FIELD SEQUENCE DETECTOR, METHOD AND VIDEO DEVICE

• US 20070258014A1
• Filed: 05/02/2006
• Published: 11/08/2007
• Est. Priority Date: 05/02/2006
• Status: Active Grant

• Current Assignee: ATI Technologies ULC (Advanced Micro Devices, Inc.)
• Original Assignee: ATI Technologies Incorporated (Advanced Micro Devices, Inc.)
• Inventors: Doswald, Daniel
• Granted Patent: US 7,953,293 B2
• Time in Patent Office: Days
• US Class Current: 348/701
• CPC Class Codes: H04N 17/00 Diagnosis, testing or measuring for television systems or their details
• CPC Class Codes: H04N 7/0115 with details on the detection of a particular field or frame pattern in the incoming video signal, e.g. 3:2 pull-down pattern
• CPC Class Codes: H04N 7/012 Conversion between an interlaced and a progressive signal for graphics images G09G2310/0229

Abstract

A field sequence detector determines the field sequence of a series of fields of video by assessing the vertical frequency content of hypothetical de-interlaced images. Hypothetical images are formed from a currently processed field and an adjacent (e.g. previous or next) field. If the vertical frequency content is relatively high (e.g. above ½ the Nyquist frequency for the image), the hypothetical image is assessed to be formed of improperly interlaced fields, belonging to different frames. If the frequency content is relatively low, the hypothetical image is assessed to be properly assembled from fields of the same frame. The field sequence in the series of fields may be detected from the assessed frequency content for several of said series of fields. Known field sequence, such as 3:2 pull-down, 2:2 pull down, and more generally m:n:l:k pull-down sequences.

Claims

1. A method of detecting a field sequence of a series of fields of video, said method comprising:

processing each of said series of fields by assessing the frequency content along the vertical direction of a first hypothetical image formed by interleaving alternating rows from the currently processed field and rows from an adjacent field, to determine the presence of high frequency vertical artifacts in said first hypothetical image;

detecting said field sequence in said series of fields, from the sequence of first hypothetical images determined to have high frequency vertical artifacts.

2. The method of claim 1, wherein said assessing comprises performing a frequency analysis of pixels within a column about addressed pixels of said first hypothetical image.

3. The method of claim 2, wherein said assessing comprises performing a wavelet analysis of said pixels within said column about said addressed pixels.

4. The method of claim 2, wherein said assessing comprises performing a Fourier analysis of said pixels within said column about said addressed pixels.

5. The method of claim 1, wherein said detecting comprises recognizing high frequency content along the vertical direction of said first hypothetical image, for a defined pattern of said series of fields.

6. The method of claim 1, wherein said assessing comprises using multiple frequency analyses along columns of said first hypothetical image to classify said first hypothetical image as having or not having high frequency content in a vertical direction.

7. The method of claim 1, wherein said assessing comprises determining magnitudes of dominant frequency components in the vertical direction, above a threshold frequency, proximate addressed pixels in said first hypothetical image.

8. The method of claim 7, wherein said assessing comprises analysing a distribution of said magnitudes and classifying said first hypothetical image as having high frequency content in the vertical direction, based on said distribution.

9. The method of claim 8, further comprising filtering said distribution of magnitudes, using a distribution of magnitudes of dominant frequencies, above said threshold, in a hypothetical image not having vertical artifacts.

10. The method of claim 1, wherein said field sequence is detected as a 3:

2 pull down sequence, in response to assessing two non-sequential ones of said first hypothetical images for every five fields as having high frequency content in the vertical direction.

11. The method of claim 1, wherein said field sequence is detected as a 2:

2 pull down sequence in said series of fields, in response to assessing said first hypothetical image as having high vertical frequency content for every alternating one of said series of fields.

12. The method of claim 1,

wherein said first hypothetical image is formed by interleaving alternating rows from the currently processed field and rows from an immediately next field,

said method further comprising:

sequentially processing each of said series of fields by constructing at least a portion of a second hypothetical image formed by interleaving alternating rows from the currently processed field and rows from an immediately previous field;

assessing the frequency content along the vertical direction of said second hypothetical image to determine the presence of a high frequency vertical artifacts in said second hypothetical image.

13. The method of claim 12, wherein said currently processed field and immediately next field are de-interlaced in response to detecting relatively low frequency content along the vertical direction of said first hypothetical image.

14. The method of claim 12, wherein said currently processed field and immediately previous field are de-interlaced in response to detecting relatively low frequency content along the vertical direction of said second hypothetical image.

15. The method of claim 13, further comprising forming differences between pixels in said previous field and pixels in said next field, and wherein said assessing comprises using said differences, for each of said currently processed fields to classify each of said first and second hypothetical images.

16. The method of claim 1, wherein said field sequence is detected as a m:

n;

l;

k pull-down sequence in response to detecting a pattern of relatively high (H) and relatively low (L) frequency content in the vertical direction, for m+n+l+k sequential ones of said first hypothetical images, said pattern corresponding to (L)_m-1H(L)_n-1H(L)_l-1H(L)_k-1H(L)_m-1 . . .

32. Computer readable medium storing processor readable instructions, that when loaded at a computing device operable to decode video into a plurality of video fields, adapt said computing device to detect a field sequence of said series of video fields in accordance with the method of claim 1.

34. The method of claim 1, wherein said currently processed field and said adjacent field are formed from adjacent frames in a series of video frames, to detect frame repetition.

17. A detector for detecting a field sequence of a series of fields of video, said detector comprising:

a first frequency analyzer for assessing vertical frequency content in first hypothetical images formed by interleaving alternating rows from a currently processed field and rows from an adjacent field;

a processor in communication with said first frequency analyzer to determine said sequence of said series of fields, from said assessed vertical frequency content for several currently processed fields of said series of fields.

18. The detector of claim 17, wherein said first frequency analyzer determines vertical frequency content proximate individual pixels in said first hypothetical images.

19. The detector of claim 18, further comprising an address decoder for decoding addresses of pixels in said currently processed field and said adjacent field to assess said vertical frequency content proximate said individual pixels.

20. The detector of claim 19, wherein said address decoder decodes addresses for selected ones of said individual pixels in a programmed range.

21. The detector of claim 18, wherein said first frequency analyzer is operable to determine a magnitude of a dominant frequency, above a threshold frequency, in said vertical frequency content proximate each ones of said individual pixels in said first hypothetical images.

22. The detector of claim 21, further comprising an accumulator bank to store a distribution of said magnitudes of said dominant frequency for each of said first hypothetical images, said accumulator bank in communication with said processor, and wherein said processor classifies each hypothetical image as having high frequency content in a vertical direction, based on said accumulator bank.

23. The detector of claim 22, wherein said processor is further operable to adjust a content of said accumulator bank, to filter said content of said accumulator bank based on a distribution of said magnitudes for a hypothetical image without artifacts.

24. The detector of claim 17, wherein said first frequency analyzer analyzes hypothetical images formed by interleaving alternating rows from the currently processed field and rows from an immediately next field,

said detector further comprising:

a second frequency analyzer for assessing vertical frequency content in second hypothetical images formed by interleaving alternating rows from the currently processed field and rows from a previous field.

25. The detector of claim 24, wherein

said first analyzer is operable to determine a magnitude of a dominant frequency above a threshold frequency, proximate individual pixels in said first hypothetical images;

and

said second analyzer is operable to determine a magnitude of a dominant frequency above a threshold frequency, proximate individual pixels in said second hypothetical images.

26. The detector of claim 25, wherein said first and second analyzers perform wavelet analyses.

27. The detector of claim 25, wherein said first and second analyzers perform Fourier analyses.

28. The detector of claim 25, further comprising a difference calculator for calculating the difference between pixels in said immediately next field and said immediately previous field.

29. The detector of claim 17, wherein said processor is operable to detect a 3:

2 pull down sequence in said series of fields, in response to determining relatively high assessed vertical frequency content for two non-sequential ones of said first hypothetical images for every five of said currently processed fields.

30. The detector of claim 17, wherein said processor is operable to detect a 2:

2 pull down sequence in said series of fields, in response to determining relatively high assessed vertical frequency content for every alternating one of said series of fields.

33. A video device comprising a detector for determining a field sequence, formed in accordance with claim 25.

31. A device for detecting a field sequence of a series of fields of video, said device comprising:

means for processing each of said series of fields to construct at least a portion of a first hypothetical image formed by interleaving alternating rows from the currently processed field and rows from an adjacent field;

means for assessing the frequency content along the vertical direction of said first hypothetical image to determine the presence of high frequency vertical artifacts in said first hypothetical image;

means for detecting said field sequence in said series of fields, by analyzing said assessed frequency content for several of said series of fields.

35. A method comprising:

for each field of a plurality of video fields, constructing at least a portion of an associated hypothetical image formed by interleaving rows from the field and rows from an adjacent field in the plurality of video fields;

assessing the frequency content along the vertical direction of each of the associated hypothetical images to determine the presence of high frequency vertical artifacts therein;

determining a field sequence in the plurality of fields from the sequence of those associated hypothetical images determined to have high frequency vertical artifacts.

36. A method of processing a series of fields of video, said method comprising:

for each of the fields, assessing frequency content along the vertical direction of an associated hypothetical image formed by interleaving alternating rows from the field and rows from an adjacent field;

classifying each of the associated hypothetical images as having or not having high frequency vertical artifacts based on the assessing.

37. The method of claim 36, further comprising

determining a field sequence in the series of fields from the sequence of the associated hypothetical images classified as having high frequency vertical artifacts.

38. The method of claim 36, further comprising

forming frames of de-interlaced video, by combining fields and adjacent fields in the series of fields, for each field for which its associated hypothetical image is classified as not having high frequency vertical artifacts.

39. The method of claim 36, wherein the assessing comprises performing a frequency analysis of sets of pixels within columns about addressed pixels of the hypothetical image.

40. The method of claim 39, wherein the assessing comprises performing a wavelet analysis of the sets of pixels.

41. The method of claim 39, wherein the assessing comprises performing a Fourier analysis of the sets of pixels.

42. The method of claim 37, wherein the determining comprises recognizing a defined pattern of fields for which the associated hypothetical images are classified as having high frequency vertical artifacts.

43. The method of claim 36, wherein the assessing comprises performing multiple frequency analyses along columns of each associated hypothetical image.

44. The method of claim 36, wherein the assessing comprises determining magnitudes of dominant frequency components in a vertical direction, above a threshold frequency, proximate addressed pixels in the associated hypothetical image.

45. The method of claim 44, wherein the assessing comprises analysing a distribution of the magnitudes for multiple addressed pixels and the classifying is performed based on the distribution.

46. The method of claim 45, further comprising filtering the distribution of the magnitudes, using a distribution of magnitudes of dominant frequencies, above the threshold, in a hypothetical image not having vertical artifacts.

47. The method of claim 37, wherein the field sequence is determined to be a 3:

2 pull down sequence, in response to classifying associated hypothetical images for two non-sequential fields in every five fields of the series of fields, as having high frequency vertical artifacts.

48. The method of claim 37, wherein the field sequence is detected as a 2:

2 pull down sequence, in response to classifying associated hypothetical images for every second field of the series of fields, as having high frequency vertical artifacts.

49. The method of claim 36,

wherein each associated hypothetical image is formed by interleaving alternating rows from a field and rows from its immediately next field,

the method further comprising:

for each of the fields, assessing frequency content along the vertical direction of a second associated hypothetical image formed by interleaving alternating rows from the field and rows from its immediately previous field;

classifying each of the second associated hypothetical images as having or not having high frequency vertical artifacts based on the assessing of frequency content along the vertical direction of the second associated hypothetical image.

50. The method of claim 49, wherein a field and its immediately previous field are de-interlaced in response to classifying its second associated hypothetical image as not having high frequency vertical artifacts.

51. The method of claim 49, further comprising forming a difference between pixels in the immediately previous field and pixels in the immediately next field, and wherein said assessing comprises using the difference, for each of said fields to classify each of the associated hypothetical images and the second associated hypothetical images.

52. The method of claim 37, wherein the field sequence is detected as a m:

n;

l;

k pull-down sequence, in response to only classifying, in m+n+l+k adjacent fields, the m^th;

m+n^th;

m+n+l^th and m+n+l+k^th associated hypothetical images as having high frequency vertical artifacts.