Motion detection for interlaced video

• US 8,582,032 B2
• Filed: 09/07/2007
• Issued: 11/12/2013
• Est. Priority Date: 09/07/2006
• Status: Active Grant

• Current Assignee: Texas Instruments, Inc.
• Original Assignee: Texas Instruments, Inc.
• Inventors: Zhai, Fan, Chang, Weider Peter
• Primary Examiner(s): ALAM, MUSHFIKH I
• Application Number:
  US11/852,215
• Publication Number:
  US 20080062309A1
• Time in Patent Office: 2,258 Days
• Field of Search: 348/448, 348/452, 348/700
• US Class Current: 348/452
• CPC Class Codes: G06T 3/4084 in the transform domain, e.g. fast Fourier transform [FFT] domain scaling
• CPC Class Codes: G06T 7/254 involving subtraction of images
• CPC Class Codes: H04N 5/144 Movement detection for video coding H04N19/503; analysis of motion in general G06T7/20
• CPC Class Codes: H04N 7/012 Conversion between an interlaced and a progressive signal for graphics images G09G2310/0229
• CPC Class Codes: H04N 7/0135 involving interpolation processes interpolation-based image scaling G06T3/4007; interpolation for video coding H04N19/587, H04N19/59
• CPC Class Codes: H04N 7/0137 dependent on presence/absence of motion, e.g. of motion zones H04N7/014 takes precedence; movement detection in television signals H04N5/144
• CPC Class Codes: H04N 7/0142 the interpolation being edge adaptive edge detection in television signals H04N5/142; edge-driven scaling G06T3/403

Abstract

Motion detection in interlaced video fields, as useful in de-interlacing, includes spatial-temporal maximum filtering, temporal IIR filtering dependent upon spatial-temporal variance, and spatial variance dependent moving-still interpolation blending factor.

Claims

1. A method of motion detection for interlaced fields, comprising the steps of:

(a) for a pixel (x,y) of parity differing from the parity of a first field, computing a first motion measure for said pixel (x,y) from differences between the values of pixels in a neighborhood of (x,y) in the field preceding said first field and the values of corresponding neighborhood pixels in the field following said first field;

(b) filtering said first motion measure for said pixel (x,y) to give a second motion measure for said pixel (x,y) wherein said filtering includes texture-adaptive spatial-temporal filtering and an infinite-impulse-response (IIR) filtering with filter coefficient adapting to a measure of spatial-temporal variance about said pixel (x,y) for capturing fast and periodic motion in interlaced video sequence, wherein the IIR filtering is reset at video scene changes when a level of difference between two neighboring fields with the same parity is greater than a threshold; and

(c) computing an output motion measure for said pixel (x,y) from said second motion measure for said pixel (x,y).

2. The method of claim 1, wherein said filtering said first motion measure includes spatial maximum filtering prior to said IIR filtering.

3. The method of claim 1, wherein said filtering said first motion measure includes spatial-temporal maximum filtering prior to said IIR filtering.

4. The method of claim 1, wherein said filter coefficient adapting to a measure of spatial-temporal variance includes adapting to a scene change detection.

5. The method of claim 1, wherein said computing a first motion measure includes lowpass filtering prior to said differences.

6. The method of claim 1, wherein said computing an output motion measure uses a measure of spatial variance about said pixel (x,y).

7. A method of video de-interlacing, comprising the steps of:

(a) detecting motion for a pixel (x,y) of parity differing from the parity of a first field;

(b) computing a value for said pixel (x,y) as a blend of a spatial interpolation for said pixel (x,y) and a temporal interpolation for said pixel (x,y) where said blend depends upon said motion detection;

(c) wherein said detecting motion includes (i) computing a first motion measure for said pixel (x,y) from differences between the values of pixels in a neighborhood of (x,y) in the field preceding said first field and the values of corresponding pixels in the field following said first field;

(ii) filtering said first motion measure for said pixel (x,y) to give a second motion measure for said pixel (x,y) wherein said filtering includes texture-adaptive spatial-temporal filtering and an infinite-impulse-response (IIR) filtering with filter coefficient adapting to a measure of spatial-temporal variance about said pixel (x,y) for capturing fast and periodic motion in interlaced video sequence, wherein the IIR filtering is reset at video scene changes when a level of difference between two neighboring fields with the same parity is greater than a threshold; and

(iii) computing an output motion measure for said pixel (x,y) from said second motion measure for said pixel (x,y).

8. The method of claim 7, wherein said computing as a blend includes use of a lookup table indexed by said motion detection.

9. The method of claim 7, wherein said filtering said first motion measure includes spatial maximum filtering prior to said IIR filtering.

10. The method of claim 7, wherein said filtering said first motion measure includes spatial-temporal maximum filtering prior to said IIR filtering.

11. The method of claim 7, wherein said filter coefficient adapting to a measure of spatial-temporal variance includes adapting to a scene change detection.

12. The method of claim 7, wherein said computing a first motion measure includes lowpass filtering prior to said differences.

13. The method of claim 7, wherein said computing an output motion measure uses a measure of spatial variance about said pixel (x,y).