CN1674669A

Movatterモバイル変換

Info

Publication number: CN1674669A
Application number: CN 200410031223
Authority: CN
Inventors: 赵柏伟
Original assignee: Realtek Semiconductor Corp
Current assignee: Realtek Semiconductor Corp
Priority date: 2004-03-26
Filing date: 2004-03-26
Publication date: 2005-09-28
Anticipated expiration: 2024-03-26
Also published as: CN100385937C

Abstract

The invention provides a method and apparatus for de-interlacing video data by motion compensation, the video data including first, second and third fields, the method comprising: performing displacement estimation on a first pixel of the first field and a second target field, and generating a first displacement vector, wherein the second target field corresponds to the second field; generating a first reference pixel corresponding to the first pixel according to the first displacement vector and the second field; and generating a target pixel of a target frame according to the first pixel and the first reference pixel.

Description

Utilize the image uninterleaving method and the device of motion vector compensation

Technical field

The present invention relates to a kind of image uninterleaving method and apparatus, particularly relate to a kind of method and apparatus that utilizes the image uninterleavingization of bit shift compensation (motion compensation).

Background technology

Traditional interleaved type scanning (interlaced scanning) technology is that staggered the demonstration done in two fields forming this frame.These two fields are made up of strange field and even place.For formula scanning technique in proper order, be called noninterlaced scanning again, its then be earlier with a Qi Chang with an even occasion and become a corresponding frame, show this frame with the horizontal frequency that doubles then, thus, the quality of picture and stability all can promote.

With two occasions and before becoming a frame, need the correct relation that detects in the video signal data between each each other, that is need detect a whether corresponding film mode (for example being to be a 3-2pull down film mode or a 2-2 pull down film mode) of video signal data itself, impose correct deinterleave operation again, if this video signal data is a film mode, then suitable two adjacent fields are merged, and if this video signal data is not a film mode, then need interpolation is carried out correctly to finish the deinterleave processing in each field of this video signal data.Known technology merely utilizes interpolation field to mend interpolation (inter-field interpolation) between (intra-fieldinterpolation), field, or displacement adjusts the processing method of (motion-adaptive), and is disclosed with No. 6512550 as No. the 6577345th, United States Patent (USP).

Yet, how to lower the wrongheaded situation of pixel value, and make the image after the deinterleave processing have preferable picture quality, or reduce the related hardware cost, just become an important topic.

Summary of the invention

The purpose of this invention is to provide a kind of method and apparatus that utilizes bit shift compensation to come a video signal data is carried out the deinterleave processing, to promote picture quality and to reduce the related hardware cost.

The invention provides a kind of bit shift compensation with method with a video signal data deinterleaveization, this video signal data includes first, second and third (field), this method includes: this one first pixel of first and one second target field are carried out displacement estimate, and produce one first motion vector, wherein this second target field is corresponding with this second; According to this first motion vector and this second, produce one first reference pixel corresponding to this first pixel; And, produce an object pixel of a target frame (frame) according to this first pixel and this first reference pixel.

A kind of image uninterleaving circuit, be used for deinterleave one video signal data, this video signal data comprises first, second and third, this image uninterleaving circuit includes: module is estimated in one first displacement, being used for that this one first pixel of first and one second target field are carried out displacement estimates, produce one first motion vector, wherein this second target field is corresponding with one second; One first pixel selection module according to this first motion vector and this second, produces one first reference pixel corresponding to one first pixel; And an occasion and a module, according to this first pixel and this first reference pixel, to produce an object pixel of a target frame (frame).

The present invention finds at least one pixel reference value of a corresponding pixel by correct motion vector (motion vector), then this pixel reference value is made suitable weighted blend (weightblending), avoid causing image quality to descend, thereby can obtain a comparatively smooth-going image because of predicting the outcome of mistake.

Description of drawings

Fig. 1 is the flow chart of first embodiment of image uninterleaving method of the present invention.

Fig. 2 is the functional block diagram of imageuninterleaving circuit 200 of the present invention.

Fig. 3 is the flow chart of second embodiment of image uninterleaving method of the present invention.

Fig. 4 is the functional block diagram of image uninterleaving circuit 300 of the present invention.

Embodiment

Please consult Fig. 1 and Fig. 2 simultaneously, Fig. 1 is the flow chart of one first embodiment of image uninterleaving method of the present invention.This method is to utilize bit shift compensation to carry out the deinterleave computing.The inventive method successively is respectively second F by receiving three adjacent fields in this video signal data in regular turn_N-1, first F_nWith the 3rd F_N+1Fig. 2 is the functional block diagram of imageuninterleaving circuit 200 of the present invention.This imageuninterleaving circuit 200 includes aninterpolation field module 10,module 11, apixel selection module 12, adecision logic module 14, aninternal memory 15 and an occasion andmodule 16 are estimated in a displacement.Displacement is estimatedmodule 11 and is included a displacement and estimate unit 11a;Pixel selection module 12 includespixel selection unit 12a and 12b;Decision logic module 14 includes decision logic unit 14a.The running of this first embodiment is as described below:

Step 202 interpolation field step:interpolation field module 10 receives and first F of foundation_nOne first pixel obtain the interpolation field reference pixel P_intra of a pixel P_mc among the target field F_mc.Wherein, interpolation field reference pixel P_intra can be produced by known interpolation operation (intra-fieldinterpolation);

Step is estimated instep 204 displacement: displacement is estimated unit 11a and is received also according to first F_nCarry out displacement with a preceding target field F_mc-1 and estimate, and produce corresponding first F_nThe one first motion vector MV_1 and the one first similarity S of this first pixel₁Wherein, the first motion vector MV_1 represents the displacement of the possible respective pixel of this first pixel and preceding target field F_mc-1, if the value of the first motion vector MV_1 is 0, represents that then the pixel P_mc of this first pixel and target field F_mc does not have displacement; The first similarity S₁Represent the similar situation of this first pixel and the possible respective pixel of preceding target field F_mc-1.Among one embodiment, the first similarity S₁Ask method directly to subtract each other to draw a difference value with this first pixel and its respective pixel, as the big more expression first similarity S of difference value₁More little, as the more little expression first similarity S of difference value₁Big more;

Step

206 and 208 is the pixel selection step:

Step 206:pixel selection unit 12a receives and the foundation first motion vector MV_1 and one second F_N-1, produce one first reference pixel P_inter_n-1 corresponding to this first pixel;

Step 208: pixel selection modular unit 12b receives and the foundation first motion vector MV_1 and one the 3rd F_N+1, produce one second reference pixel P_inter_n+1 corresponding to this first pixel;

Step 212: the pixel P_mc ofinternal memory 15 receiving target field F_mc, pixel P_mc and this first pixel of occasion andmodule 16 receiving target field F_mc merge, and export the object pixel of a target frame (frame) Fr_n; And

Step 214: finish.

In another embodiment, can not consider interpolation field reference pixel P_intra, thenstep 202 is omissible.In another embodiment, can not consider the second reference pixel P_inter_n+1, thenstep 208 is omissible.In another embodiment, do not consider the interpolation field reference pixel P_intra and the second reference pixel P_inter_n+1, then step 202,208 and 210 is omissible.

If first F_nBe one strange, then target field F_mc is an idol field, and if first F_nBe an idol field, then target field F_mc is Yi Qichang.

Fig. 3 utilizes bit shift compensation for the inventive method and carries out the flow chart of second embodiment of deinterleave processing.The inventive method successively is respectively second F in regular turn by receiving three adjacent fields in this video signal data_N-1, first F_nWith the 3rd F_N+1Fig. 4 is the functional block diagram of second kind of image uninterleaving circuit 300 of the present invention.Image uninterleaving circuit 300 includes aninterpolation field module 10,module 11, apixel selection module 12, adecision logic module 14, aninternal memory 15 and an occasion andmodule 16 are estimated in a displacement.Wherein, displacement is estimatedmodule 11 and is included displacement and estimate unit 31,32 and 33;Pixel selection module 12 includes pixel selection unit 34,35,36 and 37;Decision logic module 14 includes decision logic unit 41,42,43 and 44.Please consult Fig. 3 and Fig. 4 simultaneously, the running of this second embodiment is as described below:

Step 302 is identical withstep 202;

Step 304,306 and 308 is estimated step for displacement:

Step 304: displacement is estimated unit 31 and is received also according to first F_nCarry out displacement with a preceding target field F_mc-1 and estimate, produce corresponding first F_nThe one first motion vector MV_1 and the one first similarity S of this first pixel₁

Step 306: displacement is estimated unit 32 and is received first F_nWith one second F_N-1, export one second motion vector MV_2 and one second similarity S₂

Step 308: displacement is estimated unit 33 and is received first F_nWith one the 3rd F_N+1, export a triple motion vector M V_3 and a third phase and seemingly spend S₃

Step 310,312,314 and 316 is the pixel selection step:

Step 310: pixel selection unit 34 is according to the first motion vector MV_1 and second F_N-1Corresponding pixel, produce one first reference pixel P_inter_n-1_a corresponding to this first pixel;

Step 312: pixel selection unit 35 is according to the second motion vector MV_2 and second F_N-1Corresponding pixel, produce one the 3rd reference pixel P_inter_n-1_b corresponding to this first pixel;

Step 314: pixel selection unit 36 is according to triple motion vector M V_3 and one the 3rd F_N+1Corresponding pixel, produce one the 4th reference pixel P_inter_n+1_a corresponding to this first pixel;

Step 316: pixel selection unit 37 is according to the first motion vector MV_1 and one the 3rd F_N+1Corresponding pixel, produce one second reference pixel P_inter_n+1_b corresponding to this first pixel;

Step 318,320,322 and 324 is a similarity determining step:

Step 318: decision logic unit 41 is according to this first and second similarity S₁, S₂, adjust the weighting ratio w1 of the first reference pixel P_inter_n-1_a and the 3rd reference pixel P_inter_n-1_b, and produce a Wucan respectively according to this weighting w1 and examine pixel P_inter_n-1 and one the 4th similarity S₄In one embodiment, its to ask method be S₄=w₁* S₁+ (1-w₁) * S₂, P_inter_n-1=w₁* P_inter_n-1_a+ (1-w₁) * P_inter_n-1_b;

Step 320: decision logic unit 42 according to this first and third phase seemingly spend S₁, S₃, judge the weighting ratio of the 4th reference pixel P_inter_n+1_a and the second reference pixel P_inter_n+1_b, to produce one the 6th reference pixel P_inter_n+1 and one the 5th similarity S₅

Step 322: decision logic unit 43 is according to the 4th and the 5th similarity S₄, S₅, judge that Wucan examines the weighting ratio of pixel P_inter_n-1 and the 6th reference pixel P_inter_n+1, to produce one the 7th reference pixel P_inter and one the 6th similarity S₆

Step 324: decision logic unit 44 is according to the 6th similarity S₆Adjust the weighting ratio w6 of interpolation field reference pixel P_intra and the 7th reference pixel P_inter, to produce pixel P_mc to target field F_mc that should first pixel.In the present embodiment, P_mc=w6 * P_inter+ (1-w6) * P_intra;

Step 326: the pixel P_mc ofinternal memory 15 receiving target field F_mc, pixel P_mc and this first pixel of occasion andmodule 16 receiving target field F_mc merge, and export the object pixel of a target frame (frame) Fr_n; And

Step 328: finish.

Above-mentioned steps is the preferred embodiment of this second embodiment, but also clipped step certainly, and the pixel selection step of step or part or similarity determining step partly etc. are estimated in the displacement of for example omitting step 302 or part.

Wherein, among the embodiment, the determination methods of decision logic unit 41: as first and second similarity S₁, S₂During all greater than a critical value, P_inter_n-1=w then₁* P_inter_n-1_a+ (1-w₁) * P_inter_n-1_b, wherein, weighting w₁=(S₁/ (S₁+ S₂)); If the first similarity S₁Greater than this critical value and the second similarity S₂Less than this critical value, then P_inter_n-1=P_inter_n-1_a; In addition, the 4th similarity S₄Obtain with above-mentioned judgment mode.Wherein this decision logic unit 42,43 and 44 determination methods can be similar.

In the present embodiment, target field that last place is obtained system is used for the auxiliary corresponding target field of obtaining present processing midfield, so the aforesaid operations flow process just can be used to calculate in the video signal data a plurality of corresponding target field, reaches the purpose of deinterleaveization at last.Second embodiment also utilizes three field F_N-1, F_n, F_N+1Predict other possibility pixel value (that is P_inter_n+1_a and P_inter_n+1_b) of each pixel in the target field, therefore when carrying out the subsequent weight computing and set the pixel value of each pixel in this target field, present embodiment can produce preferable operation result and reach and avoid the influence of inappropriate possible pixel value to actual pixel value.

The above only is preferred embodiment of the present invention, and all equalizations of doing according to claim of the present invention change and modify, and all should belong to the covering scope of patent of the present invention.

Claims

1. method is used for deinterleave one video signal data, and this video signal data includes first, second and third, and this method includes:

This one first pixel of first and one second target field are carried out displacement estimate, and produce one first motion vector, wherein this second target field is corresponding with this second;

According to this first motion vector and this second, produce one first reference pixel corresponding to this first pixel; And

According to this first pixel and this first reference pixel, produce an object pixel of a target frame.

2. the method for claim 1, wherein estimate in the step in this displacement, also produce one first similarity, this method also includes:

Carry out this interpolation field of first, produce an interpolation field reference pixel; And

According to this first similarity, adjust the weighting ratio of those reference pixels, to produce a bit shift compensation pixel, in order to replace this first reference pixel.

3. the method for claim 1, wherein estimate in the step in this displacement, also produce one first similarity, this method also includes:

According to this first motion vector and the 3rd, produce one second reference pixel corresponding to this first pixel; And

4. the method for claim 1, wherein estimate in the step in this displacement, also produce one first similarity, this method also includes:

Carry out displacement to this first and this second and estimate, and produce one second motion vector and one second similarity;

According to this second motion vector and this second, produce one the 3rd reference pixel corresponding to this first pixel; And

According to those similarities, adjust the weighting ratio of those reference pixels, to produce a bit shift compensation pixel, in order to replace this first reference pixel.

5. method as claimed in claim 4, this method also includes:

This first is carried out displacement and estimate with the 3rd, and produce triple motion vector and a third phase like degree; And

According to this triple motion vector with the 3rd, produce one the 4th reference pixel corresponding to this first pixel.

6. as claim 3 or 4 or 5 described methods, this method also includes:

Carry out this interpolation field of first, produce an interpolation field reference pixel.

7. image uninterleaving circuit is used for deinterleave one video signal data, and this video signal data comprises first, second and third, and this image uninterleaving circuit includes:

The unit is estimated in one first displacement, is used for that this one first pixel of first and one second target field are carried out displacement and estimates, and produces one first motion vector, and wherein this second target field is corresponding with this second;

One first pixel selection unit according to this first motion vector and this second, produces one first reference pixel corresponding to one first pixel; And

One occasion and module are according to this first pixel and this first reference pixel, to produce an object pixel of a target frame.

8. image uninterleaving circuit as claimed in claim 7, wherein, this first displacement is estimated the unit and is produced one first similarity, and this image uninterleaving circuit also includes:

One interpolation field module is used for producing an interpolation field reference pixel according to this first; And

The weighting ratio of those reference pixels according to this first similarity, is adjusted in one first decision logic unit, to produce a bit shift compensation pixel, in order to replace this first reference pixel.

9. image uninterleaving circuit as claimed in claim 7, wherein, this first displacement is estimated the unit and is produced one first similarity, and this image uninterleaving circuit also includes:

One second pixel selection unit according to this first motion vector and the 3rd, produces one second reference pixel corresponding to this first pixel;

The weighting ratio of those reference pixels according to this first similarity, is adjusted in one second decision logic unit, to produce a bit shift compensation pixel, in order to replace this first reference pixel.

10. image uninterleaving circuit as claimed in claim 7, wherein, this first displacement is estimated the unit and is produced one first similarity, and this image uninterleaving circuit also includes:

The unit is estimated in one second displacement, carries out displacement to this first and this second and estimates, and produce one second motion vector and one second similarity;

One the 3rd pixel selection unit according to this second motion vector and this second, produces one the 3rd reference pixel corresponding to this first pixel; And

The weighting ratio of those reference pixels according to those similarities, is adjusted in one the 3rd decision logic unit, to produce a bit shift compensation pixel, to replace this first reference pixel.

11. image uninterleaving circuit as claimed in claim 10, this image uninterleaving circuit also includes:

One triple motion is estimated the unit, this first is carried out displacement and estimates with the 3rd, and produce triple motion vector and a third phase like degree; And

One the 4th pixel selection unit, according to this triple motion vector with the 3rd, produce one the 4th reference pixel corresponding to this first pixel.

12., also include as claim 9 or 10 or 11 described image uninterleaving circuit:

One interpolation field module is used for exporting an interpolation field reference pixel according to this first.