CN101197997A - De-interlacing method and system based on dynamic threshold value movement and edge self-adaption - Google Patents

Abstract

The invention discloses a dynamic-threshold-based movable and edge adaptive de-interlacing method and a system thereof. The method comprises the following steps of: obtaining four-field data of an image with a third field being a current field and the current point of the current field being a pending insertion point; dynamically obtaining a maximum threshold and a minimum threshold; obtaining differences between the four fields concerning the current point, and upper, lower, left and right points and taking the maximum value; comparing the five differences respectively with periphery pixel similarity degree thresholds, a corresponding point of a difference being a stagnant point if the former is less than the later, and obtaining the number of the stagnant points surrounding the current point; obtaining motion detection information according to the acquired maximum value and number of stagnant points; obtaining the direction with greatest correlativity according to the current field data, interpolating in the direction and obtaining the interpolation result; obtaining the de-interlaced image according to the motion detection information and the interpolation result. The invention effectively protects the edges of an object and prevents useful information from blurring on the condition that the complexity of an integrated circuit is not very high.

Description

Based on the motion of dynamic threshold and the interlace-removing method and the system of edge self-adaption

Technical field

The present invention relates to technical field of integrated circuits, relate in particular to a kind of based on the motion of dynamic threshold and the interlace-removing method and the system of edge self-adaption.

Background technology

Interlace signal is to constitute by two that interlock: first comprises all odd-numbered lines, and second comprises all even number lines.Deinterlacing technique can transfer the TV signal of interlacing to the progressive signal that display can be discerned, and to eliminate the defective of interlaced video sequence, promptly eliminates row structure line, eliminates the flicker between scan line, improves the definition of display frame.The application of deinterlacing technique helps high-quality display liquid crystal, isoionic popularization and application line by line.Fu Za interlace-removing method effect is good more, but realizes but very difficult on integrated circuit.

People have designed a large amount of interlace-removing methods, are 1421098A as publication number, and the Chinese patent that open day is on May 28th, 2003 discloses a kind of motion and edge adaptive deinterlacing method.This method is divided into one section with the plural pixel on the scan line, and each pixel is shared a public motion value in the section, according to whether motion value is determined by different look-up tables in the edge.This method adopts the method for section, though saved the time, when judging the motion at details place, is easy to generate wrong motion detection result.In addition, publication number is 1599447A, and the Chinese patent that open day is on March 23rd, 2005 discloses a kind of directional correlation motion compensation process of Digital Television reprocessing deinterlacing technique.This method comprises that directional correlation filtering, pixel motion with de-noising function are estimated, motion compensation three steps, because adopted estimation, so method is complicated, and when estimation is made mistakes, occurs artifact easily.

Summary of the invention

In view of this; the technical problem to be solved in the present invention provides a kind of based on the motion of dynamic threshold and the interlace-removing method and the system of edge self-adaption; it can protect object edge effectively under the integrated circuit complexity is not very high situation, prevent the fuzzy of useful information.

In order to solve the problems of the technologies described above, the invention provides a kind of based on the motion of dynamic threshold and the interlace-removing method of edge self-adaption.Described method comprises the steps: that a. obtains four field data of image, and described four is first, second, third and fourth, and wherein the 3rd for working as the front court, when the current point of front court for being inserted into a little; B. according to data, dynamically obtain maximum, minimum threshold T as front court and the 4th _Max, T _MinC. the upper and lower, left and right that obtain current point and current point are difference E between 5 four ₁, E ₂, E ₃, E ₄, E ₅, get five difference E ₁, E ₂, E ₃, E ₄, E ₅In maximum be E; D. with described five difference E ₁, E ₂, E ₃, E ₄, E ₅Respectively with surrounding pixel similarity degree threshold value T _dRelatively, if the former is less than the latter, then the point of this difference correspondence is a rest point, and then obtains at described 5 and be the number of rest point; E. the number of maximum E that tries to achieve according to step b and the step c rest point of trying to achieve obtains motion detection information I _mF. according to the direction that obtains the correlation maximum when the data of front court,, obtain interpolation result F then at the enterprising row interpolation of the direction of correlation maximum _IntraG. according to motion detection information I _mAnd the interpolation result F of step b _Intra, obtain to go interlacing image afterwards.

In order to solve the problems of the technologies described above, the present invention also provides a kind of application system for carrying out said process.This system comprises and dynamically obtains threshold module, estimates the motion detection information module, interpose module, output module; Wherein information is dynamically obtained being input as when front court and the 4th field data of threshold module, is output as maximum, minimum threshold T _Max, T _MinThat estimates that computing detects information module is input as four field data and T _Max, T _Min, be output as motion detection information I _mInterpose module be input as data as front court and the 4th, be output as interpolation F as a result _IntraOutput module be input as motion detection information I _m, interpolation F as a result _Intra, and the 4th data, be output as the image after the interlacing.

With respect to prior art; interlace-removing method provided by the invention and system; can effectively discern moving object and marginal portion; make under the situation that hardware is easily realized; the stationary body vertical detail is more, and dynamic object smear and distortion can not occur, under the integrated circuit complexity is not very high situation; protect object edge effectively, prevented the fuzzy of useful information.Dynamically ask for threshold value according to different situations in addition, realized can both be correct in different situations detection motion and standstill object, well gone the interlacing effect.

Description of drawings

To the description of one embodiment of the invention, can further understand purpose, specific structural features and the advantage of its invention by following in conjunction with its accompanying drawing.Wherein, accompanying drawing is:

Fig. 1 is locus, place, an insertion point schematic diagram;

Fig. 2 a and Fig. 2 b are E _rAt [T _m, T _b] between the time, threshold value is asked for schematic diagram;

Fig. 3 is motion and the interlace-removing method of edge self-adaption and the block diagram of system based on dynamic threshold provided by the invention;

Fig. 4 be provided by the invention based on dynamic threshold motion and the flow chart of the interlace-removing method of edge self-adaption.

Embodiment

Specify motion and the interlace-removing method of edge self-adaption and the preferred forms that realizes the system of this method based on dynamic threshold provided by the invention below in conjunction with accompanying drawing.

The invention provides and a kind ofly see also Fig. 3 and Fig. 4 based on the motion of dynamic threshold and the interlace-removing method of edge self-adaption, the method includes the steps of:

Step 1 is dynamically obtained maximum, minimum threshold

The judgment object a variety of modes of whether having moved, choosing that threshold value is simple is wherein a kind of method, but but can not adapt to all motion conditions.Such as: for a lot of slightly images of motion (as at the fine hair of the dog of motion, the Chi Shui that bobbles etc.), if choose very big threshold value, only greater than this threshold value just can think to move may think the fine hair of small size motion by mistake to be static like this.If static, two directly merge, and will make the picture division, cause visual fuzzy.Therefore, for this type games situation, max-thresholds, minimum threshold respectively to choose one.

In addition, motion for common test video pendulum, below OK is that odd field has white line in odd-numbered line, even field has white line in even number line, therefore must just can identify the motion and standstill of this type objects more than 4, the minimum threshold of this type games should be located at more than 10, and max-thresholds is more than 30, but so big threshold value is not suitable for other normal motion conditions.Therefore, at first the peculair motion situation that whether belongs to pendulum was judged before whether judgment object moves that for the ease of narration, this paper is called the peculair motion of pendulum " situation of odd even single line ".

Er＝(a ₃₂+b ₃₂)/2-c ₄₂ (1)

Fig. 1 has described and has been used to differentiate object of which movement and four static fields: first n-2, the 2nd n-1, the 3rd n, the 4th n+1.See also Fig. 1, Er represents upper and lower 2 a of current point ₃₂, b ₃₂Average and next corresponding points c ₄₂Difference.If Er is very big, expression has the horizontal line of single pixel to have different with next corresponding place.If Er is very little, belong to normal condition.

if?Er＞T _b

T _max＝T ₁

T _min＝T ₂

elseif Er＞T _m

T _max＝T _1s+(Er-T _m)*(T _b-T _m)/(T ₂-T _1s) (2)

T _min＝T _1s+(Er-T _m)*(T _b-T _m)/(T ₁-T _1s)

else

T _max＝T _1s

T _min＝T _1s

Here T ₁, T ₂The maximin of setting under the situation for the odd even single line, T ₁Be defaulted as 30, T ₂Be defaulted as 10; T _1sMotor point boundary for setting under the normal condition is defaulted as 4; T _bFor judging whether to belong to the big threshold value of odd even single line situation, the OK as in the pendulum is defaulted as 200; T _mFor judging whether to belong to the middle threshold value of odd even single line situation, be defaulted as 70, mean that black and white contrast is not distinct especially single pixel line, the horizontal line that causes as moving object.

T _Min, T _MaxBeing normalized threshold, is a variable.When belonging to the situation of odd even single line, normalized threshold is selected T ₁, T ₂, greater than T ₁Certainly motion is less than T ₂Certainly do not move.T _1sBe the motor point boundary of setting under the normal condition, relatively more conservative, greater than T _1sThink to belong to motion certainly, less than T _1sBelong to static, motion detection information only selects 0 and 1.[T _m, T _b] between situation be in order to prevent that the uneven situation of transition from appearring in picture at the boundary place, middle T _Min, T _MaxMild transition once.When Er at [T _m, T _b] between the time, threshold value is asked for schematic diagram as shown in Figure 2.

Step 2 obtains motion detection information I _m

Step 2.1 is asked 5 differences between four n-2, n-1, n, n+1, gets maximum

E＝max(E ₁，E ₂，E ₃，E ₄，E ₅) (3)

Wherein, E ₁=| a ₃₂-a ₁₂|, E ₂=| b ₃₂-b ₁₂|, E ₃=| c ₄₂-c ₂₂|, E ₄=| c ₄₁-c ₂₁|, E ₅=| c ₄₃-c ₂₃|, be inserted into a little the locus schematic diagram as shown in Figure 1, the 3rd for work as the front court, being with the central point of grid is current pre-interpolation point, just above-mentioned current point.The maximum (in full with) of bracket interior element is got in max (...) expression, and E is the maximum of difference between 54 of current points, upper and lower, left and right, judges motion whether information with this.

Step 2.2 asks on every side be the number of rest point at 5

Why ask on every side be too little at 5 because the max-thresholds that has is provided with for the number of rest point, require too tight to quiescent conditions, some rest images can be misinterpreted as dynamic image and interpolation, this wrongheaded direct result is exactly that image is discontinuous on time domain, brings visual noise or fine motion.Therefore simple can not accurately judge movable information sometimes with max-thresholds, also needs to use the information of five points on every side.When judging by accident to motion when still image because Rule of judgment is strict, adopt and judge adjacent 5 each values, if wherein there are 3 points very approaching, can be understood as is still image.

At first when whether the calculating every bit moves, all establish following initial value:

still_num＝0

if E ₁＜T _d

still_sum＝still_sum+1

end

if E ₂＜T _d

still_sum＝still_sum+1

end

if E ₃＜T _d

still_sum＝still_sum+1

end

if E ₄＜T _d

still_sum＝still_sum+1

end

if E ₅＜T _d

still_sum＝still_sum+1 (4)

end

Here still_sum is meant at point up and down, left-right dots and next corresponding points thereof of current point 5 points totally, is worth the number (number of rest point) of close point, T _dFor surrounding pixel similarity degree threshold value, be made as 7.Still_sum for around 5 may be the number of static point, it is greater than 2, that is to say as long as around have in 53 static, just think that the central point of these 5 encirclements is static.

Step 2.3 is obtained motion detection information I _m

(1), when motion conditions belongs to the situation of odd even single line, motion detection information I _mFollowing acquisition:

if?Er＞T _b

if?E≥T _max

I _m＝0

elseif?E＞T _min

$I_m=\frac{T_1-E}{T_1-T_2}---\left(5\right)$

else

I _m＝1

end

end

(2), when motion conditions belongs to normal situation, motion detection information I _mFollowing acquisition:

if?Er≤T _m

if?E≥T _max

I _m＝0

if(E＜(T _d+10))

if?still_sum＞2

I _m＝1 (6)

end

end

else

I _m＝1

end

end

Above content shows, because T _MaxWhat be provided with is too little, and some rest image can be misinterpreted as dynamic image, causes slight the moving of rest image on the time domain.If still_sum has the point more than three to be differentiated for static around satisfying, can think that central point also is static, this moment, correction motion detected information I _m=1, promptly static.

(3), when situation belongs to the transition situation of the situation of odd even single line and normal condition, motion detection information I _mFollowing acquisition:

if?T _m＜Er≤T _b

if?E≥T _max

I _m＝0

if(E＜(T _d+10))

if?still_sum＞2

I _m＝1

end

end

elseif?E＞T _min

$I_m=\frac{E-T_{\min }}{T_{\min }-T_{\min }}$

if?(E＜(T _d+10))

if?still_sum＞2

I _m＝1

end

end

else (7)

I _m＝1

end

end

Following formula shows, because inevitably comprise noise between two, sets a little threshold value T _Min, assert as difference less than T _MinThen belong to stagnant zone certainly, as greater than bigger threshold value T _Max, then belong to the moving region certainly, if between then obtains the weights between [0,1].For preventing that some rest image can be misinterpreted as dynamic image, if still_sum has the point more than three to be differentiated for static around satisfying, can think that central point also is static in addition, this moment, correction motion detected information I _m=1, promptly static.

Step 3, image interpolation in

The present invention adopts the medium filtering interpolation algorithm of judging based on the four direction edge; can judge the direction that the degree of correlation is the highest according to picture material; carry out interpolation in that this side up then, obvious this interpolation algorithm can be protected object boundary effectively, suppresses the fuzzy of useful information.

Step 3.1 is got correlation reckling in the four direction

Defining 4 directional dependency is:

D ₁＝(|a ₃₁-a ₃₃|+|b ₃₁-b ₃₃|+|a ₃₁-a ₃₂|+|b ₃₂-b ₃₃|)/2

D ₂＝|a ₃₁-b ₃₁|+|a ₃₃-b ₃₃|

D ₃＝|a ₃₂-b ₃₁|+|a ₃₃-b ₃₂| (8)

D ₄＝|a ₃₁-b ₃₂|+|a ₃₂-b ₃₃|

D＝min(D ₁，D ₂，D ₃，D ₄)

D wherein ₁Be horizontal direction, the vertical line detection of single pixel be horizontal direction, need add in order to prevent mistake | a ₃₁-a ₃₂|+| a ₃₂-a ₃₃| judgement; D ₂Be vertical direction; D ₃Be 63 degree angular direction; D ₄Be 116 degree angular direction.The minimum value (in full together) of bracket interior element is got in min (...) expression, and D is D ₁, D ₂, D ₃, D ₄Middle minimum value.D ₁, D ₂, D ₃, D ₄Numerical values recited represent the make progress correlation of pixel of interpolation and counterparty, it is strong more to be worth more little correlation, carries out interpolation at the direction of correlation maximum.

Step 3.2 is carried out interpolation to the correlation reckling

Carry out interpolation according to direction.The directional dependency maximum, D=D ₂, then:

F _intra＝(max(min(a ₃₁，a ₃₂，b ₃₁)，min(b ₃₁，b ₃₂，a ₃₃)，min(a ₃₂，b ₃₂))+

(9)

min(max(a ₃₁，a ₃₂，b ₃₁)，max(b ₃₁，b ₃₂，a ₃₃)，max(a ₃₂，b ₃₂)))/2

D ₃The directional dependency maximum, D=D ₃, then:

F _intra＝(max(min(a ₃₂，b ₃₁)，min(a ₃₃，b ₃₂)，min(a ₃₂，b ₃₂))+

(10)

min(max(a ₃₂，b ₃₁)，max(a ₃₃，b ₃₂)，max(a ₃₂，b ₃₂)))/2

D ₄The directional dependency maximum, D=D ₄, then

F _intra＝(max(min(a ₃₁，b ₃₂)，min(a ₃₂，b ₃₃)，min(a ₃₂，b ₃₂))+

(11)

min(max(a ₃₁，b ₃₂)，max(a ₃₂，b ₃₃)，max(a ₃₂，b ₃₂)))/2

If D ₁The directional dependency maximum, D=D ₁, then:

F _intra＝med(c ₄₂，a ₃₁，a ₃₂?a ₃₃，b ₃₁，b ₃₂，b ₃₃) (12)

F _IntraBe that the n interpolation field is mended the value that obtains.There is not corresponding information because of this on the horizontal direction, so can adopt the corresponding points c of next ₄₂Information.Adopt medium filtering, promptly by take off corresponding points and this up and down the intermediate value of totally six points obtain.

Step 4: interlaced picture is removed in output

After motion detection information judges, obtain the image of intrafield interpolation after, calculate the image obtain after the interlacing with following formula:

$F_0(x,y,n)=\left\{\begin{array}{ccc}F(x,y,n)& \mathrm{if}& \mathrm{<figure-callout\; id="2"\; label="x\; mod"\; filenames="S2007101731540E00011.png,S2007101731540E00031.png"\; state="\{\{state\}\}">x</figure-callout>}\mathrm{mod}2=n\mathrm{mod}2\\ (1-I_m)*F_{\mathrm{intra}}+I_m*F(x,y,n+1)& & \mathrm{else}\end{array}\right.---\left(13\right)$

Here F (x, y, n), F ₀(x, y, the n) x in, y represents the row and column of insertion point; (n) n in represents to work as front court, F to F for x, y ₀(n) n in represents the present frame after interpolation is carried out in the front court for x, y; F (x, y, the n) value before the current field interpolation of expression, (x, y n+1) represent the value that next (the 4th) interpolation is preceding, F to F ₀(x, y, n) the final output of expression remove image after the interlacing, be the information of a frame.Mod represents to get surplus operation.Following formula shows, if I _m=1 o'clock, static, expression was when front court and next two simple weave in.An interpolation between this formula is represented stagnant zone adopted is treated the pixel value of interpolated point with the pixel value conduct of opposite field same position, to improve its vertical definition; Work as I _m＜1 o'clock, adopt interpolation field to mend to the moving region, to avoid the adopting caused motion blur of interpolation between the field, reach best effect.

The present invention also provides the system that uses above-mentioned method for designing, and the functional-block diagram of this system handles as shown in Figure 3.

This system comprises four modules: dynamically obtain threshold module, estimate the motion detection information module, interpose module, output module.Dynamically obtain threshold module be input as when front court (the 3rd) and after (the 4th a) information, be output as T _Min, T _MaxWhat the maximum estimated computing detected information module is input as four field information and T _Min, T _Max, be output as motion detection information I _mInterpose module be input as when the front court and after one information, be output as interpolation F as a result _Intra(n), this also is a field information for x, y; Output module be input as motion detection information I _m, interpolation is F as a result _Intra(x, y n), and back one information, are output as the image F after the interlacing ₀(n), this is a frame information for x, y.

" if " expression that relates to herein in addition, " if "; Else represent otherwise; End represents to finish.

Under the integrated circuit complexity is not very high situation, the invention provides a kind of motion and the interlace-removing method of edge self-adaption and system that uses this method for designing based on dynamic threshold, can protect object edge effectively, prevent the fuzzy of useful information; Simultaneously can dynamically ask for threshold value according to different situations, different situations can both be correct detection motion and standstill object, well gone the interlacing effect.

Claims (9)

1. one kind based on the motion of dynamic threshold and the interlace-removing method of edge self-adaption, it is characterized in that described method comprises the steps:

A. obtain four field data of image, described four is first, second, third and fourth, and wherein the 3rd for working as the front court, when the current point of front court for being inserted into a little;

B. according to data, dynamically obtain maximum, minimum threshold T as front court and the 4th _Max, T _Min

C. the upper and lower, left and right that obtain current point and current point are difference E between 5 four ₁, E ₂, E ₃, E ₄, E ₅, get five difference E ₁, E ₂, E ₃, E ₄, E ₅In maximum be E;

D. with described five difference E ₁, E ₂, E ₃, E ₄, E ₅Respectively with surrounding pixel similarity degree threshold value T _dRelatively, if the former is less than the latter, then the point of this difference correspondence is a rest point, and then obtains the number of rest point;

E. the number of maximum E that tries to achieve according to step b and the step c rest point of trying to achieve obtains motion detection information I _m

F. according to the direction that obtains the correlation maximum when the data of front court,, obtain interpolation result F then at the enterprising row interpolation of the direction of correlation maximum _Intra

G. according to motion detection information I _mAnd the interpolation result F of step b _Intra, obtain to go interlacing image afterwards.

2. as claimed in claim 1ly it is characterized in that, obtain maximum, minimum threshold T among the step b based on the motion of dynamic threshold and the interlace-removing method of edge self-adaption _Max, T _MinThe method that adopts is:

if?Er＞T _b

T _max＝T ₁

T _min＝T ₂

elseif Er＞T _m

T _max＝T _1s+(Er-T _m)*(T _b-T _m)/(T ₂-T _1s)

T _min＝T _1s+(Er-T _m)*(T _b-T _m)/(T ₁-T _1s)

else

T _max＝T _1s

T _min＝T _1s

Wherein Er represents the upper and lower average of current point and the difference of next corresponding points at 2; T ₁, T ₂For object of which movement is the maximin of setting under the situation of odd even single line; T _1sFor object of which movement is the motor point boundary of setting under the normal condition; T _bThe big threshold value that whether belongs to odd even single line situation for the judgment object motion; T _mThe middle threshold value that whether belongs to odd even single line situation for the judgment object motion.

3. as claimed in claim 2ly it is characterized in that based on the motion of dynamic threshold and the interlace-removing method of edge self-adaption, in step e, motion detection information I _mAdopt one of following three kinds of methods to obtain:

One, when object of which movement belongs to the situation of odd even single line, obtains motion detection information I _mMethod be:

if?Er＞T _b

if?E≥T _max

I _m＝0

elseif?E＞T _min

$I_m=\frac{T_1-E}{T_1-T_2}$

else

I _m＝1

end

end

Two, when object of which movement belongs to normal condition, obtain motion detection information I _mMethod be:

if?Er≤T _m

if?E≥T _max

I _m＝0

if(E＜(T _d+10))

if?still_sum＞2

I _m＝1

end

end

else

I _m＝1

end

end

Wherein still_sum represents the number of rest point in the steps d;

Three, when object of which movement belongs between the situation of odd even single line and the transition situation between the normal condition, obtain motion detection information I _mThe employing method is:

if?T _m＜Er≤T _b

if?E≥T _max

I _m＝0

if?(E＜(T _d+10))

if?still_sum＞2

I _m＝1

end

end

elseif?E＞T _min

$I_m=\frac{E-T_{\min }}{T_{\min }-T_{\min }}$

if(E＜(T _d+10))

if?still_sum＞2

I _m＝1

end

end

else

I _m＝1

end

end

4. as claimed in claim 1ly it is characterized in that based on the motion of dynamic threshold and the interlace-removing method of edge self-adaption, in step f by selected four direction D ₁, D ₂, D ₃, D ₄Obtain the direction of correlation maximum.

5. as claimed in claim 4ly it is characterized in that described D based on the motion of dynamic threshold and the interlace-removing method of edge self-adaption ₁Be horizontal direction, D ₂Be vertical direction, D ₃Be 63 degree angular direction, D ₄Be 116 degree angular direction.

6. describedly it is characterized in that as claim 4 or 5 that the method that obtains the direction D of correlation maximum is based on the motion of dynamic threshold and the interlace-removing method of edge self-adaption:

D ₁＝(|a ₃₁-a ₃₃|+|b ₃₁-b ₃₃|+|a ₃₁-a ₃₂|+|b ₃₂-b ₃₃|)/2

D ₂＝|a ₃₁-b ₃₁|+|a ₃₃-b ₃₃|

D ₃＝|a ₃₂-b ₃₁|+|a ₃₃-b ₃₂|

D ₄＝|a ₃₁-b ₃₂|+|a ₃₂-b ₃₃|

D＝min(D ₁，D ₂，D ₃，D ₄)

Wherein, a ₃₁, a ₃₂, a ₃₃, b ₃₁, b ₃₂, b ₃₃Last 3 and following 3 points of representing current point respectively; Min represents to get minimum value, and the direction D of correlation maximum is D ₁, D ₂, D ₃, D ₄Middle minimum value.

7. as claimed in claim 6ly it is characterized in that, in step f, adopt one of following four kinds of methods to obtain interpolation result F based on the motion of dynamic threshold and the interlace-removing method of edge self-adaption _Intra:

One, D=D ₂, obtain interpolation result F _IntraMethod be:

F _intra＝(max(min(a ₃₁，a ₃₂，b ₃₁)，min(b ₃₁，b ₃₂，a ₃₃)，min(a ₃₂，b ₃₂))+

min(max(a ₃₁，a ₃₂，a ₃₁)，max(b ₃₁，b ₃₂，b ₃₃)，max(a ₃₂，a ₃₂)))/2

Two, D=D ₃, obtain interpolation result F _IntraMethod be:

F _intra＝(max(min(a ₃₂，b ₃₁)，min(a ₃₃，b ₃₂)，min(a ₃₂，b ₃₂))+

min(max(a ₃₂，b ₃₁)，max(a ₃₃，b ₃₂)，max(a ₃₂，b ₃₂)))/2

Three, D=D ₄, obtain interpolation result F _IntraMethod be:

F _intra＝(max(min(a ₃₁，b ₃₂)，min(a ₃₂，b ₃₃)，min(a ₃₂，b ₃₂))+

min(max(a ₃₁，b ₃₂)，max(a ₃₂，b ₃₃)，max(a ₃₂，b ₃₂)))/2

Four, D=D ₁, obtain interpolation result F _IntraMethod be:

F _intra＝med(c ₄₂，a ₃₁?a ₃₂?a ₃₃，b ₃₁，b ₃₂，b ₃₃)

C wherein ₄₂Be current o'clock corresponding points at the 4th; Max represents to get maximum; Min represents to get minimum value; Med represents to get intermediate value.

8. as claimed in claim 7ly it is characterized in that, obtain to go the method for the image after the interlacing to be based on the motion of dynamic threshold and the interlace-removing method of edge self-adaption:

$F_0(x,y,n)=\left\{\begin{array}{ccc}F(x,y,n)& \mathrm{if}& x\mathrm{mod}2=n\mathrm{mod}2\\ (1-I_m)*F_{\mathrm{intra}}+I_m*F(x,y,n+1)& & \mathrm{else}\end{array}\right.$

Wherein (n) n in represents to work as front court, F to F for x, y ₀(x, y, n) n in represent current place present frame; F (x, y, the n) value before the current field interpolation of expression, F (x, y, the n+1) value before expression the 4th field interpolation, F ₀(n) frame information of the output image after the interlacing is removed in expression for x, y; Mod represents to get surplus operation.

One kind use as claimed in claim 1 based on dynamic threshold motion and the system of the interlace-removing method of edge self-adaption, it is characterized in that described system comprises and dynamically obtains threshold module, estimates the motion detection information module, interpose module, output module; Wherein information is dynamically obtained being input as when front court and the 4th field data of threshold module, is output as maximum, minimum threshold T _Max, T _MinThat estimates that computing detects information module is input as four field data and T _Max, T _Min, be output as motion detection information I _mInterpose module be input as data as front court and the 4th, be output as interpolation F as a result _IntraOutput module be input as motion detection information I _m, interpolation F as a result _Intra, and the 4th data, be output as the image after the interlacing.

Images