CN100481926C

CN100481926C - Method and apparatus for digital video de-interlacing and field sequence detection thereof

Info

Publication number: CN100481926C
Application number: CNB2006100985850A
Authority: CN
Inventors: 汪源; 邱嵩
Original assignee: Vimicro Corp
Current assignee: Beijing Vimicro Artificial Intelligence Chip Technology Co ltd
Priority date: 2006-07-12
Filing date: 2006-07-12
Publication date: 2009-04-22
Anticipated expiration: 2026-07-12
Also published as: CN1878282A

Abstract

The invention relates to a field sequence checking method used in digit video intersection solution and relative device. Wherein, it needs to collect the sum of pixel difference absolute values (SAD) used in field sequence check and select the minimum SAD value as SADMin; judges if the field sequence of SADMin is the motion field, when the result is yes, based on the position PosMin of SADMin to judge if there is copy field represented periodically, and when the result is yes, confirming the received field sequence is folded field sequence. The invention can improve the image quality when playing video flow.

Description

A kind of method and apparatus that is used for a digital video de-interlacing and a Sequence Detection thereof

Technical field

The present invention relates to digital video technology, be specifically related to a kind of method and apparatus that is used for a digital video de-interlacing and a Sequence Detection thereof.

Background technology

Folding (Pulldown) technology of 3:2 is the technology that the film video circulation is turned to television video stream of present institute widespread usage, the film video circulation that this technology can be lined by line scan for 24 frame/seconds changes into and meets (the National Television System Committee of National Television System Committee, NTSC) about 60/second interleaved video field forms of requiring of standard television video frequency signal make the video flowing after the conversion can be by the TV normal play.Usually, described the video flowing that is changed into by described film video circulation can be called 3:2Pulldown field sequence.Referring to Fig. 1, the generating principle of 3:2Pulldown field sequence has been shown among Fig. 1, this principle is: the method for utilizing the field to duplicate generates 10 TV signal according to per 4 two field pictures.Particularly, the even number line by moving-picture frame of the idol among Fig. 1 generates, and strange field is then generated by the odd-numbered line of moving-picture frame; Can see, because N field and N+2 field are all idol and from same frame, be the identical duplicate field of a pair of content therefore; In like manner, N+5 field and N+7 field also are a pair of duplicate field.

Along with the development of TV tech, the TV of lining by line scan is popularized just gradually.In order on the TV of lining by line scan, normally to watch TV programme, need to use the interlaced field conversion of signals that TV can be received and be the release of an interleave technology of frame signal of lining by line scan.At present, the release of an interleave pattern that is adopted at general interlaced scan tv video flowing is commonly referred to as the ordinary video pattern; The release of an interleave pattern that is adopted at 3:2Pulldown field sequence is commonly referred to as television movie mode.Particularly, the release of an interleave pattern that is adopted at 3:2Pulldown field sequence normally realizes with such processing method: with the duplicate field in the sequence of 3:2Pulldown field (as (N+2) field and (N+7) field) deletion, and with remaining video field by being combined (as N field and (N+1), (N+3) and (N+4) field); This processing method is commonly referred to as anti-3:2Pulldown, and progressive frame signal original after handling through anti-3:2Pulldown is perfectly recovered, thereby its video effect is well more a lot of than adopting general mode.

In actual applications, because the TV signal that progressive-scan tv is received comprises the mixed video stream of normal interlaced scanning television video stream and 3:2Pulldown field sequence often, if progressive-scan tv can detect the classification of the video flowing of receiving, and carry out different release of an interleaves at normal interlaced scanning television video stream with 3:2Pulldown field sequence and handle, so just can guarantee all to reach best image quality for different video flowing classifications.

The principle of present applied 3:2Pulldown field Sequence Detection is: searching can embody the regular duplicate field (be to have to be duplicate field in per 5, and the position of each duplicate field in 5 remaining unchanged) that occurs of being of 3:2Pulldown field sequence signature.Concrete 3:2Pulldown field Sequence Detection principle as shown in Figure 1, calculate pixel value difference absolute value sum (the Sum of Absolute Difference between per two adjacent same polarity fields (idol field or strange field), SAD), and per 5 sad values that will calculate be defined as one and take turns; Take turns etc. in limited the

wheel

2 or 3, if minimum sad value remains unchanged residing position in 5 sad values that this is taken turns in each wheel, be 3:2Pulldown field sequence with regard to the video flowing of determining to receive so, and then use processing method such as anti-3:2Pulldown and carry out the pairing release of an interleave of television movie mode and handle.

In actual applications, often there are the following problems for present applied 3:2Pulldown field sequence detecting method: owing to adopt fixing or single thresholding causes and can't correctly discern 3:2Pulldown field sequence, this will cause 3:2Pulldown field sequence can not obtain the pairing release of an interleave processing of television movie mode probably, the shown picture quality that goes out is on the low side after making video flowing by release of an interleave, and then has reduced user satisfaction.

Summary of the invention

In view of this, main purpose of the present invention is to provide a kind of field sequence detecting method and device that is used for digital video de-interlacing, and the picture quality when flowing to improve displaying video improves user satisfaction.

Another object of the present invention is to provide a kind of method and apparatus of realizing digital video de-interlacing, the picture quality when flowing to improve displaying video improves user satisfaction.

For achieving the above object, technical scheme of the present invention is achieved in that

The invention discloses a kind of field sequence detecting method that is used for digital video de-interlacing, comprising:

Under the ordinary video pattern, whether there is Foldable field sequence to occur in the checkout area sequence;

Under television movie mode, whether there is ordinary video field sequence to occur in the checkout area sequence;

Wherein, described under the ordinary video pattern, whether there is Foldable field sequence to occur in the checkout area sequence, comprising:

A. collect the sad value be used to carry out a Sequence Detection, and select wherein minimum sad value SADMin, judge the described SADMin that selects at the field sequence whether be moving scene, when judged result when being, enter step B;

B. upgrade the value of reference position PrevPosMin with the residing position PosMin of the sad value of described minimum; According to the residing position PosMin of SADMin in the described sad value of collecting, judge whether to exist to be the regular duplicate field that occurs, when judged result is that definite field sequence that is received is a Foldable field sequence when being; When judged result is returned steps A for not the time;

Described under television movie mode, whether there is ordinary video field sequence to occur in the checkout area sequence, comprising:

C. collect the sad value be used to carry out a Sequence Detection, judge the sad value that wherein is positioned at PrevPosMin at the field sequence whether be moving scene or the motion and the static boundary scene of ordinary video field sequence, when judged result enters step D when being;

D. judge whether repeatedly to occur moving scene or the motion and the static boundary scene of ordinary video field sequence, and be to determine that field sequence that is received is an ordinary video field sequence when being, otherwise definite field sequence that is received not an ordinary video field sequence in judged result.

In the steps A, the method for collecting described sad value is:

Sad value in the calculated field sequence between per two adjacent same polarity fields, order obtain such one and take turns sad value, and the sad value that obtains is stored according to the order of sequence.

In the steps A, judge that whether described sequence is that the method for moving scene is:

Judge that whether described SADMin is less than the first absolute thresholding that sets in advance, and judged result be when being with the described sad value of collecting in sad value except that described SADMin multiply each other with first relative threshold that sets in advance respectively, and judge described SADMin whether than take advantage of each result in any one is all little, if, determine described SADMin at the field sequence be moving scene; Otherwise, determine described SADMin at the field sequence be not moving scene.

The span of described first relative threshold is: between 0 to 0.5.

Among the step B, described determination methods is: judge whether the residing position PosMin of SADMin that respectively takes turns in the sad value that once collected is identical, if define the duplicate field that is regular appearance; Otherwise, determine not to be the regular duplicate field that occurs.

Among the step C, judge whether described sequence is that the moving scene of ordinary video field sequence or the method for motion and static boundary scene are:

Judge and whether be positioned at the sad value of PrevPosMin less than the second absolute thresholding that sets in advance, and judged result be when being with the described sad value of collecting in sad value except that the sad value that is positioned at PrevPosMin multiply each other with second relative threshold respectively, and judge be arranged in PrevPosMin sad value whether greater than take advantage of any one of four results, if, determine to be positioned at PrevPosMin sad value at the field sequence be moving scene or the motion and the static boundary scene of ordinary video field sequence; Otherwise, determine to be positioned at PrevPosMin sad value at the field sequence be not moving scene and the motion and the static boundary scene of ordinary video field sequence.

In the steps A, when judged result be when being further accumulative total collect effective wheel number of effective sad value,

Among the step C, when the result of described judgement reduces described effective wheel number of collecting when being;

Among the step D, described determination methods is:

Judge that whether the described effective wheel number reduced reaches the field EOS thresholding that sets in advance, if reach, then determines repeatedly to occur moving scene or the motion and the static boundary scene of ordinary video field sequence; Otherwise, definite moving scene or motion and static boundary scene that ordinary video field sequence repeatedly do not occur.

This method further comprises:

When the release of an interleave mode switch frequency of statistics does not reach the release of an interleave mode switch thresholding that sets in advance as yet, judge whether video flowing is stable, and when definite video flowing is stablized, improve release of an interleave mode switch frequency;

When the release of an interleave mode switch frequency of statistics has reached described release of an interleave mode switch thresholding, reduce release of an interleave mode switch frequency.

Judge whether stable method is video flowing:

When the pattern of current release of an interleave is the ordinary video pattern, whether judge to collect maximum that the wheel numerical value of sad value reached greater than the maximum count thresholding that sets in advance, if greater than, determine the video flowing instability; Otherwise, determine that video flowing is stable;

And/or, when the pattern of current release of an interleave is the telecine process pattern, whether judge to collect minimum value that the wheel numerical value of sad value reached less than the least count thresholding that sets in advance, if less than, determine the video flowing instability; Otherwise, determine that video flowing is stable.

Described sequence is 3:2Pulldown field sequence.

The invention also discloses a kind of method that realizes digital video de-interlacing, under the ordinary video pattern, this method may further comprise the steps:

A. collect and be used to carry out the sad value of a Sequence Detection and select wherein minimum sad value SADMin, judge the described SADMin that selects at the field sequence whether be moving scene, when judged result when being, enter step b;

Wherein, describedly judge that whether described sequence is that the method for moving scene is:

Judge that whether SADMin is less than the first absolute thresholding that sets in advance, and judged result be when being with the described sad value of collecting in sad value except that SADMin multiply each other with first relative threshold that sets in advance respectively, and judge SADMin whether than take advantage of each result in any one is all little, if, determine SADMin at the field sequence be moving scene; Otherwise, determine SADMin at the field sequence be not moving scene;

B. according to the residing position PosMin of SADMin in the described sad value of collecting, judge whether to exist the duplicate field that is regular appearance, when judged result is to determine that field sequence that is received is a Foldable field sequence when being, and the video flowing of receiving is carried out the pairing release of an interleave processing of television movie mode; When judged result is returned step a for not the time.

The invention also discloses a kind of field sequence detecting apparatus that is used for digital video de-interlacing, this device comprises the SAD calculating that links to each other successively and adds up module, SAD judging module, a sequence signature judging module;

Wherein, SAD calculates and the accumulative total module, is used to collect the sad value that carries out a Sequence Detection and selects wherein minimum sad value SADMin at the sad value of collecting, and more selected SADMin and present position PosMin thereof is sent to the SAD judging module;

The SAD judging module, be used to judge the SADMin that receives at the field sequence whether be moving scene, also in judged result notice sequence signature judging module when being; Under television movie mode, judge the sad value institute be positioned at described PrevPosMin at the field sequence whether be moving scene or the motion and the static boundary scene of ordinary video field sequence, notify described field sequence signature judging module when being in judged result;

Field sequence signature judging module is used for determining whether to exist the duplicate field that regularity occurs that is that can embody a sequence signature according to PosMin, and determines that the field sequence that is received is a Foldable field sequence when definite existence is the duplicate field of regular appearance; Be used under television movie mode, according to the judged result of SAD judging module at the sad value of PrevPosMin, judge whether repeatedly to occur moving scene or the motion and the static boundary scene of ordinary video field sequence, and be to determine that the field sequence that is received is an ordinary video field sequence when being in judged result.

Described sequence signature judging module is further used for controlling continuous tupe judging module video flowing sent to telecine release of an interleave resume module.

Described sequence signature judging module is further used for controlling continuous tupe judging module video flowing sent to ordinary video release of an interleave resume module.

This device further comprises the threshold adaptive adjusting module that links to each other with described sequence signature judging module, be used to the release of an interleave mode switch frequency of knowing that a sequence signature judging module is added up, and when release of an interleave mode switch frequency does not reach the release of an interleave mode switch thresholding that sets in advance as yet, judge whether video flowing is stable, when definite video flowing is stablized, improve release of an interleave mode switch frequency again; Also when release of an interleave mode switch frequency has reached described release of an interleave mode switch thresholding, reduce release of an interleave mode switch frequency.

Described video flowing is the received interlacing scan data of data storage, and this data storage is described to link to each other with the tupe judging module.

The invention also discloses a kind of device of realizing digital video de-interlacing, this device comprises the SAD calculating that links to each other successively and adds up module, SAD judging module, a sequence signature judging module;

Described SAD judging module is used for according to the first absolute thresholding and first relative threshold that set in advance, judge the SADMin that receives at the field sequence whether be moving scene, also notify described field sequence signature judging module when being in judged result;

Described sequence signature judging module, be used for determining whether to exist the duplicate field that is regular appearance that can embody a sequence signature according to described PosMin, determine that when determining to have the duplicate field that is regular appearance the field sequence that is received is a Foldable field sequence, and the tupe judging module that control links to each other sends to telecine release of an interleave resume module with video flowing.

Compared with prior art, field sequence detecting method and the device that is used for digital video de-interlacing provided by the present invention, all need to collect and be used to carry out the sad value of a Sequence Detection and select wherein minimum sad value SADMin, the described SADMin that judge to select at the field sequence whether be moving scene, when judged result be when being according to the described sad value of collecting in the residing position PosMin of SADMin, judge whether to exist to be the regular duplicate field that occurs, when judged result is that definite field sequence that is received is a Foldable field sequence when being.The method and apparatus of realization digital video de-interlacing provided by the present invention all need carry out the pairing release of an interleave of television movie mode to the video flowing of receiving and handle when determine that the field sequence that is received is Foldable field sequence.

As seen, the inventive method and device all can determine whether to be necessary to carry out at the video flowing of receiving the conversion of release of an interleave pattern in time, exactly, thereby can obviously improve the displaying video picture quality in when stream, and improve user satisfaction.Have, the inventive method and device can also further effectively be controlled the frequency of release of an interleave mode switch again, the picture quality when this also can improve displaying video stream, and improve user satisfaction.

Description of drawings

Fig. 1 is that the 3:2Pulldown field sequence of prior art generates and detect principle schematic;

Fig. 2 is the 3:2Pulldown field Sequence Detection flow chart of a preferred embodiment of the present invention;

Fig. 3 is the flow chart of switching frequency between two kinds of release of an interleave patterns of control telecine, ordinary video of a preferred embodiment of the present invention;

Fig. 4 is the 3:2Pulldown field sequence detecting apparatus and the principle schematic of a preferred embodiment of the present invention.

Embodiment

Below in conjunction with drawings and the specific embodiments to the detailed description of the invention.

Through in-depth analysis, can sum up and draw following two kinds of field sequence essential distinctions under different scenes and corresponding preferable release of an interleave processing method Foldable field sequence and common sequence:

Particularly, investigating at per 5 one group sad value is the important component part of 3:2 Foldable field sequence testing process:

1) at the moving scene in common the sequence, 5 sad values are all very big, even it is very little to have occurred certain sad value once in a while, the position that this sad value occurs also can be very at random; Preferably the video flowing of receiving being carried out the pairing release of an interleave of ordinary video pattern in this case handles;

2) at the static scene in common the sequence, 5 sad values are all very little, need not limit the release of an interleave pattern;

3), have very little and its position of 1 sad value to remain unchanged in 5 sad values, and this sad value is obviously little a lot of than other 4 sad values at the moving scene in the Foldable field sequence; Preferably the video flowing of receiving being carried out the pairing release of an interleave of television movie mode in this case handles;

4) at the static scene in the Foldable field sequence, 5 sad values are all very little, need not limit the release of an interleave pattern;

5) at motion and static boundary scene in common the sequence, 5 sad values vary, and do not have any rule; Preferably the video flowing of receiving being carried out the pairing release of an interleave of ordinary video pattern in this case handles;

6) at motion in the Foldable field sequence and static boundary scene, the sad value between duplicate field generally is minimum, and the size of all the other 4 sad values is then irregular; Preferably the video flowing of receiving being carried out the pairing release of an interleave of television movie mode in this case handles.

By the above as seen and since under static scene, adopt which kind of release of an interleave pattern all can, therefore can guarantee when running into static scene, not change current release of an interleave pattern as far as possible.

Therefore, adopt above-mentioned analysis conclusion, field sequence detecting method and the device that is used for digital video de-interlacing provided by the present invention, all need to collect and be used to carry out the sad value of a Sequence Detection and select wherein minimum sad value SADMin, the described SADMin that judge to select at the field sequence whether be moving scene, when judged result be when being according to the described sad value of collecting in the residing position PosMin of SADMin, judge whether to exist to be the regular duplicate field that occurs, when judged result is that definite field sequence that is received is a Foldable field sequence when being.The method and apparatus of realization digital video de-interlacing provided by the present invention all need carry out the pairing release of an interleave of television movie mode to the video flowing of receiving and handle when determine that the field sequence that is received is Foldable field sequence.

Referring to Fig. 2, Fig. 2 is the 3:2Pulldown field Sequence Detection flow chart of a preferred embodiment of the present invention, and this flow process may further comprise the steps:

Step 201: detect the detector of 3:2Pulldown field sequence at being arranged at being used in the progressive-scan tv, carry out the system initialization operation.

Described system initialization operation generally includes:

Empty SAD counter SADCntr, the value that makes SADCntr is 0; Present anti-3:2PD state I nvPDStatus is set to close;

Empty anti-3:2PD state toggle count device InvPDSwitchCntr, the value that makes InvPDSwitchCntr is 0.

When carrying out the system initialization operation, also current release of an interleave pattern is defaulted as the ordinary video pattern usually, and initial setting up is used to write down the residing position of effective minimum SAD: reference position PrevPosMin.

Step 202: field sequence detector receives new one data send to progressive-scan tv, calculates the sad value between this and its previous same polarity field, and should value be stored in according to the order of sequence in the SAD buffer; Also SADCntr is added 1.

Particularly, can adopt 3 components when calculating sad value: brightness signal Y component, color signal Cb and Cr component, also can only adopt the brightness signal Y component; And, pixel that participate in to calculate sad value in can being all pixels or all pixels half, 1/4 even pixel still less.

Step 203: judge whether the value that SADCntr adds up equals 5, if enter step 204; Otherwise, return step 202.

Step 204: use prior art and judge the current telecine process pattern that whether is in, if enter step 214; Otherwise, directly enter step 205.

Particularly, owing to when carrying out the system initialization operation, current release of an interleave pattern is defaulted as the ordinary video pattern; When therefore carrying out the decision operation of this step at first, can be that the current telecine process pattern that is not in enters step 205 generally because of judged result.

Step 205 is to step 206: select the sad value SADMin of a minimum and write down its residing position PosMin in 5 sad values in 5 sad values that the SAD buffer is stored, judge again whether SADMin satisfies judgment condition 1, if satisfy, enter step 208; Otherwise, enter step 207.

The principle of described judgment condition 1 is: SADMin is enough little, and SADMin is little more a lot of than other 4 in described 5 sad values.Because this principle, judgment condition 1 can realize by the first absolute thresholding and first relative threshold that set in advance, as: whether judge SADMin less than the first absolute thresholding, if not, determine that judgment condition 1 is satisfied; Otherwise, further multiply each other with first relative threshold respectively with described other 4 sad values in 5 sad values, and judge SADMin whether greater than take advantage of four results in any one, if determine that judgment condition 1 does not satisfy; Otherwise, determine that judgment condition 1 satisfies.

The span of described first relative threshold is usually located between 0 to 0.5, its role is to duplicate field and non-duplicate field are effectively distinguished; Certainly, first relative threshold can have different spans, as long as can determine whether SADMin is all little more a lot of than other 4 in described 5 sad values by first relative threshold.

Step 207: the InvPDSwitchCntr that resets and set in advance, the count value that makes InvPDSwitchCntr is 0, directly enters step 222 afterwards.

Step 208: whether the count value of judging InvPDSwitchCntr is greater than 0, if enter step 210; Otherwise, enter step 209.

Step 209: determine that first effective SADMin is found, and upgrade PrevPosMin, InvPDSwitchCntr=1 is set, directly enter step 222 afterwards with its position.

Step 210: whether the PosMin that is write down in the determining step 205 is identical with the current PrevPosMin that stores, if identical, enters step 211; Otherwise, enter step 209.

Step 211: the count value of InvPDSwitchCntr is added 1.

Step 212: whether the count value of judging InvPDSwitchCntr has reached the initial thresholding CntrThrOn of 3:2Pulldown field sequence that sets in advance, if reach, enters step 213; Otherwise, directly enter step 222.

Particularly, can during more than or equal to CntrThrOn, determine that the count value of InvPDSwitchCntr has reached CntrThrOn in the count value of InvPDSwitchCntr.

Step 213: determine to have detected 3:2Pulldown field sequence, further can start anti-3:2pulldown function, the video flowing of receiving is carried out the pairing release of an interleave of television movie mode handle so that use processing method such as the pairing anti-3:2pulldown of anti-3:2pulldown function at the video flowing of receiving; Present anti-3:2PD state I nvPDStatus is set to open; Afterwards, directly enter step 222.

Step 214 is to step 215: select to be positioned at the sad value of PrevPosMin in 5 sad values that the SAD buffer is stored, and judge whether selected sad value satisfies judgment condition 2, if satisfy, enter step 216; Otherwise, enter step 219.

Have, the principle of described judgment condition 2 is again: the sad value that is positioned at PrevPosMin is enough little, and this sad value is big too many unlike other 4 in described 5 sad values.Because this principle, judgment condition 2 can realize by the second absolute thresholding and second relative threshold that set in advance, as: whether judge the sad value that is positioned at PrevPosMin less than the second absolute thresholding, if not, determine that judgment condition 2 is satisfied; Otherwise, further multiply each other with second relative threshold respectively with described other 4 sad values in 5 sad values, and judge be arranged in PrevPosMin sad value whether greater than take advantage of any one of four results, if determine that judgment condition 2 does not satisfy; Otherwise, determine that judgment condition 2 satisfies.

The span of described second relative threshold is usually located between 1 to 2, its role is to judge whether duplicate field also exists; Certainly, second relative threshold can have different spans, as long as whether can determine to be arranged in the sad value of PrevPosMin by second relative threshold big more too many than other 4 of described 5 sad values.

Step 216: the count value of InvPDSwitchCntr is added 1.

Step 217: whether the count value of judging InvPDSwitchCntr has surpassed CntrThrOn, if surpass, enters step 218; Otherwise, directly enter step 222.

Particularly, can during greater than CntrThrOn, determine that the count value of InvPDSwitchCntr has surpassed CntrThrOn in the count value of InvPDSwitchCntr.

Step 218: the value of CntrThrOn is composed to InvPDSwitchCntr, the count value of InvPDSwitchCntr is bound in CntrThrOn.

Step 219: the count value of InvPDSwitchCntr is subtracted 1.

Step 220: whether the count value of judging InvPDSwitchCntr has reached the 3:2Pulldown field EOS thresholding CntrThrOff that sets in advance, if reach, enters step 221; Otherwise, directly enter step 222.In fact, when the count value of InvPDSwitchCntr has reached CntrThrOff, just mean that also being the regular duplicate field that occurs has not existed.

Particularly, can during smaller or equal to CntrThrOff, determine that the count value of InvPDSwitchCntr has reached CntrThrOff in the count value of InvPDSwitchCntr.

Step 221: determine that 3:2Pulldown field sequence finishes, further can close anti-3:2pulldown function, no longer the video flowing of receiving being carried out the pairing release of an interleave of television movie mode handles, then being carried out the pairing release of an interleave of ordinary video pattern, handles the video flowing of receiving, and present anti-3:2PD state I nvPDStatus is set to close, and InvPDSwitchCntr=0 is set.

Step 222: empty SAD buffer and SADCntr, so that begin to carry out 3:2Pulldown field Sequence Detection next time; Afterwards, return step 202.

Among Fig. 2, the value of the first absolute thresholding and the second absolute thresholding can equate also can be unequal.Have, because the influence of factors such as noise, video editing, the sad value between duplicate field is not to be 0 but less relatively value again, and therefore necessary enough height of the first absolute thresholding and the second absolute thresholding are with the sad value between the duplicate field that covers this non-zero; And the first absolute thresholding and the second absolute thresholding are also low as far as possible, so that the sad value under the sad value between duplicate field and other moving scene is separated.

As shown in Figure 2, step 201 to the key effect of step 203 is to collect the basic element that can carry out 3:2Pulldown field Sequence Detection: 5 sad values.

Step 205 to the key effect of step 212 is: judge whether the SADMin in 5 sad values collecting is enough little and whether all little a lot of than other 4 in described 5 sad values.Both did like this and the sad value under the sad value between duplicate field and other moving scene can be separated, and can guarantee that again the static scene in the video flowing can't make judgment condition 1 satisfy.Obviously, select SADMin simply from 5 sad values after, judgment condition 1 provides technical support for the validity of determining SADMin exactly.

Certainly, step 205 to the key effect of step 212 also is: after the validity of having determined SADMin, seek as far as possible and can embody the regular duplicate field that occurs of being of 3:2Pulldown field sequence signature; The mode that searching is the regular duplicate field that occurs then is: each collected 5 sad values with effective SADMin are defined as effective one take turns, and add up effectively wheel number with InvPDSwitchCntr; In effective wheel of CntrThrOn number, if the SADMin in each wheel remains unchanged residing position in 5 sad values that this is taken turns, then determine the regular duplicate field that occurs to have occurred being, thereby can determine that the video flowing that receives is a 3:2Pulldown field sequence.

Step 214 to the key effect of step 222 is: judge that whether big too many than other 4 of described 5 sad values whether enough little the sad value that is arranged in the PosMin position in 5 sad values that the SAD buffer stored and.Both did like this and the sad value under the sad value between duplicate field and other moving scene can be separated, the moving scene of ordinary video field sequence or motion and static boundary scene are distinguished; Can guarantee that again the static scene in the video flowing can make judgment condition 2 satisfy as far as possible, makes the conversion of release of an interleave pattern seldom occur.Do like this because no matter in the video flowing of ordinary video pattern or in the video flowing of television movie mode, 5 sad values are all very little when static scene, and the difference between each sad value neither be very big; In this case, there is no need to carry out the conversion of release of an interleave pattern.

Step 214 to the key effect of step 222 also is: judge whether the number of times that the sad value be positioned at the PosMin position does not satisfy judgment condition 2 reaches CntrThrOff, and determine not existed in the video flowing sequence peculiar duplicate field that is regular appearance in 3:2Pulldown field when reaching, thereby definite 3:2Pulldown field sequence finishes.

By the above as seen, flow process shown in Figure 2 is utilized absolute thresholding and relative threshold by reasonably making up, can determine the validity of SADMin exactly, and determine based on this whether the video flowing that receives is 3:2Pulldown field sequence, and then can determine whether to be necessary between ordinary video pattern and television movie mode, to carry out the conversion of release of an interleave pattern in time, exactly.Obviously, timely, the picture quality when accurately conversion certainly will be able to significantly improve displaying video stream of release of an interleave pattern, thereby can obviously improve user satisfaction.

In actual applications, if the frequency of release of an interleave mode switch is too high between ordinary video pattern and the television movie mode, also can influence and handle the shown picture quality that goes out behind the video flowing, therefore, can also be further the frequency of release of an interleave mode switch be control effectively.Particularly, can dynamically adjust the frequency of carrying out the release of an interleave mode switch between ordinary video pattern and the television movie mode by the value of adjusting CntrThrOn, the corresponding visible flow process shown in Figure 3 of method of adjustment, this flow process may further comprise the steps:

Step 301: the release of an interleave mode switch number of times ModeSwitchCntr that the statistics certain hour is interior at interval, and judge whether the ModeSwitchCntr that is added up does not reach the release of an interleave mode switch thresholding MODESWCHTHR that sets in advance as yet, if enter step 302; Otherwise, enter step 308.

Step 302: use prior art and judge the current telecine process pattern that whether is in, if enter step 304; Otherwise, enter step 303.

Step 303: whether judge maximum that InvPDSwitchCntr reaches greater than the maximum count thresholding CntrMaxThr that sets in advance, if, process ends; Otherwise, enter step 305.

Step 304: whether judge maximum that InvPDSwitchCntr reaches less than the least count thresholding CntrMinThr that sets in advance, if, process ends; Otherwise, enter step 305.

Step 305: the value of CntrThrOn is subtracted 1.

Step 306: whether the value of judging CntrThrOn is less than the initial thresholding MinCntrThrOn of minimum 3:2Pulldown field sequence that sets in advance, if enter step 307; Otherwise, process ends.

Step 307: the value of MinCntrThrOn is composed to CntrThrOn, the value of CntrThrOn is bound in MinCntrThrOn.

Step 308: the value of CntrThrOn is added 1.

Step 309: whether the value of judging CntrThrOn is greater than the initial thresholding MaxCntrThrOn of maximum 3:2Pulldown field sequence that sets in advance, if enter step 310; Otherwise, process ends.

Step 310: the value of MaxCntrThrOn is composed to CntrThrOn, the value of CntrThrOn is bound in MaxCntrThrOn.

In principle, why carry out step 303, be because when the release of an interleave mode switch be not very frequent and current when not being in the telecine process pattern, the duplicate field that can embody 3:2Pulldown field sequence signature can seldom occur, and makes that the value of InvPDSwitchCntr is smaller all the time.In this case, if the maximum that InvPDSwitchCntr reached is less than CntrMaxThr, illustrate that video flowing is more stable at present, thereby can suitably reduce the value of CntrThrOn, this can make InvPDSwitchCntr relatively easily reach CntrThrOn, so just can make the sensitivity of release of an interleave mode detection recover normal; Otherwise, illustrate that video flowing is not very stable at present, thereby had better not the value of CntrThrOn be changed, but wait until that video flowing changes CntrThrOn when more stable again.

In like manner, why carry out step 304, be because when the release of an interleave mode switch be not very frequent and current when being in the telecine process pattern, the duplicate field that can embody 3:2Pulldown field sequence signature can frequent appearance, makes that the value of InvPDSwitchCntr is bigger all the time.In this case, if the minimum value that InvPDSwitchCntr reached is less than CntrMinThr, illustrate that video flowing is not very stable at present, thereby had better not the value of CntrThrOn be changed, but wait until that video flowing changes CntrThrOn when more stable again; Otherwise, illustrate that video flowing is more stable at present, thereby can suitably reduce the value of CntrThrOn, this can make InvPDSwitchCntr relatively easily reach CntrThrOn, makes the sensitivity of carrying out the release of an interleave mode detection between ordinary video pattern and the television movie mode be improved normal condition.

Having, why carry out step 308, is because under the too frequent situation of release of an interleave mode switch, can suitably increase the value of CntrThrOn; This can make InvPDSwitchCntr be not easy to reach CntrThrOn relatively, thereby can reduce the frequency of carrying out the release of an interleave mode switch between ordinary video pattern and the television movie mode.

In addition, also can before step 302, increase the step of the initial thresholding MinCntrThrOn of minimum Foldable field sequence that judges whether CntrThrOn equals to set in advance, and if judged result for being, process ends then, otherwise enter step 302 again.Why adopting this step, is because when release of an interleave mode switch when not being very frequent, if CntrThrOn is a normal value, think that then current detection sensitivity remains on normal condition, can not adjust the value of CntrThrOn.

Need to prove that in the flow process of execution graph 3, in the time of CntrThrOn can be set be initial value, be it and reach normal value, corresponding current video release of an interleave mode detection sensitivity also is normal condition.

As seen, flow process shown in Figure 3 can dynamically be adjusted the frequency of carrying out the release of an interleave mode switch between ordinary video pattern and the television movie mode rationally, effectively.

In order to guarantee that flow process can be carried out smoothly shown in Fig. 2, Fig. 3, need in progressive-scan tv, carry out setting as shown in Figure 4.Among Fig. 4, the SAD that links to each other successively in the field sequence detector 420 calculates and accumulation module 421, SAD judging module 422 and a sequence signature judging module 423 are the Primary Components that are used for flow process shown in execution graph 2, Fig. 3.

Particularly, data storage 410 is used for the interlaced scanning video stream that buffer memory is received, and the interlaced scanning video stream of buffer memory is sent to SAD calculating and accumulation module 421 and the tupe judging module 430 that self links to each other; SAD calculates and accumulates module 421 according to the video field that is comprised in the video flowings of receiving, collects sad value, determines SADMin and PosMin, and these SAD information are sent to SAD judging module 422; SAD judging module 422 is according to calculating from SAD and the SAD information of accumulation module 421, use judgment condition 1 judge the SADMin institute of receiving at field sequence whether be moving scene, and notify a field sequence signature judging module 423 when being in judged result; Sequence signature judging module 423 searchings in field can embody the regular duplicate field that occurs of being of 3:2Pulldown field sequence signature, and when the duplicate field that is regular appearance because of discovery determines that the video flowing that receives is 3:2Pulldown field sequence, send the current detection sign that is in television movie mode of expression to tupe judging module 430.

When tupe judging module 430 is received the current detection sign that is in television movie mode of expression, will send to telecine release of an interleave module 441 from the video flowing of data storage 410; Processing methods such as the pairing anti-3:2pulldown of the telecine release of an interleave module 441 anti-3:2pulldown functions of application are carried out the pairing release of an interleave processing of television movie mode to the interlaced scanning video stream of receiving, and formed progressive scanned video stream sends to continuous display device demonstration after will finishing the release of an interleave processing.

SAD judging module 422 self set InvPDSwitchCntr that can also when judgment condition 1 does not satisfy, reset; Field sequence signature judging module 423 can also be determined that first effective SADMin is found, and upgrade the original PosMin that stores with its position when judgment condition 1 satisfy but the inconsistent SADMin in position occurred.

Certainly, SAD calculates and accumulates module 421 further can select to be positioned at PrevPosMin in 5 sad values that the SAD buffer is stored sad value, and the sad value of selecting is sent to SAD judging module 422; SAD judging module 422 further can use judgment condition 2 judge the sad value institute that is positioned at PrevPosMin at field sequence whether be moving scene or the motion and the static boundary scene of ordinary video field sequence, and notify a field sequence signature judging module 423 when being in judged result; Field sequence signature judging module 423 further can judge whether repeatedly to occur moving scene or the motion and the static boundary scene of ordinary video field sequence according to the judged result of SAD judging module 422 at the sad value that is positioned at PrevPosMin.

When judged result field sequence signature judging module 423 for not the time can further suitably increase the value of InvPDSwitchCntr; When judged result is that field sequence signature judging module 423 can further send the current detection sign that is in the ordinary video pattern of expression to tupe judging module 430 when being.When tupe judging module 430 is received the current detection sign that is in the ordinary video pattern of expression, will send to ordinary video release of an interleave module 441 from the video flowing of data storage 410; 441 of ordinary video release of an interleave modules are carried out the pairing release of an interleave of ordinary video pattern to the interlaced scanning video stream of receiving and are handled, and will finish release of an interleave and handle the formed progressive scanned video stream in back and send to continuous display device and show.

In actual applications, when a 3:2Pulldown field Sequence Detection finishes, field sequence signature judging module 423 also will notify SAD to calculate and accumulation module 421 empties SAD buffer and SADCntr, so that begin to carry out 3:2Pulldown field Sequence Detection next time.Have again, can also further be provided with in the field sequence detector 420 and a threshold adaptive adjusting module 424 that sequence signature judging module 423 links to each other, be used for knowing the frequency of release of an interleave mode switch from field sequence signature judging module 423, and judge whether that according to the frequency of knowing the value of tackling CntrThrOn increases and decreases, with a judged result notice sequence signature judging module 423, the CntrThrOn that self preserves is increased and decreased processing accordingly again by field sequence signature judging module 423.

In sum, in actual applications, if the value bit wide of pixel is 8 bits, the span of described absolute thresholding is usually located between 4 to 16, if the value bit wide of pixel is other values, then the span of absolute thresholding increases and decreases in proportion: for example pixel is represented that numerical digit is wide and is increased 1, is 9 bit numbers, then the span of absolute thresholding takes advantage of 2, is 8～32; Pixel is represented numerical digit reductions 1, and then the span of absolute thresholding takes advantage of 1/2, is 2～8, and then the rest may be inferred for other; The span of CntrThrOn is usually located between 3 to 24; The span of CntrThrOff is usually also between 3 to 24, but less than the value of CntrThrOn.

When the CntrThrOn adaptive change, this also is minimum value (also the being initial value) MinCntrThrOn of CntrThrOn; The excursion of CntrThrOn be from MinCntrThrOn to maximum MaxCntrThrOn, the scope of MaxCntrThrOn is 24～48, according to the corresponding increasing of the MinCntrThrOn that chooses or reduce;

The statistics time interval of threshold adaptive adjusting module 424 can be made as 4～16 seconds, and value is more little, and reaction speed is fast more, but accuracy rate is low more;

Thresholding MODESWCHTHR has determined that having switched how many times in the timing statistics internal schema just calculates " too frequent ", generally can select 2～8, and statistics time interval is long more usually, and thresholding is high more;

After CntrThrOn improved, the switching times of state had dropped to below the thresholding MODESWCHTHR, thereby just should consider that at this moment when reducing CntrThrOn returns to initial value; For this reason in each statistics time interval, the variation of counter InvPDSwitchCntr is also at the row of supervision: under general mode, the maximum that InvPDSwitchCntr reaches goes on record, if it is less than thresholding CntrMaxThr (can be made as with MinCntrThrOn and equate), then CntrThrOn begins to successively decrease; Similarly, under film mode, the minimum value that InvPDSwitchCntr reaches goes on record, if it is greater than thresholding CntrMinThr (can be made as current C ntrThrOn-2), then CntrThrOn begins to successively decrease.

In actual applications, judge sad value at the field sequence whether be moving scene, and judge sad value at the field sequence whether be that the moving scene of ordinary video field sequence or the method for motion and static boundary scene have multiple.Absolute thresholding and relative threshold that can application of aforementioned be judged, also can use other method and carry out corresponding judgement.

In fact, the method and apparatus that is used for a digital video de-interlacing and a Sequence Detection thereof described above is not only applicable to the 3:2Pulldown technology, also is applicable to other similar Pulldown technology such as 24:1Pulldown technology.In principle, no matter be applicable to which kind of Pulldown technology, in every N field of video field sequence row 1 duplicate field must be arranged, as: the 3:2Pulldown technology is in per 5 video fields 1 duplicate field to be arranged; Difference when being applicable to different Pulldown technology generally only is: each takes turns collected SAD number difference, as: when being applicable to the 3:2Pulldown technology, each takes turns collected SAD number is 5.

By the above as can be seen, field sequence detecting method and the device that is used for digital video de-interlacing provided by the present invention, and the method and apparatus of realizing digital video de-interlacing, the picture quality in the time of all can obviously improving displaying video stream, and improve user satisfaction.

Claims

1, a kind of field sequence detecting method that is used for digital video de-interlacing is characterized in that, comprising:

A. collect the pixel value difference absolute value sum sad value be used to carry out a Sequence Detection, and select wherein minimum sad value SADMin, judge the described SADMin that selects at the field sequence whether be moving scene, when judged result when being, enter step B;

2, the method for claim 1 is characterized in that, in the steps A, the method for collecting described sad value is:

3, the method for claim 1 is characterized in that, in the steps A, judges that whether described sequence is that the method for moving scene is:

4, method as claimed in claim 3 is characterized in that, the span of described first relative threshold is: between 0 to 0.5.

5, as each described method of claim 1 to 3, it is characterized in that among the step B, described determination methods is: judge whether the residing position PosMin of SADMin that respectively takes turns in the sad value that once collected is identical, if define the duplicate field that is regular appearance; Otherwise, determine not to be the regular duplicate field that occurs.

6, the method for claim 1 is characterized in that, among the step C, judges whether described sequence is that the moving scene of ordinary video field sequence or the method for motion and static boundary scene are:

7, as claim 5 or 6 described methods, it is characterized in that, in the steps A, when judged result be when being further accumulative total collect effective wheel number of effective sad value,

Among the step D, described determination methods is:

8, the method for claim 1 is characterized in that, this method further comprises:

When the release of an interleave mode switch frequency of statistics does not reach the release of an interleave mode switch thresholding that sets in advance as yet, judge whether video flowing is stable, and when definite video flowing is stablized, recover the sensitivity of normal release of an interleave mode detection;

9, method as claimed in claim 8 is characterized in that, judges whether stable method is video flowing:

10, the method for claim 1 is characterized in that, described sequence is the folding Pulldown field of 3:2 sequence.

11, a kind of method that realizes digital video de-interlacing is characterized in that, under the ordinary video pattern, this method may further comprise the steps:

A. collect and be used to carry out the pixel value difference absolute value sum sad value of a Sequence Detection and select wherein minimum sad value SADMin, judge the described SADMin that selects at the field sequence whether be moving scene, when judged result when being, enter step b;

B. according to the residing position PosMin of SADMin in the described sad value of collecting, judge whether to exist

12, a kind of field sequence detecting apparatus that is used for digital video de-interlacing is characterized in that, this device comprises the SAD calculating that links to each other successively and adds up module, SAD judging module, a sequence signature judging module;

The SAD judging module, be used to judge the SADMin that receives at the field sequence whether be moving scene, also notify described sequence signature judging module when being in judged result; Under television movie mode, judge the sad value institute be positioned at described PrevPosMin at the field sequence whether be moving scene or the motion and the static boundary scene of ordinary video field sequence, notify described field sequence signature judging module when being in judged result;

Field sequence signature judging module is used for determining whether to exist the duplicate field that regularity occurs that is that can embody a sequence signature according to described PosMin, and determines that the field sequence that is received is a Foldable field sequence when definite existence is the duplicate field of regular appearance; Be used under television movie mode, according to the judged result of SAD judging module at the sad value of PrevPosMin, judge whether repeatedly to occur moving scene or the motion and the static boundary scene of ordinary video field sequence, and be to determine that the field sequence that is received is an ordinary video field sequence when being in judged result.

13, device as claimed in claim 12 is characterized in that, described sequence signature judging module is further used for controlling continuous tupe judging module video flowing is sent to telecine release of an interleave resume module.

14, device as claimed in claim 12 is characterized in that, described sequence signature judging module is further used for controlling continuous tupe judging module video flowing is sent to ordinary video release of an interleave resume module.

15, as each described device of claim 12 to 14, it is characterized in that, this device further comprises the threshold adaptive adjusting module that links to each other with described sequence signature judging module, be used to the release of an interleave mode switch frequency of knowing that a sequence signature judging module is added up, and when release of an interleave mode switch frequency does not reach the release of an interleave mode switch thresholding that sets in advance as yet, judge whether video flowing is stable, when definite video flowing is stablized, improve release of an interleave mode switch frequency again; Also when release of an interleave mode switch frequency has reached described release of an interleave mode switch thresholding, reduce release of an interleave mode switch frequency.

As claim 13 or 14 described devices, it is characterized in that 16, described video flowing is the received interlacing scan data of data storage, this data storage links to each other with described tupe judging module.

17, a kind of device of realizing digital video de-interlacing is characterized in that, this device comprises the SAD calculating that links to each other successively and adds up module, SAD judging module, a sequence signature judging module;

Wherein, described SAD calculates and the accumulative total module, be used to collect the pixel value difference absolute value sum sad value that carries out a Sequence Detection and select wherein minimum sad value SADMin, more selected SADMin and present position PosMin thereof are sent to the SAD judging module at the sad value of collecting;