Deinterlacing method for video signals based on motion-compensated interpolation

• US 6,606,126 B1
• Filed: 09/01/2000
• Issued: 08/12/2003
• Est. Priority Date: 09/03/1999
• Status: Expired due to Fees

• Current Assignee: LG Electronics, Inc. (LG Corporation)
• Original Assignee: LG Electronics, Inc. (LG Corporation)
• Inventors: Min, Cheol Hong, Lim, Il Taek, Lim, Kyoung Won
• Primary Examiner(s): Kostak, Victor R.
• Application Number:
  US09/654,168
• Time in Patent Office: 1,075 Days
• Field of Search: 348/452, 348/448, 348/700, 348/459, 348/701, 382/299, 382/300
• US Class Current: 348/452
• CPC Class Codes: H04N 7/012 Conversion between an interlaced and a progressive signal for graphics images G09G2310/0229
• CPC Class Codes: H04N 7/014 involving the use of motion vectors motion estimation and compensation in video coding H04N19/51

Abstract

Disclosed is a motion-compensated interpolation (MCI)-based deinterlacing method for converting video signals of an interlaced scanning format into those of a progressive scanning format. The deinterlacing method involves a motion vector estimation step, a video signal deinterlacing step, a deinterlacing side-effect check step, and a deinterlaced result select step. The motion vector estimation step includes the steps of estimating a motion vector by the unit of a ½ pixel, using the previous frame, the current field, and the future field, and estimating a motion vector by the unit of a ¼ pixel, using the previous frame and the future field. Based oh a motion vector estimated at the motion vector estimation step, the video signal is deinterlaced. The deinterlacing side-effect check is achieved by conducting a check for determining whether or not the estimated motion vector has a temporal consistency, a check for determining whether or not the estimated motion vector has a critical velocity, and a check for determining whether or not the estimated motion vector has a backward matching accuracy. When the result of the side-effect check is satisfactory, the result of the deinterlacing is accepted. On the other hand, when the result of the side-effect check is unsatisfactory, the result of a deinterlacing conducted based on a method other than the MIC is accepted.

Claims

1. A deinterlacing method for video signals based on a motion-compensated interpolation comprising:

(a) estimating a motion vector of a video signal by a first fractional unit of a pixel and the unit of a second fractional unit of a pixel, respectively, using a previous frame of the video signal, a current field of the video signal, and a future field of the video signal;

(b) deinterlacing the current field by interpolating empty pixels of the current field by pixels of the previous frame designated by the estimated motion vector, thereby deinterlacing the current field;

(c) conducting a side-effect check for determining whether a result of the deinterlacing is satisfactory; and

(d) selectively outputting the result of the deinterlacing as a deinterlaced value of the current field based upon the result of the side effect check.

2. The deinterlacing method according to claim 1, wherein the motion vector estimation step comprises the steps of:

estimating a motion vector by the unit of a ½

pixel, using the previous frame, the current field, and the future field; and

estimating a motion vector by the unit of a ¼

pixel, using the previous frame and the future field.

3. The deinterlacing method according to claim 2, wherein the motion vector estimation is carried out, using a basic block having a size of 4×

4 applied to a frame.

4. The deinterlacing method according to claim 2, wherein the motion vector estimation is carried out, using a basic block having a size of 8×

8 defined by a basic block of 4×

4, applied to a frame, and pixels added to each edge of the basic block of 4×

4, based on a guard region.

5. The deinterlacing method according to claim 1, wherein when the deinterlacing is conducted for empty pixels of the current field by bringing pixels of the previous frame designated by the motion vector, the motion vector estimated by the unit of a ¼

pixel is derived by conducting a weighted average for four pixels, arranged around a position of the pixel to be brought, in accordance with a bi-linear interpolation.

6. The deinterlacing method according to claim 1, wherein the side-effect check is made by conducting at least one of a check for determining whether or not the estimated motion vector has a temporal consistency, a check for determining whether or not the estimated motion vector has a critical velocity, ,and a check for determining whether or not the estimated motion vector has a backward matching accuracy.

7. The deinterlacing method according to claim 6, wherein the check with respect to a temporal consistency of the estimated motion vector comprises:

determining whether a difference between a motion vector currently derived and a motion vector previously derived is less than a predetermined critical value;

determining whether mean absolute differences respectively derived by the unit of a ½

pixel and the unit of a ¼

pixel are determined not to be considerably different from each other while being sufficiently small; and

wherein when a difference between the motion vector currently derived and the motion vector previously derived is less than the predetermined critical value, and the mean absolute differences respectively derived by the unit of a ½

pixel and the unit of a ¼

pixel are determined not to be considerably different from each other while being sufficiently small, then determining the estimated motion vector to have the temporal consistency.

8. The deinterlacing method according to claim 6, wherein the check with respect to a critical velocity of the estimated motion vector comprises:

checking a vertical component of the estimated motion vector to determine whether the estimated motion vector is within a predetermined critical velocity range; and

wherein when the estimated motion vector is within the predetermined critical velocity range, then outputting the estimated motion vector from the side-effect check step, and otherwise preventing the estimated motion vector from being outputted from the side-effect check step.

9. The deinterlacing method according to claim 6, wherein the check with respect to a backward matching accuracy comprises:

when a current block is projected onto the previous frame, based on the estimated motion vector, determining whether there is a video similar to that of the current block in a block of the previous frame onto which the current block is projected; and

wherein when there is a video similar to that of the current block in the projected block of the previous frame, then outputting the estimated motion vector from the side-effect check, and otherwise preventing the estimated motion vector from being outputted from the side-effect check.

10. The deinterlacing method according to claim 9, wherein the determination of whether or not the estimated motion vector has a backward matching accuracy is conducted using, as a reference, a largest one of respective absolute values of differences between corresponding pixels of a block of the current field and a corresponding block of the previous frame.

11. The deinterlacing method according to claim 6, wherein the result of the deinterlacing based on the currently derived motion vector is outputted when the motion vector has a temporal consistency, when the motion vector is not within a predetermined critical velocity range, and when the motion vector has a reliable backward matching accuracy.

15. The deinterlacing method according to claim 1, wherein the selectively outputting comprises:

if a result of the side-effect check is satisfactory, then outputting the result of the deinterlacing as a deinterlaced value of the current field; and

if not, then outputting the result of a deinterlacing conducted in accordance with a method other than that of the step (b).

16. The deinterlacing method according to claim 1, wherein the first and second fractional units are less than 1.

17. The deinterlacing method according to claim 16, wherein the first fractional unit is a ½

pixel and wherein the second fractional unit is a ¼

pixel.

18. The deinterlacing method according to claim 1, wherein the side effect check determines whether the result of the deinterlacing meets prescribed characteristics.

19. The deinterlacing method according to claim 1, wherein the previous frame, the current field, and the future field correspond to sequential fields of the video signals.

12. A deinterlacing method for video signals based on a motion-compensated interpolation comprising:

(a) estimating a motion vector of a video signal by the unit of a ½

pixel, using a previous frame of the video signal, a current field of the video signal, and a future field of the video signal, and estimating a motion vector by the unit of a ¼

pixel based on the motion vector estimated by the unit of a ½

pixel while using the previous frame and the future field, in at least two search areas of different sizes each including a block for which the motion vector estimation is to be carried out;

(b) deinterlacing the video signal by bringing pixels of the previous frame designated by a motion vector estimated at the step (a) to the current field;

(c) conducting a side effect check for determining whether the deinterlaced video signal is reliable by checking whether the estimated motion vector has at least one of a temporal consistency, a critical velocity and a backward matching accuracy in each search area of the deinterlaced video signal; and

(d) if the result of the side-effect check is satisfactory, then outputting the result of the deinterlacing as a deinterlaced value of the current field, and if not,then outputting the result of a deinterlacing conducted in accordance with a method other than that of the step (b).

13. The deinterlacing method according to claim 12, wherein:

the deinterlacing of the video signal is carried out, based on the estimated motion vectors for each of at least two search areas of different sizes;

the reliability of the deinterlacing in association with the search areas is determined according to the result of the temporal consistency check, the critical velocity check and the backward matching accuracy check of the motion vector for each deinterlaced video signal; and

if there are at least two reliable motion vectors, a deinterlaced video signal derived based on the motion vector associated with a smaller one of the search areas is selected from the deinterlaced video signals, and outputted.

14. The deinterlacing method according to claim 13, wherein the at least two search areas comprises three search areas respectively having a size of m×

n (m, n>

0) in horizontal and vertical directions, a size of m×

0, and a size of 0×

0 within a pixel area ranging from “

−

Mh”

to “

+Mh”

in a horizontal direction while ranging from “

−

Mv”

to “

+Mv”

in a vertical direction, with respect to a relation between the previous frame and a future field.

20. A deinterlacing method for video signals comprising:

(a) estimating a motion vector of a video signal by a first fractional unit of a pixel and a second fractional unit of a pixel respectively, using a previous frame of the video signal, a current field of the video signal, and a future field of the video signal in at least two search areas of different sizes each including a block for which the motion vector estimation is to be carried out;

(b) deinterlacing the current field by interpolating empty pixels of the current field by pixels of the previous frame designated by the estimated motion vector to deinterlaced the current field;

(c) conducting a side-effect check for determining whether a result of the deinterlacing meets prescribed characteristics in each search area of the deinterlaced video signal; and

(d) selectively outputting the result of the deinterlacing as a deinterlaced value of the current field based upon the result of the side effect check.

21. The deinterlacing method of claim 20 wherein the side-effect check is made by conducting at least one of a check for determining whether the estimated motion vector has a temporal consistency, a check for determining whether the estimated motion vector has a critical velocity, and a check for determining whether or not the estimated motion vector has a backward matching accuracy in each search area of the deinterlaced video signal.

22. The deinterlacing method according to claim 20, wherein the first and second fractional units are less than 1.

23. The deinterlacing method according to claim 20, wherein the first fractional unit is a ½

pixel and wherein the second fractional unit is a ¼

pixel.

24. A deinterlacing method for video signals comprising:

(a) deinterlacing a video signal using motion compensation;

(b) conducting a side-effect check by conducting at least one of a check for determining whether an estimated motion vector has a temporal consistency, a check for determining whether the estimated motion vector has a critical velocity, and a check for determining whether the estimated motion vector has a backward matching accuracy; and

(c) selectively outputting the result of the deinterlacing as a deinterlaced value of the current field based upon the result of the side effect check.

25. She deinterlacing method according to claim 24, wherein a motion vector of a video signal is estimated by using a first fractional unit of a pixel and a second fractional unit of a pixel respectively, using a previous frame of the video signal, a current field of the video signal, and a future field of the video signal and deinterlacing the current field is performed by interpolating empty pixels of the current field by pixels of the previous frame designated by the estimated motion vector to deinterlaced the current field.

26. The deinterlacing method according to claim 24, wherein estimating a motion vector by a unit of a ¼

pixel based on a motion vector estimated by a unit of ½

pixel using the previous frame and the future field, in at least two search areas of different sizes each including a block for which the motion vector estimation is to be carried out.