Coder matched layer separation and interpolation for compression of compound documents

• US 6,941,024 B2
• Filed: 01/14/2002
• Issued: 09/06/2005
• Est. Priority Date: 01/14/2002
• Status: Expired due to Fees

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Mukherjee, Debargha
• Primary Examiner(s): Mehta, Bhavesh M.
• Assistant Examiner(s): Hung, Yubin
• Application Number:
  US10/047,289
• Publication Number:
  US 20030133617A1
• Time in Patent Office: 1,331 Days
• Field of Search: 382/164, 382/166, 382/173, 382/224, 382/239, 382/240, 382248-251, 382/276, 382/282, 382/283, 375/240.19, 375/240.18
• US Class Current: 382/248
• CPC Class Codes: H04N 1/41 Bandwidth or redundancy reduction by scanning H04N1/17 ; methods or arrangements for coding, decoding, compressing or decompressing digital video signals H04N19/00

Abstract

A method of preparing an image for efficient wavelet transform compression includes the steps of separating the image into foreground and background image layers, and a mask layer. A smoothing filter is applied to interpolate irrelevant pixel values in the foreground and background layers for coder efficiency. The irrelevant pixel values are interpolated as a function of relevant pixels and causal irrelevant pixels within the smoothing filter window.

Claims

1. A method of decomposing an image comprising the steps of:

a) decomposing the image into a plurality of stripes;

b) decomposing each stripe into foreground and background image layers, and a mask layer; and

c) applying a smoothing filter to interpolate irrelevant pixel values in the foreground and background layers for wavelet encoding efficiency,

wherein step c) further comprises the steps of;

i) determining a layer base color and offsets to a common reduced area of each layer to identify image and mask layer values for all regions except an overlapped common reduced area; and

ii) separating the overlapped common reduced area into foreground and background layers.

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

d) encoding the foreground, background, and mask layers with a forward discrete wavelet transformation encoder.

3. The method of claim 2 wherein the foreground and background are JPEG 2000 encoded, wherein the mask is encoded with one of a JBIG and a JBIG2 encoder.

4. The method of claim 1 wherein step c) further comprises the steps:

iii) classifying each pixel within a selected layer as relevant or irrelevant; and

iv) applying a smoothing filter to each irrelevant pixel, p_c, proceeding in a raster scan order to interpolate a value for that irrelevant pixel.

5. The method of claim 4 wherein a normalized weighted average of the relevant pixels and causal irrelevant pixels contribute to the interpolated value.

6. The method of claim 4 wherein the smoothing filter is a weighted Gaussian filter.

7. The method of claim 6 wherein each element of the smoothing filter is of the form w_klV_kb wherein V_kl is a non-weighted filter value, wherein W_kl is a function of its associated pixel causality and relevance.

8. The method of claim 7 wherein w_kl=0 for the center pixel (p_c) and any non-causal irrelevant pixel.

9. The method of claim 7, wherein w_kl is set to m₁ if its associated pixel is a relevant pixel and w_kl is set to m₂ if the associated pixel is a causal irrelevant pixel such that $\frac{m_1}{m_2}>$

1.

10. The method of claim 7 wherein w_kl is set to m₁ if its associated pixel is a relevant pixel and w_kl is set to m₂ if the associated pixel is a causal irrelevant pixel such that $\frac{m_1}{m_2}=2.$

11. A method of decomposing an image comprising the steps of:

a) decomposing the image into a plurality of stripes;

b) decomposing each stripe into foreground and background image layers, and a mask layer; and

c) applying a smoothing filter to interpolate irrelevant pixel values in the foreground and background layers for wavelet encoding efficiency

wherein step b) further comprises the steps of;

i) dividing a selected layer into a plurality of decision regions(D_ij) and associated analysis regions (A_ij), wherein each D_ij⊂

A_ij; and

ii) assigning the entire region D_ij to one of the background and foreground layers, if a contrast of A_ij does not exceed a pre-determined threshold.

12. The method of claim 11 wherein the entire region D_ij is assigned to the foreground or background layers based on whether the average pixel value AVG(Dij) is closer to an average pixel value of neighboring foreground regions or neighboring background regions.

13. The method of claim 11 wherein step b) further comprises the steps of:

iii) distributing the pixels of D_ij between the background and foreground layers, if a contrast of A_ij exceeds a pre-determined threshold.

14. The method of claim 13, wherein step b)(ii) further comprises the steps of:

i) separating the pixels of A_ij into two groups, GROUP_1 and GROUP_2;

ii) compute an average (AVG_1, AVG_2) for each group; and

iii) mutually exclusively assigning the pixels of D_ij GROUP_1 and GROUP_2 to a selected one of the foreground and background layers based on a comparison of the relative luminance of GROUP_1 and GROUP_2.

15. A method of preparing an image for efficient wavelet transform compression, comprising the steps of:

a) separating the image into foreground and background image layers, and a mask layer; and

b) applying a smoothing filter to interpolate irrelevant pixel values in the foreground and background layers for coder efficiency,

wherein step a) further comprises

i) dividing a selected layer into a plurality of decision regions(D_ij) and associated analysis regions (A_ij), wherein each D_ij ⊂

A_ij.

ii) assigning the entire region D_ij to one of the background and foreground layers, if a contrast of A_ij does not exceed a pre-determined threshold, and

iii) distributing the pixels of D_ij between the background and foreground layers, if a contrast of A_ij exceeds a pre-determined threshold.

16. The method of claim 15 wherein a normalized weighted average of the relevant pixels and the causal irrelevant pixels contribute to the interpolated value.

17. The method of claim 15 wherein the smoothing filter is a weighted Gaussian filter.

18. The method of claim 15 wherein each element of the smoothing filter is of the form w_kl V_kb wherein V_klis a non-weighted filter value, wherein w_kl is a function of its associated pixel causality and relevance.

19. The method of claim 18 wherein w_kl =0 for the center pixel (p_c) and any non-causal irrelevant pixel.

20. The method of claim 18 wherein w_kl is set to m₁ if its associated pixel is a relevant pixel and w_klis set to m₂ if the associated pixel is a causal irrelevant pixel such that $\frac{m_1}{m_2}>$

1.

21. The method of claim 18 wherein w_kl is set to m₁ if its associated pixel is a relevant pixel and w_kl is set to m₂ if the associated pixel is a causal irrelevant pixel such that $\frac{m_1}{m_2}=2.$