Parallel inverse discrete wavelet transform

• US 6,738,523 B1
• Filed: 09/14/2000
• Issued: 05/18/2004
• Est. Priority Date: 09/14/2000
• Status: Expired due to Fees

• Current Assignee: Eastman Kodak Company
• Original Assignee: Eastman Kodak Company
• Inventors: Small, Jeffrey A., Brown, Mark D.
• Primary Examiner(s): Johnson, Timothy M.
• Assistant Examiner(s): Bayat, Ali
• Application Number:
  US09/662,273
• Time in Patent Office: 1,342 Days
• Field of Search: 382/232, 382/233, 382/242, 382/248, 382/234, 382/238, 382/240, 382/260, 382/263, 382/264, 382/266, 382/305, 382/173, 375/240.11, 375/240.18, 375/240.19, 348/398.1
• US Class Current: 382/240
• CPC Class Codes: G06F 17/148 Wavelet transforms
• CPC Class Codes: G06T 9/007 Transform coding, e.g. discrete cosine transform

Abstract

It is, therefore, an object of the present invention to provide a structure and method of processing signals, which includes transforming a signal into subbands by applying a forward discrete wavelet transform (FDWT) to the signal, partitioning the subbands into overlapping subband subsets, inverse transforming the overlapping subband subsets into signal subsets by applying at least one inverse discrete wavelet transform (IDWT) to the overlapping subband subsets, and combining the signal subsets to reproduce the signal. The IDWT is applied to all of the subband subsets in parallel.

Claims

1. A method of processing signals comprising:

transforming a signal into subbands by applying a forward discrete wavelet transform (FDWT) to said signal;

partitioning said subbands into overlapping subband subsets;

inverse transforming said overlapping subband subsets into signal subsets by applying at least one inverse discrete wavelet transform (IDWT) to said overlapping subband subsets; and

combining said signal subsets to reproduce said signal.

2. The method in claim 1, wherein said IDWT is applied to all of said subband subsets in parallel.

3. The method in claim 1, wherein said inverse transforming includes performing lifting upon said subband subsets to minimize an overlap between adjacent overlapping subband subsets.

4. The method in claim 1, wherein said partitioning is performed so as to minimize an overlap between adjacent overlapping subband subsets.

5. The method in claim 1, further comprising discarding excess sample values that occur as a result of overlap between adjacent overlapping subband subsets.

6. The method in claim 1, wherein said transforming comprises dividing said subbands into a highpass subband group and a lowpass subband group.

7. The method in claim 1, wherein said partitioning comprises a multiple-level FDWT and said IDWT is simultaneously applied to each level of said multiple-level FDWT.

8. The method in claim 1, wherein said signal subsets overlap.

9. The method in claim 1, wherein said lifting minimizes an overlap between adjacent overlapping subband subsets.

10. The method in claim 1, wherein said partitioning is performed so as to minimize an overlap between adjacent overlapping subband subsets.

11. The method in claim 1, wherein said transforming comprises dividing said FDWT coefficients into a highpass subband group and a lowpass subband group.

12. A method of image compression utilizing inverse discrete wavelet transform (IDWT), comprising:

partitioning a set of forward discrete wavelet transform (FDWT) coefficients into two or more overlapping subband subsets;

simultaneously performing an IDWT of each subset using lifting, such that an overlap between adjacent subband subsets is minimized;

recombining said subband subsets; and

discarding excess sample values that occur as a result of overlap between adjacent subband subsets.

13. The method in claim 12, wherein said partitioning comprises a multiple-level FDWT and said IDWT is simultaneously applied to each level of said multiple-level FDWT.

14. The method in claim 12, wherein said partitioning comprises a two-level FDWT and said IDWT is simultaneously applied to each level of said two-level FDWT.

15. The method in claim 12, wherein said partitioning comprises a Mallat two-dimensional FDWT.

16. A computer program product for processing signals, said computer program product comprising a computer readable storage medium having a computer program stored thereon for performing a method comprising:

transforming a signal into subbands by applying a forward discrete wavelet transform (FDWT) to said signal;

partitioning said subbands into overlapping subband subsets;

inverse transforming said overlapping subband subsets into signal subsets by applying at least one inverse discrete wavelet transform (IDWT) to said overlapping subband subsets; and

combining said signal subsets to reproduce said signal.

17. The computer program product in claim 16, wherein said IDWT is applied to all of said subband subsets in parallel.

18. The computer program product in claim 16, wherein said inverse transforming includes performing lifting upon said subband subsets to minimize an overlap between adjacent overlapping subband subsets.

19. The computer program product in claim 16, wherein said partitioning is performed so as to minimize an overlap between adjacent overlapping subband subsets.

20. The computer program product in claim 16, further comprising discarding excess sample values that occur as a result of overlap between adjacent overlapping subband subsets.

21. The computer program product in claim 16, wherein said transforming comprises dividing said subbands into a highpass subband group and a lowpass subband group.

22. The computer program product in claim 16, wherein said partitioning comprises a multiple-level FDWT and said IDWT is simultaneously applied to each level of said multiple-level FDWT.

23. The computer program product in claim 16, wherein said signal subsets overlap.