APPARATUS AND METHOD FOR VIDEO DATA PROCESSING

• US 20130003870A1
• Filed: 07/01/2011
• Published: 01/03/2013
• Est. Priority Date: 07/01/2011
• Status: Abandoned Application

• Current Assignee: Advanced Micro Devices, Inc., ATI Technologies ULC (Advanced Micro Devices, Inc.)
• Original Assignee: Advanced Micro Devices, Inc., ATI Technologies ULC (Advanced Micro Devices, Inc.)
• Inventors: Swan, Philip L., Spencer, Christopher L.
• Application Number:
  US13/174,873
• Publication Number:
  US 20130003870A1
• Time in Patent Office: Days
• US Class Current: 375/240.26
• CPC Class Codes: H04N 19/43 Hardware specially adapted for motion estimation or compensation
• CPC Class Codes: H04N 19/436 using parallelised computational arrangements

Abstract

Methods and apparatus for accelerating the processing of image data are disclosed that are particularly useful in conducting graphical pattern searches. Embodiments of the invention conduct and implement comparative calculations of reference and search image pel data on a multi-pel comparative basis, particularly, sum of the absolute differences (SAD) based calculation comparisons.

Claims

1. An apparatus for facilitating the processing of video image data:

a first input configured to receive pel data for a set of a predetermined number N of pels of a reference image, where N is at least 2;

a second input configured to receive pel data for a set of a predetermined number M of pels of a search image where M is greater than N;

a plurality of K arithmetic circuits, each configured to process the pel data for the set of N pels received via the first input with pel data for a search image pel subset of N pels of the set of M pels received via the second input such that the search image pel subset used by each arithmetic circuit contains pel data with respect to different subsets of the set of M pels; and

an output configured to output a set of K data elements where each element corresponds to a comparison result generated with respect to one of the search image pel subsets.

2. The apparatus of claim 1 further comprising:

a third input configured to receive a set of K elements previously produced from the output with respect to the subsets of the set of M pels for which pel data is being received by the second input;

wherein each arithmetic circuit is configured to accumulate into the comparison result that it generates with respect to one of the search image pel subsets a corresponding subset data element of the set of K elements received via the third input.

3. The apparatus of claim 2 wherein N=4, M=8 and K=4 and pel data for each pel is a byte of eight bits.

4. The apparatus of claim 1 wherein:

the first input is configured to receive pel data for N consecutive pels of the reference image;

the second input is configured to receive pel data for M consecutive pels of the search image; and

each arithmetic circuit is configured to process the pel data for N consecutive pels of the reference image with pel data for a respective search image pel subset of N consecutive pels of the search image.

5. The apparatus of claim 1 wherein each arithmetic circuit is configured to process the pel data for the set of N pels of the reference image with pel data for the respective search image pel subset by calculating the absolute value of the difference of a pel value of each pel of the N reference image pels with a pel value of a corresponding pel of the respective search image pel subset.

6. The apparatus of claim 1 where the reference image is defined by object pels and background pels such that the pel data for an object pel represents a value within a predetermined range and the pel data for a background pel is set to a fixed value outside the predetermined range, the IC wherein:

each arithmetic circuit includes;

N processing circuits, each configured to process the pel data for a pel of the set of N pels of the reference image with pel data of a corresponding pel of the respective search image subset such that the a comparative value is produced where the pel of the reference image is an object pel and a zero value is produced where the pel of the reference image is a background pel; and

a combiner component configured to selectively combine the values produced by the N processing circuits and to produce a collective comparative value with respect to the respective search image pel subset as one of the K data elements.

7. The apparatus of claim 6 further comprising:

a third input configured to receive a set of K elements previously produced from the output with respect to the subsets of the set of M pels for which pel data is being received by the second input;

wherein each arithmetic circuit is configured to accumulate into the comparison result that it generates with respect to one of the search image pel subsets a corresponding subset data element of the set of K elements received via the third input.

8. The apparatus of claim 6 wherein:

the first input is configured to receive pel data for N consecutive pels of the reference image;

the second input is configured to receive pel data for M consecutive pels of the search image; and

each arithmetic circuit is configured to process the pel data for N consecutive pels of the reference image with pel data for a respective search image pel subset of N consecutive pels of the search image.

9. The apparatus of claim 6 wherein each arithmetic circuit is configured to process the pel data for the set of N pels of the reference image with pel data for the respective search image pel subset by calculating the absolute value of the difference of a pel value of each pel of the N reference image pels with a pel value of a corresponding pel of the respective search image pel subset.

10. The apparatus of claim 9 wherein N=4, M=8 and K=4 and pel data for each pel is a byte of eight bits.

11. The apparatus of claim 1 configured as a video capture device.

12. The apparatus of claim 1 configured as a video display device.

13. A method of processing of video image data:

inputting pel data for a set of a predetermined number N of pels of a reference image, where N is at least 2;

inputting pel data for a set of a predetermined number M of pels of a search image where M is greater than N;

conducting K arithmetic processes with respect to the inputted pels, each arithmetic process processing the pel data for the set of N pels of the reference image with pel data for a search image pel subset of N pels of the set of M pels of the search image such that the search image pel subset used by each arithmetic process contains pel data with respect to different subsets of the set of M pels; and

outputting a set of K data elements where each element corresponds to a comparison result generated with respect to one of the search image pel subsets.

14. The method of claim 13 further comprising:

inputting a set of K elements previously produced from the output with respect to the subsets of the set of M pels for which pel data is input;

wherein each arithmetic process accumulates into the comparison result that is respectively generated with respect to one of the search image pel subsets a corresponding subset data element of the inputted set of K elements.

15. The method of claim 14 performed where N=4, M=8 and K=4 and pel data for each pel is a byte of eight bits.

16. The method of claim 13 wherein:

pel data for N consecutive pels of the reference image is input;

pel data for M consecutive pels of the search image is input; and

each arithmetic process processes the pel data for N consecutive pels of the reference image with pel data for a respective search image pel subset of N consecutive pels of the search image.

17. The method of claim 13 wherein each arithmetic process processes the pel data for the set of N pels of the reference image with pel data for the respective search image pel subset by calculating the absolute value of the difference of a pel value of each pel of the N reference image pels with a pel value of a corresponding pel of the respective search image pel subset.

18. The method of claim 13 where the reference image is defined by object pels and background pels such that the pel data for an object pel represents a value within a predetermined range and the pel data for a background pel is set to a fixed value outside the predetermined range, the method wherein:

each arithmetic process includes;

N comparison sub processes, each processing the pel data for a pel of the set of N pels of the reference image with pel data of a corresponding pel of the respective search image subset such that the a comparative value is produced where the pel of the reference image is an object pel and a zero value is produced where the pel of the reference image is a background pel; and

a combiner process that selectively combines the values produced by the N comparison sub processes to produce a collective comparative value with respect to the respective search image pel subset as one of the K data elements.

19. The method of claim 18 further comprising:

inputting a set of K elements previously produced from the output with respect to the subsets of the set of M pels for which pel data is input;

wherein each arithmetic process accumulates into the comparison result that it generates with respect to one of the search image pel subsets a corresponding subset data element of the inputted set of K elements.

20. The method of claim 18 wherein:

pel data for N consecutive pels of the reference image is input;

pel data for M consecutive pels of the search image is input; and

each arithmetic process processes the pel data for N consecutive pels of the reference image with pel data for a respective search image pel subset of N consecutive pels of the search image.

21. The method of claim 18 wherein each arithmetic process processes the pel data for the set of N pels of the reference image with pel data for the respective search image pel subset by calculating the absolute value of the difference of a pel value of each pel of the N reference image pels with a pel value of a corresponding pel of the respective search image pel subset.

22. The method of claim 21 performed where N=4, M=8 and K=4 and pel data for each pel is a byte of eight bits.

23. The method of claim 24 performed by an integrated circuit in response to a microinstruction that identifies a Masked Quad SAD operation to be performed.

24. The method of claim 15 performed by an integrated circuit in response to a microinstruction that identifies a Quad SAD operation to be performed.

25. A computer-readable storage medium storing a set of instructions for execution by a processor to facilitate processing of video image data that is adapted to:

receive pel data for a set of a predetermined number N of pels of a reference image, where N is at least 2;

receive pel data for a set of a predetermined number M of pels of a search image where M is greater than N;

conduct K arithmetic processes with respect to the received pels, each arithmetic process processing the pel data for the set of N pels of the reference image with pel data for a search image pel subset of N pels of the set of M pels of the search image such that the search image pel subset used by each arithmetic process contains pel data with respect to different subsets of the set of M pels; and

output a set of K data elements where each element corresponds to a comparison result generated with respect to one of the search image pel subsets.

26. The computer-readable storage medium of claim 25, wherein the instructions are identified by a single OpCode that is adapted to:

receive 32 bits of pel data representing four consecutive pels of the reference image;

receive 64 bits of pel data representing eight consecutive pels of the search image; and

conduct four arithmetic processes that each process the pel data for the four consecutive pels of the reference image with pel data for the respective search image pel subset by calculating the absolute value of the difference of a pel value of each pel of the four reference image pels with a pel value of a corresponding pel of the respective search image pel subset; and

output four 16 bit data elements.

27. The computer-readable storage medium of claim 25, wherein the set of instructions is for execution by the processor to facilitate processing of video image data where the reference image is defined by object pels and background pels such that the pel data for an object pel represents a value within a predetermined range and the pel data for a background pel is set to a fixed value outside the predetermined range, wherein:

each conducted arithmetic process includes;

N comparison sub processes that each process the pel data for a pel of the set of N pels of the reference image with pel data of a corresponding pel of the respective search image subset such that the a comparative value is produced where the pel of the reference image is an object pel and a zero value is produced where the pel of the reference image is a background pel; and

a combiner process that selectively combines the values produced by the N comparison sub processes to produce a collective comparative value with respect to the respective search image pel subset as one of the K data elements.

28. The computer-readable storage medium of claim 27, wherein the instructions are identified by a single OpCode that is adapted to:

receive 32 bits of pel data representing four consecutive pels of the reference image;

receive 64 bits of pel data representing eight consecutive pels of the search image; and

conduct four arithmetic processes that each process the pel data for the four consecutive pels of the reference image with pel data for the respective search image pel subset by calculating the absolute value of the difference of a pel value of each pel of the four reference image pels with a pel value of a corresponding pel of the respective search image pel subset; and

output four 16 bit data elements.

29. A computer-readable storage medium storing a set of instructions for execution by one or more processors to facilitate manufacture of circuitry within an integrated circuit that is adapted to:

receive pel data for a set of a predetermined number N of pels of a reference image, where N is at least 2;

receive pel data for a set of a predetermined number M of pels of a search image where M is greater than N;

conduct K arithmetic processes with respect to the received pels, each arithmetic process processing the pel data for the set of N pels of the reference image with pel data for a search image pel subset of N pels of the set of M pels of the search image such that the search image pel subset used by each arithmetic process contains pel data with respect to different subsets of the set of M pels; and

output a set of K data elements where each element corresponds to a comparison result generated with respect to one of the search image pel subsets.

30. The computer-readable storage medium of claim 29, wherein the instructions are hardware description language (HDL) instructions used for the manufacture of a device.

31. The computer-readable storage medium of claim 29, wherein the set of instructions is to facilitate manufacture of circuitry within an integrated circuit to facilitate processing of video image data where the reference image is defined by object pels and background pels such that the pel data for an object pel represents a value within a predetermined range and the pel data for a background pel is set to a fixed value outside the predetermined range, wherein:

each conducted arithmetic process includes;

N comparison sub processes that each process the pel data for a pel of the set of N pels of the reference image with pel data of a corresponding pel of the respective search image subset such that the a comparative value is produced where the pel of the reference image is an object pel and a zero value is produced where the pel of the reference image is a background pel; and

a combiner process that selectively combines the values produced by the N comparison sub processes to produce a collective comparative value with respect to the respective search image pel subset as one of the K data elements.

32. The computer-readable storage medium of claim 31, wherein the instructions are hardware description language (HDL) instructions used for the manufacture of a device.

33. An apparatus for facilitating the processing of video image data:

a first input configured to receive pel data for a plurality of pels of a reference image defined by object pels and background pels such that the pel data for an object

pel represents a value within a predetermined range and the pel data for a background pel is set to a fixed value outside the predetermined range;

a second input configured to receive pel data for a corresponding plurality of pels of a search image;

a corresponding plurality of processing circuits, each configured to process the pel data for a pel of the reference image received via the first input with pel data of a corresponding pel of the search image received via the second input such that a comparative value is produced where the pel of the reference image is an object pel and a zero value is produced where the pel of the reference image is a background pel; and

a combiner component configured to selectively combine the values produced by the plurality of processing circuits and to output a collective comparative value with respect to the plurality of search image pels.

34. The apparatus of claim 33 further comprising:

a third input configured to receive a collective comparative value previously produced by the combiner with respect to the plurality of pels for which pel data is received by the second input;

wherein the combiner is configured to accumulate the previously produced collective comparative value into the collective comparative value that it produces with respect to the plurality of pels for which pel data is received by the second input.

35. The apparatus of claim 34 where pel data for each pel is a byte of eight bits wherein:

the first input is configured to receive pel data for a set of four pels of the reference image such that the pel data for a background pel is set to a zero value;

the second input is configured to receive pel data for a set of four pels of the search image; and

the plurality of processing circuits is four in number.

36. The apparatus of claim 33 wherein the processing circuits are configured to process the pel data for the pels of the reference image with pel data for pels of the search image by calculating the absolute value of the difference of a pel value of each pel of the reference image set of pels with a pel value of a corresponding pel of the search image set of pels.

37. The apparatus of claim 36 where pel data for each pel is a byte of eight bits wherein:

the first input is configured to receive pel data for a set of four pels of the reference image such that the pel data for a background pel is set to a zero value;

the second input is configured to receive pel data for a set of four pels of the search image; and

the plurality of processing circuits is four in number.

38. The apparatus of claim 33 configured as a video capture device.

39. The apparatus of claim 33 configured as a video display device.

40. A method of processing of video image data:

inputting pel data for a plurality of pels of a reference image defined by object pels and background pels such that the pel data for an object pel represents a value within a predetermined range and the pel data for a background pel is set to a fixed value outside the predetermined range;

inputting pel data for a corresponding plurality of pels of a search image;

conducting a corresponding plurality of processes with respect to the inputted pel data, each process performed with respect to pel data for a different pel of the reference image and pel data of a corresponding different pel of the search image such that;

a comparative value is produced where the pel of the reference image is an object pel, and

a zero value is produced where the pel of the reference image is a background pel; and

selectively combining the values produced by the plurality of processing circuits and outputting a collective comparative value with respect to the plurality of search image pels.

41. The method of claim 40 further comprising:

inputting a collective comparative value previously produced with respect to the plurality of pels of the search image for which pel data is input;

wherein the combining accumulates the inputted previously produced collective comparative value into the collective comparative value that is output.

42. The method of claim 41 where pel data for each pel is a byte of eight bits wherein:

pel data for a set of four pels of the reference image is input where the pel data for a background pel is set to a zero value;

pel data for a set of four pels of the search image is input; and

four of the processes are conducted with respect to the inputted pel data.

43. The method of claim 40 wherein the plurality of processes process the pel data for the pels of the reference image with pel data for pels of the search image by calculating the absolute value of the difference of a pel value of each pel of the reference image set of pels with a pel value of a corresponding pel of the search image set of pels.

44. The method of claim 43 where pel data for each pel is a byte of eight bits wherein:

pel data for a set of four pels of the reference image is input where the pel data for a background pel is set to a zero value;

pel data for a set of four pels of the search image is input; and

four of the processes are conducted with respect to the inputted pel data.

45. The method of claim 44 performed by an integrated circuit in response to a microinstruction that identifies a Masked SAD operation to be performed.