labeled pixels to identify non-background region pixels comprises the steps of:

using the pixel labels to generate a background mask comprising pixels from the plurality of pixels with a background label; and

performing a connected component analysis on non-background pixels of the background mask and responsive to a connected component group of non-background pixels being below a threshold size, selecting pixels corresponding to the connected component group as non-background region pixels.

4. The method of claim 3 wherein the threshold size of the connect component group refers to the number of pixels of the connect component group or to a size of a region that bounds the connect component group.

5. The method of claim 1 wherein the step of using the labeled pixels to identify non-background region pixels comprises the steps of:

generating an edge mask comprising pixels labeled as edge pixels;

using at least some of the labeled pixels and the edge mask to identify solid region pixels; and

including the solid region pixels as non-background pixels in the foreground image layer.

6. The method of claim 5 wherein the step of using at least some of the labeled pixels and the edge mask to identify solid region pixels comprises the steps of:

using line scans to identify a non-edge-pixel segment bounded by a first edge pixel segment and by a second edge pixel segment;

computing a first color value for the first edge pixel segment, a second color value for the second edge pixel segment, and a non-edge-pixel segment color value for the non-edge-pixel segment; and

responsive to the difference between the non-edge-pixel segment color value and the average of the first and second color values being below a threshold value, and responsive to a threshold size limit being satisfied, identifying the combined first edge pixel segment, second edge pixel segment, and non-edge-pixel segment as solid region pixels.

7. The method of claim 6 further comprising the steps of:

excluding from the solid region pixels any pixels with a background label.

8. A tangible computer-readable medium comprising one or more sequences of instructions to direct an instruction-executing device to perform at least the steps of claim 1.

9. A method for compressing an image comprising pixels, the method comprising:

labeling each pixel from a plurality of pixels selected from the image;

using the pixel labels to generate a background mask comprising pixels from the plurality of pixels with a background label;

performing a connected component analysis on non-background pixels of the background mask and responsive to a connected component group of non-background pixels being below a threshold size, selecting pixels corresponding to the connected component group as non-background region pixels;

generating a foreground image layer comprising the non-background region pixels;

generating a background image layer comprising pixels with a background label;

generating a mask layer that identifies from which of the foreground and background image layers a pixel should be selected when generating a representation of the plurality of pixels; and

compressing the mask layer differently than at least one of the foreground and background image layers.

10. The method of claim 9 wherein the step of labeling each pixel from the plurality of pixels selected from the image comprises the steps of:

identifying edge pixels using gradient pixels identified in the plurality of pixels, in a first low-pass filtered set of the plurality of pixels, and in a second low-pass filtered set of the plurality of pixels;

determining an edge pixel type for each of the identified edge pixels; and

using at least some of the identified edge pixels to label each pixel of the plurality of pixels.

11. The method of claim 9 further comprising the steps of:

generating an edge mask comprising pixels labeled as edge pixels;

using at least some of the labeled pixels and the edge mask to identify solid region pixels; and

including the solid region pixels as non-background pixels in the foreground image layer.

12. The method of claim 11 wherein the step of using at least some of the labeled pixels and the dilated edge mask to identify solid region pixels comprises the steps of:

using line scans to identify a non-edge-pixel segment bounded by a first edge pixel segment and a second edge pixel segment;

computing a first color value for the first edge pixel segment, a second color value for the second edge pixel segment, and a non-edge-pixel segment color value for the non-edge-pixel segment; and

responsive to the difference between the non-edge-pixel segment color value and the average of the first and second color values being below a threshold value, and responsive to a threshold size limit being satisfied, identifying the combined first edge pixel segment, second edge pixel segment, and non-edge-pixel segment as solid region pixels.

13. The method of claim 9 wherein the threshold size of the connect component group refers to the number of pixels of the connect component group or to a size of a region that bounds the connect component group.

14. A tangible computer-readable medium comprising one or more sequences of instructions to direct an instruction-executing device to perform at least the steps of claim 9.

15. A system for compressing an image comprising:

a pixel labeler, coupled to receive an image comprising pixels, that labels each pixel of a plurality of pixel selected from the image;

a background estimator, coupled to receive information from the pixel labeler, that attempts to estimate a background color of the plurality of pixels and responsive to estimating the background color, relabels the label associated with at least one pixel of the plurality of pixels;

a background mask and edge mask generator, coupled to receive information from the background estimator, that generates a background mask comprising pixels from the plurality of pixels with a background label and that generates an edge mask comprising pixels with an edge pixel label;

a non-background region extractor, coupled to receive information from the background mask generator, that performs a connected component analysis on the non-background pixels of the background mask and responsive to a connected component group of non-background pixels being below a threshold size, identifying pixels that correspond to the connected component group as non-background pixels;

an image layer separator, coupled to receive information from the non-background region extractor, that generates a foreground image layer comprising non-background pixels, a background image layer comprising pixels with a background label, and generates a mask layer that identifies from which of the foreground and background image layers a pixel should be selected when generating a representation of the plurality of pixels; and

a layer-based compression module, coupled to receive information from the image layer separator, that compresses the mask layer differently than at least one of the foreground and background image layers.

16. The system of claim 15, wherein the pixel labeler labels the plurality of pixels selected from the image by:

identifying edge pixels using gradient pixels identified in the plurality of pixels, in a first low-pass filtered set of the plurality of pixels, and in a second low-pass filtered set of the plurality of pixels;

determining an edge pixel type for each of the identified edge pixels; and

using at least some of the identified edge pixels to label each pixel of the plurality of pixels.

17. The system of claim 15 further comprising:

a mode selector, that receives for a selected mode and provides information to the image layer separator to generate the foreground layer according to the selected mode; and

a solid region extractor, coupled to receive information from the mode selector, that, responsive to a specific selected mode, identifies solid region pixels in the edge mask using line scans so that the identified solid region pixels may be included as non-background pixels in the foreground image layer.

18. The system of claim 17 wherein the solid region extractor identifies solid region pixels by performing the steps comprising:

using line scans in the edge mask to identify a non-edge-pixel segment bounded by a first edge pixel segment and a second edge pixel segment;

computing a first color value for the first edge pixel segment, a second color value for the second edge pixel segment, and a non-edge-pixel segment color value for the non-edge-pixel segment; and

responsive to the difference between the non-edge-pixel segment color value and the average of the first and second color values being below a threshold value, and responsive to a threshold size limit being satisfied, identifying the combined first edge pixel segment, second edge pixel segment, and non-edge-pixel segment as solid region pixels.

19. The system of claim 18 wherein the solid region extractor excludes from the solid region pixels any pixels with a background label.

20. The system of claim 15 further comprising:

a layer-based color enhancer, coupled to receive information from the image layer separator, that smoothes and color enhances at least some of the non-background pixels in the foreground image layer; and

a background color suppressor, coupled to receive information about the background image layer and coupled to provide information to the layer-based compression module, that, responsive to estimating a light colored background, adjusts at least one background pixel'"'"'s color of the background image layer toward white.