CN104202555B - Interlace-removing method and device - Google Patents
Interlace-removing method and device Download PDFInfo
- Publication number
- CN104202555B CN104202555B CN201410515279.7A CN201410515279A CN104202555B CN 104202555 B CN104202555 B CN 104202555B CN 201410515279 A CN201410515279 A CN 201410515279A CN 104202555 B CN104202555 B CN 104202555B
- Authority
- CN
- China
- Prior art keywords
- pixel
- interpolation
- value
- motion
- field
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Television Systems (AREA)
Abstract
The present invention provides a kind of interlace-removing method and device, this method includes application preceding field pixel corresponding with interpolation pixel with subsequent fields, difference between the corresponding pixel value of next interpolation pixel, and the motion weighted value of a upper interpolating pixel calculates the motion weighted value of interpolation pixel, calculate the inter-field interpolation and field interpolation of interpolation pixel, and apply inter-field interpolation, field interpolation and the motion weighted value of interpolation pixel calculate the original pixel values of interpolation pixel, according to the correlation of the mathematic interpolation current field multiple directions between the pixel value of the pixel in current field in multiple directions, the pixel value of multiple pixels on the direction of correlation maximum and the corresponding pixel of preceding field interpolation pixel is selected to be filtered, calculate compensated pixel value, and according to original pixel values, compensated pixel value and field interpolation calculate interpolation pixel pixel value really.The device is to realize de-interlaced device using the above method.Operand of the present invention is small, requires relatively low to hardware device, and the picture quality exported is good.
Description
Technical field
The present invention relates to video signal processing field, specifically a kind of be converted into going for progressive scan by interlacing scan
Interlacing method and the device for realizing this method.
Background technology
Digital broadcast television technology is needed as a data frame to transmit vision signal, and the data transmitted are frequently included
Multirow data.And digital broadcast television technology includes the technology and interleaved technology of progressive scan, as shown in figure 1, by
The packet that line sweep technique is transmitted contains the data per a line, as shown in Fig. 1 three rows data therein, the first row data
It is to include three pixels a, b, c, the second row includes three pixel x, y, z, the third line includes three pixels d, e, f.Such as
Each field data that digital broadcast television technology is transmitted includes the data per a line, then needs the data volume of transmission excessive,
Very high to bandwidth requirement, therefore, traditional broadcast television technique is typically to save bandwidth using horizontal-interlace technique.
In the data that horizontal-interlace technique is transmitted, the number of each field data only data comprising odd-numbered line or even number line
According to, and the odd even line number of data that two adjacent field datas are transmitted is just complementary.As shown in Figure 2, preceding field F (n-1),
Current field F (n), subsequent fields F (n+1) are the three adjacent field datas transmitted in horizontal-interlace technique respectively.Wherein, current field
F (n) is current field F (n) phases where the field for carrying out deinterlacing, wherein pixel a, b, c and pixel d, e, f are
The data for lacking even number line or odd-numbered line, i.e. actual transmissions between adjacent odd-numbered line or even number line, two row data are not include
Pixel y, x, z.Picture in the row and subsequent fields F (n+1) of pixel k, g, h composition in preceding field F (n-1) shown in Fig. 2
The row of vegetarian refreshments l, i, j composition is institute's even number line or odd-numbered line on the scene, and preceding field F (n-1) and subsequent fields F (n+1) lack
The data of odd-numbered line or even number line.
But, horizontal-interlace technique but sacrifices picture quality, and its image transmitted can not meet spectators to image
The requirement of quality.In addition, major part display device all uses progressive scanning technology at present, therefore interleaved data are changed
For progressively scan data deinterlacing technique be current image date transmission process field important technology.It can be seen that, interlacing scan
The input field received in technology includes odd field and even field, and odd field only includes the odd-numbered line of source figure, and even field only includes the idol of source figure
Several rows, de interlacing seeks to field to be converted to frame, and frame includes all rows of image.Therefore, deinterlacing technique is exactly to need every
The even number of lines evidence lacked or odd row data are inserted between the adjacent rows data of one place transmission.With current in Fig. 2
Exemplified by field F (n), the purpose of deinterlacing technique is exactly for row data and pixel where current field F (n) pixel a, b, c
Pixel interpolating value y, x, z of the even number line lacked or odd-numbered line are inserted between row data where point d, e, f.
Assuming that current field F (n) signal frame is made up of even number line in Fig. 2, i.e. row data and pixel where pixel a, b, c
Row data where point d, e, f are even number line, and current field F (n) lacks odd row data.Deinterlacing technique is exactly to need adjacent
Even number of lines insert corresponding odd row data between, therefore deinterlacing technique is exactly to generate in current field F (n) to lack
Odd-line pixels pixel value.Pixel to be generated is commonly referred to as interpolation pixel, it is assumed that the pixel of interpolation is current
F (n) pixel x, is that current field F (n) passes through three and moved with the pixel x left adjacent pixel y being in a line
The pixel of adaptive deinterlacing technique generation, that is, a upper interpolating pixel, are in the right adjacent of a line with pixel x
Pixel z be next interpolation after the completion of pixel x interpolation pixel, that is, next interpolation pixel.
In the deinterlacing technique applied at present, the effect most preferably interlace-removing method based on motion compensation, but motion
The algorithm of compensation realizes that operand is big, requires high to hardware device, application cost is high.Rather than in movement compensating algorithm, preferably
It is the interlace-removing method based on Motion Adaptive, its step includes motion detection and weight computing, field interpolation are calculated, field interleaves
Value is calculated, output result after the weights mixing field interpolation and inter-field interpolation that are finally drawn with motion detection.For motion detection
It is inclined to as static image, weights bias toward inter-field interpolation, otherwise bias toward field interpolation.Therefore, the accuracy of motion detection
It is required that it is very high, it otherwise can cause the phenomenons such as serious sawtooth, emergence and spot.
Performance is relatively good in method for testing motion and realizes to be simply three detection methods.Three traditional detections are meters
Previous field is calculated with the difference of the corresponding pixel points on the position with interpolation pixel of latter field to judge whether interpolation point is deposited
Substantially moving, thinking that interpolation is pointed out in the presence of aobvious motion, otherwise it is assumed that interpolation pixel is quiet if difference is more than threshold value
Only.
For some motion very fast images, the difference of the pixel of front and rear field interpolation point position often very little,
So traditional three detections be easy to sentence it as by mistake it is static.Therefore, people change to three traditional method for testing motion
Enter, such as notification number be CN101600061B and CN1207906C patent of invention disclose two kinds it is different based on three motions
Although the interlace-removing method of detection method, these methods improve the performance of three motion detections, but its operand is greatly, right
The requirement of hardware is too high, has also raised the cost of implementation of interlace-removing method.
The content of the invention
The main object of the present invention is to provide the interlace-removing method that a kind of operand is small and output image quality is high.
It is a further object of the present invention to provide a kind of low to the hardware requirement and good de-interlacing apparatus of generation picture quality.
In order to realize above-mentioned main purpose, the interlace-removing method that the present invention is provided is included according in preceding field and subsequent fields
Difference between the pixel value of pixel corresponding with interpolation pixel judges interpolation pixel with the presence or absence of obvious motion, also,
Using in difference, preceding field and the subsequent fields between the pixel value of preceding field pixel corresponding with interpolation pixel with subsequent fields
The motion weighted value of difference and a upper interpolating pixel between the pixel value of pixel corresponding with next interpolation pixel is calculated
The motion weighted value of interpolation pixel, is calculated in the inter-field interpolation and field interpolation of interpolation pixel, and application inter-field interpolation, field
The motion weighted value of interpolation and interpolation pixel calculates the original pixel values of interpolation pixel, according to multiple directions in current field
On pixel pixel value between mathematic interpolation current field multiple directions correlation, select correlation maximum direction on
The pixel value of the corresponding pixel of multiple pixel and preceding field interpolation pixels is filtered, and calculates compensated pixel value, and according to first
Beginning pixel value, compensated pixel value and field interpolation calculate interpolation pixel pixel value really.
From such scheme, the first of interpolation pixel is calculated using inter-field interpolation, field interpolation and motion weighted value
Beginning pixel value, the operand of original pixel values is small, and the original pixel values that application motion weighted value is calculated can be according to pixel
Motion conditions calculate the pixel value of each pixel, and the original pixel values calculated are more accurate, and the picture quality of output is high.
Also, after the original pixel values for calculating interpolation pixel, also compensate calculating, and the operand of compensation calculation
It is small, but can effectively treat the motion state of interpolating pixel and compensate, the picture quality of output is higher, it is to avoid motion state
There are the phenomenons such as sawtooth, emergence in image caused by misjudgment.
Further scheme is, is the step of the motion weighted value for calculating interpolation pixel：Using preceding field and subsequent fields
In pixel corresponding with interpolation pixel pixel value between the state of motion value of mathematic interpolation first, using preceding field with it is follow-up
The state of motion value of mathematic interpolation second between the pixel value of pixel corresponding with next interpolation pixel, in the first motion
There are two in the motion weighted value of state value, the second state of motion value and a upper interpolating pixel when being expressed as motion, by interpolation
The motion weighted value of pixel is set to motion value, otherwise, and the motion weighted value of interpolation pixel is set into static value.
As can be seen here, weighed by the motion to the first state of motion value, the second state of motion value and a upper interpolating pixel
It is that can obtain the motion weighted value of interpolation pixel that weight values, which carry out simple logical operation, the motion weighted value of interpolation pixel
Operand it is smaller, to hardware device require it is relatively low.
One optional scheme is, is the step of the motion weighted value for calculating interpolation pixel：Using preceding field and subsequently
The state of motion value of mathematic interpolation first between the pixel value of pixel corresponding with interpolation pixel, using preceding field with after
The state of motion value of mathematic interpolation second in continuous field between the pixel value of pixel corresponding with next interpolation pixel, interpolation picture
The motion weighted value of element is the first state of motion value, the motion weighted value sum of the second state of motion value and a upper interpolating pixel
Half.
As can be seen here, weighed by calculating the first state of motion value, the second state of motion value and the motion of a upper interpolating pixel
Weight values can clearly reflect the moving situation of interpolation pixel, thus calculate the numerical value of obtained motion weighted value preferably
Reflect the motion state of interpolation pixel, the raising to the quality of output image has obvious benefits.
To realize above-mentioned another object, the de-interlacing apparatus that the present invention is provided includes motion detection block, using previous
Treated in difference, preceding field and subsequent fields between the pixel value of field pixel corresponding with interpolation pixel with subsequent fields with next
The motion weighted value of difference and a upper interpolating pixel between the pixel value of the corresponding pixel of interpolating pixel calculates interpolation picture
The motion weighted value of element；Inter-field interpolation computing module is additionally provided with, the inter-field interpolation of interpolation pixel is calculated；Field interpolation calculates mould
Block, calculates the field interpolation of interpolation pixel；And mixing computing module, using inter-field interpolation, field interpolation and interpolation
The motion weighted value of pixel calculates the original pixel values of interpolation pixel, standard of compensation computing module is additionally provided with, according to current field
The correlation of mathematic interpolation current field multiple directions between the pixel value of pixel in middle multiple directions, selects correlation maximum
Direction on multiple pixels and the pixel value of preceding field pixel corresponding with interpolation pixel be filtered, calculate compensation pixel
Value, compensates determination module, and interpolation pixel fixation element really is calculated according to original pixel values, compensated pixel value and field interpolation
Value.
From such scheme, when calculating the pixel value of interpolation pixel, with reference to motion detection, motion weight calculation, mend
The calculating of pixel value is repaid, the interpolation pixel value of calculating can be made, the pixel value of the interpolation pixel calculated is more accurate, and
And operand is small, the requirement to hardware device is low, and the picture quality generated is good.
Brief description of the drawings
Fig. 1 is the pixel map that progressive scanning technology transmits data.
Fig. 2 is the pixel map of preceding field, current field and subsequent fields in interlace-removing method.
Fig. 3 is the block diagram of de-interlacing apparatus embodiment of the present invention.
Fig. 4 is the flow chart of interlace-removing method embodiment of the present invention.
Below in conjunction with drawings and Examples, the invention will be further described.
Embodiment
The interlace-removing method of the present invention is applied in digital broadcast television transmission technology, for interleaved data to be turned
Change the technology of progressive scan into, that is, generate the pixel value of the one-row pixels point lacked in each field.The de interlacing of the present invention
Device be using interlace-removing method realized by interleaved data conversion into the data of progressive scan device.
The data that horizontal-interlace technique is transmitted are as shown in Fig. 2 include multiple adjacent fields, and each field is only included
The data of odd-numbered line or the data of even number line.For example, three adjacent fields be respectively preceding field F (n-1), current field F (n),
Subsequent fields F (n+1), current field F (n) are the fields for carrying out deinterlacing, and pixel a, b, c and pixel d, e, f are to work as
Lack the pixel of odd-numbered line between the pixel of even number line adjacent front court F (n), two row pixels.Pixel y, x, z are exactly
The pixel for the odd-numbered line for need to calculate in interlace-removing method, generating, wherein pixel x is the pixel of interpolation, pixel y
It is the upper interpolating pixel point being computed before pixel x is calculated, pixel z is after next interpolation after pixel x
Pixel.
The row of pixel k, g, h composition in preceding field F (n-1) shown in Fig. 2 is the picture of preceding field F (n-1) odd-numbered line
The row of pixel l, i, j composition in vegetarian refreshments, subsequent fields F (n+1) is the pixel of odd-numbered line in subsequent fields F (n+1), it is seen then that
Preceding field F (n-1) and subsequent fields F (n+1) lack the pixel of even number line.
Referring to Fig. 3, de-interlacing apparatus of the invention has interpolation calculation module 10 and compensation calculation module 20, wherein inserting
Being worth computing module 10 has motion detection block 11, inter-field interpolation computing module 13, field interpolation computing module 14 and mixing
Computing module 15, compensation calculation module 20 has standard of compensation computing module 21 and compensation determination module 22.
Motion detection block 11 is used to calculate and judge interpolation pixel with the presence or absence of obvious motion, and calculates interpolation picture
The motion weighted value of element.Inter-field interpolation computing module 13, field interpolation computing module 14 are respectively used to calculate interpolation pixel
Inter-field interpolation and field interpolation, motion weighted value, inter-field interpolation, the field interpolation of the mixing application interpolation pixel of computing module 15
Calculate the initial pixel of interpolation pixel.
It is higher in order to ensure the picture quality of output, the phenomenons such as sawtooth, emergence, the spot of image are reduced, interpolation is calculated
After the original pixel values of image, in addition it is also necessary to treat interpolating pixel and carry out motion compensation calculations, motion compensation calculations are counted by compensation
Module 20 is calculated to realize.Standard of compensation computing module 21 is used for the reference data for calculating compensation, and compensation determination module 22 is according to compensation
The result of calculation of benchmark module 21 calculates the concrete numerical value of compensation, obtains the final pixel value really of interpolation pixel.
The operation principle of de-interlacing apparatus modules is introduced with reference to Fig. 4, and illustrates the flow of interlace-removing method.
First, step S1 is performed, original view data is received, raw image data is the data of many, and each field is
Odd field or even field, odd field only include the pixel of odd-numbered line, and even field only includes the display of even number line.
Then, motion detection block 11 performs step S2, calculates the motion weight of interpolation pixel.For treating in Fig. 2
Interpolating pixel x, chooses in preceding field F (n-1) corresponding to pixel g corresponding with interpolation pixel x and next interpolation pixel z
Pixel h, and choose in subsequent fields F (n+1) corresponding to pixel i corresponding with interpolation pixel x and next interpolation pixel z
Pixel j, judge interpolation pixel x motion conditions with reference to current field F (n) upper interpolating pixel y.Interpolation pixel
Motion weight can be the decimal between 0 or 1 integer or 0 to 1, and the present invention provides two kinds of different computational methods.
First method is to judge whether interpolation pixel x substantially moves, and the motion weighted value that this method is calculated is only 0
Or 1, motion weighted value is 0, represents that interpolation pixel x is not substantially moved, that is, static value, and such as motion weighted value is 1,
Represent that interpolation pixel x has substantially motion, that is, motion value.Specific computational methods are, by preceding field F (n-1)
Absolute difference between the pixel value of the pixel at subsequent fields F (n+1) same position is compared with a threshold value, if
The absolute difference is less than threshold value, it can be determined that obvious motion is not present in interpolation pixel, if absolute difference is more than threshold value, sentences
There is obvious motion in disconnected interpolation pixel.
For example, choosing the pixel value of pixel i in pixel g pixel value and subsequent fields F (n+1) in preceding field F (n-1)
Between the difference of absolute value be compared with threshold value, and judge between two pixels to be in the absence of motion or there is substantially fortune
It is dynamic, and will determine that value is labeled as the first state of motion value Y1.When judging that motion is not present in two pixels, the first motion shape
State value Y1 is entered as 0, and when judging that two pixels have obvious motion, the first state of motion value Y1 is entered as 1.Assuming that threshold value is
Th, then can obtain Y1=(| fn-1 (g)-fn+1 (i) |>Th), wherein fn-1 (g) is pixel g pixel value, and fn+1 (i) is
Pixel i pixel value.
Similarly, the pixel value of pixel j in pixel h pixel value and subsequent fields F (n+1) in preceding field F (n-1) is chosen
Between absolute value differences be compared with threshold value, judge motion whether is not present between two pixels or there is substantially fortune
It is dynamic, and the second state of motion value is labeled as Y2.Therefore can obtain Y2=(| fn-1 (h)-fn+1 (j) |>Th), wherein fn-1
(h) pixel value for being pixel h, fn+1 (j) is pixel j pixel value.
Also, the motion weighted value of upper interpolating pixel y in current field F (n) is labeled as motiony.If the first fortune
There are two values to be 1 in dynamic state value Y1, the second state of motion value Y2 and a upper interpolating pixel y motion weighted value motiony, then
Judge that current field F (n) interpolation pixel x has obvious motion, otherwise it is assumed that interpolation pixel x be not present motion or
Motion is not obvious.Assuming that by interpolation pixel x motion weighted value be labeled as motionx, can obtain logical formula motionx=
Y1&&motiony||Y1&&Y2||Y2&&motion.Interpolation pixel x motion can be calculated by above-mentioned logical formula
Weighted value.Distinguishingly, when calculating first interpolation pixel, the motion weighted value of a upper interpolating pixel is marked into motiony
It is set to 1.
Second method is that interpolation pixel x motion weighted value motionx is used into fractional representation, its span
Between 0 to 1.Also, two threshold values are set, are Th1 and Th2 respectively, it is preferable that the value of two threshold values is Th1=8 respectively
With Th2=16.
When calculating interpolation pixel x motion weighted value, pixel g pixel value is chosen in preceding field F (n-1) and follow-up
Absolute value differences between F (n+1) pixels i pixel value are compared with threshold value, if two pixel point interpolations is exhausted
Threshold value Th1 is less than to value, i.e., | fn-1 (g)-fn+1 (i) |<Th1, then it is assumed that the first state of motion value motionx1=0.If
The absolute value of two pixel point interpolations is more than threshold value Th1, but less than Th2+Th1, i.e. Th1<|fn-1(g)-fn+1(i)|<Th2+
Th1, then it is assumed that 0<motionx1<1, it can for example use formula motionx1=(| fn-1 (g)-fn+1 (i) |-Th1)/Th2 meters
Calculate the value for obtaining motionx1.If the absolute value of two pixel point interpolations is more than Th2+Th1, i.e., | fn-1 (g)-fn+1 (i) |>
Th2+Th1, then it is assumed that the first state of motion value motionx1=1.
Similarly, the pixel value of pixel h pixel value and subsequent fields F (n+1) pixel j in preceding field F (n-1) are chosen
Between absolute value differences be compared with threshold value, if | fn-1 (h)-fn+1 (j) |<Th1, then it is assumed that the second state of motion value
motionx2=0；If Th1<|fn-1(h) -fn+1(j)|<Th2+Th1, then it is assumed that 0<motionx2<1, it can such as use
Formula motionx2=(| fn-1 (h)-fn+1 (j) |-Th1)/Th2 calculates the value for obtaining motionx2;If | fn-1 (h)-fn+
1(j)|>Th2+Th1, then it is assumed that the second state of motion value motionx2=1.
Finally, with reference to the first state of motion value motionx1, the second state of motion value motionx2 and a upper interpolating pixel
Motion weighted value motiony value calculate interpolation pixel x motion state and motion weighted value, interpolation pixel x fortune
Dynamic weighted value motionx is the first state of motion value motionx1, the second state of motion value motionx2 and upper interpolating pixel y
Motion weighted value motiony sums half, that is, pass through the motion weighted value that below equation calculates interpolation pixel x
motionx：Motionx=(motionx1+motionx2+motiony)/2.If result of calculation motionx>1, then it is assumed that
Motionx=1, if 0< motionx<1, then it is assumed that motionx=motionx.
After the motion weighted value for calculating interpolation pixel x, inter-field interpolation computing module 13 and field interpolation computing module 14
Step S3 is performed, interpolation pixel x inter-field interpolation and field interpolation are calculated respectively.
Calculate inter-field interpolation when, choose preceding field F (n-1) and subsequent fields F (n+1) in interpolation pixel x correspondence positions
Pixel pixel value carry out mean value calculation.For example choose pixel g and subsequent fields F (n+ in preceding field F (n-1)
1) pixel i pixel value, interpolation pixel x inter-field interpolation ft is calculated according to below equation：ft=[fn-1(g) +
fn+1(i)]/2。
When calculating field interpolation, it is averaged using the pixel value in same field with the row respective pixels of above and below interpolation pixel x two
Value obtains field interpolation, for example, choose the pixel value of the pixel b and pixel e in current field F (n), count as follows
Calculate the field interpolation fs for drawing interpolation pixel x：Fs=[fn (b)+fn (e)]/2, wherein fn (b), fn (e) are pixel b respectively
With pixel e pixel value.
Then, mixing computing module 15 performs step S4, calculates interpolation pixel x original pixel values fadp.Calculating is treated
Interpolating pixel x original pixel values fadp is, it is necessary to reference to inter-field interpolation ft, field interpolation fs and motion weighted value Motionx tri-
Individual numerical value is calculated.For example, interpolation pixel x original pixel values fadp can be calculated according to below equation：fadp=fs
×Motionx+(1-Motionx)×ft。
Above-mentioned step S1 is smaller to step S4 operands, therefore relatively low to the hardware requirement of electronic equipment, it is only necessary to count
The data of a line in preceding field F (n-1) and subsequent fields F (n+1) are calculated, so not needing line buffer to cache preceding field F
(n-1) and subsequent fields F (n+1) data.
Interpolation pixel x motion detection result is such as relied solely on to determine interpolation pixel x pixel value, result of calculation
Often may not be very accurate, the pixel of componental movement may be mistaken for static, it is therefore desirable to pass through the compensation of Motion Adaptive
Computing module 20 compensates calculating to step S4 result.
When performing compensation calculation, first, step S5 is performed, standard of compensation computing module 21 calculates current field F (n) multiple sides
To correlation, pixel a, b, c, d, e, the f chosen in the pixel g and current field F (n) in preceding field F (n-1) carry out
Computing, that is, the diagonal for calculating the pixel of two rows above and below current field F (n) represents possible edge direction, by calculating,
Compare the absolute value between the pixel on different directions, judge respective pixel and interpolation pixel x correlation in multiple directions,
The smaller correlation of absolute difference is bigger, and the corresponding direction of minimal difference is exactly edge direction.
Preferably, multiple different directions include diagonal, adjacent inclined direction and vertical direction, such as pixel
In the diagonal directions, pixel b, f are pixel b, x, e in the vertical direction on adjacent inclined direction by a, x, f.Therefore, may be used
Calculate the correlation on different directions respectively with six following formula.
X1=|fn(a)-fn(e)|+|fn(b)-fn(f)|；
X2=|fn(b)-fn(d)|+|fn(c)-fn(e)|;
X3=|fn(b)-fn(e)|×2;
K1=|fn(a)-fn(f)|;
K2= X3/2;
K3=|fn(c)-fn(d)|。
Then, the multiple pixels and the corresponding picture of preceding field interpolation pixel on a direction of selection correlation maximum
The pixel of element carries out medium filtering.Medium filtering is exactly the median for choosing pixel value in multiple pixels.Selection X1 first,
Minimum absolute difference in X2, X3, the medium filtering of 5 points of progress.If a direction in view of correlation maximum is 90
Pixel b and pixel e line directional dependency is maximum in degree, i.e. current field F (n), then 5 points of medium filtering will be moved back
The filtering of pixel b, e, g three-point median is turned to, the result calculated is more excellent.Therefore, the minimum absolute difference in X1, X2, X3 is sentenced
, it is necessary to further do following judgement after value：
If X1 is the least absolute value difference in X1, X2, X3, compare K1 and K2 size, if K1<K2, selects picture
Vegetarian refreshments a, b, e, f, g carry out medium filtering, and otherwise selected pixels point b, e, g carry out medium filtering.
If X2 is the least absolute value difference in X1, X2, X3, compare K2 and K3 size, if K3<K2, selects picture
Vegetarian refreshments b, c, d, f, g carry out medium filtering, and otherwise selected pixels point b, e, g carry out medium filtering.
If X3 is the least absolute value difference in X1, X2, X3, direct selected pixels point b, e, g carry out medium filtering.
It can be seen that, the present invention uses 5 medium filtering sides, and image can be eliminated compared to three-point median filtering method is played
In sawtooth, the picture quality of output is more preferably.Perform after above-mentioned medium filtering, interpolation pixel x benefit can be calculated
Pixel value fstd is repaid, so far, the step that is finished S6.
Then, compensation determination module 22 performs step S7, in interpolation pixel x original pixel values fadp, compensation pixel
A pixel value is chosen between value fstd and field interpolation fs and is used as interpolation pixel x pixel values, that is, carry out really
Compensation judges to calculate.Compensation judges that calculating is that absolute value differences compensation judgment calculating is carried out to above three numerical value, and selection is absolute
It is worth de-interlacing interpolation result of calculation of the less value of interpolation for interpolation pixel x, that is, determines pixel value comp.For example, such as
Really | fstd-fadp |<| fstd-fs |, it is determined that pixel value comp is original pixel values fadp, otherwise, chooses field interpolation fs
It is used as determination pixel value comp.
Preferably, compensation judges that computing module 22 can also replace above-mentioned calculating using medium filtering, i.e., by following
Formula calculate and determine pixel value comp:comp=median(fstd,fs,fadp).Wherein median is median operation.By
Above-mentioned computing can calculate determination pixel value comp, and finally perform step S8, and calculating is obtained into pixel value comp really
Output.
The pixel value for calculating interpolation pixel by above-mentioned method can compensate for the picture of odd-numbered line lacking in each field
Element or even number line pixel, and the present invention interlace-removing method operand it is small, it is low to hardware requirement, thus reduce every
The application cost of row technology.Also, after application compensation calculation, the phenomenons such as sawtooth, the emergence of image can be eliminated, image is improved
Quality.
Certainly, above-described embodiment is only the preferred embodiment of the present invention, can also there is more changes during practical application,
For example when calculating the motion weighted value of interpolation pixel, other formula can be used；Or, calculate the correlation of multiple directions
When, can be using more or a part of directions as calculating benchmark, such change has no effect on the implementation of the present invention.
Finally it is emphasized that the invention is not restricted to above-mentioned embodiment, such as judging the tool whether pixel substantially moves
The changes such as change, calculating original pixel values, change of the specific formula for calculation of compensated pixel value of body computational methods should also be wrapped
Include within the scope of the invention as claimed.
Claims (10)
1. interlace-removing method, it is characterised in that：Including：
Using the difference between the pixel value of preceding field pixel corresponding with interpolation pixel with subsequent fields, preceding field with it is follow-up
The motion weighted value of difference and a upper interpolating pixel between the pixel value of pixel corresponding with next interpolation pixel
Calculate the motion weighted value of the interpolation pixel；
Calculate the inter-field interpolation and field interpolation of the interpolation pixel, and the application inter-field interpolation, the field interpolation with
And the motion weighted value of the interpolation pixel calculates the original pixel values of the interpolation pixel；
According to the correlation of the mathematic interpolation current field multiple directions between the pixel value of the pixel in current field in multiple directions,
Select the pixel value of pixel corresponding with the interpolation pixel in the multiple pixels and preceding field on the direction of correlation maximum
Medium filtering is carried out, compensated pixel value is calculated；
The determination of the interpolation pixel is calculated according to the original pixel values, the compensated pixel value and the field interpolation
Pixel value, chooses a pixel value conduct between the original pixel values, the compensated pixel value and the field interpolation
Interpolation pixel pixel value really.
2. interlace-removing method according to claim 1, it is characterised in that：
The step of motion weighted value for calculating the interpolation pixel is：Using in preceding field and subsequent fields with the interpolation picture
The state of motion value of mathematic interpolation first between the pixel value of the corresponding pixel of element, using in preceding field and subsequent fields with it is described under
The state of motion value of mathematic interpolation second between the pixel value of the corresponding pixel of one interpolation pixel, in first motion state
There are two in the motion weighted value of value, second state of motion value and a upper interpolating pixel when being expressed as moving, by institute
The motion weighted value for stating interpolation pixel is set to motion value, otherwise, and the motion weighted value of the interpolation pixel is set into static
Value.
3. interlace-removing method according to claim 1, it is characterised in that：
The step of motion weighted value for calculating the interpolation pixel is：Using in preceding field and subsequent fields with the interpolation picture
The state of motion value of mathematic interpolation first between the pixel value of the corresponding pixel of element, using in preceding field and subsequent fields with it is described under
The state of motion value of mathematic interpolation second between the pixel value of the corresponding pixel of one interpolation pixel, the fortune of the interpolation pixel
Dynamic weighted value is first state of motion value, the motion weighted value of second state of motion value and a upper interpolating pixel
The half of sum.
4. interlace-removing method according to claim 3, it is characterised in that：
The motion weighted value of first state of motion value, second state of motion value and the interpolation pixel be 0 to 1 it
Between numerical value.
5. the interlace-removing method according to any one of claims 1 to 3, it is characterised in that：
The step of calculating the determination pixel value is to the original pixel values, the compensated pixel value and the field interpolation
Carry out medium filtering.
6. the interlace-removing method according to any one of claims 1 to 3, it is characterised in that：
In the step of calculating the compensated pixel value, multiple directions at least include diagonal, adjacent inclined direction or
Vertical direction.
7. de-interlacing apparatus, it is characterised in that：Including：
Motion detection block, using the difference between the pixel value of preceding field pixel corresponding with interpolation pixel with subsequent fields
Difference and upper interpolation picture between value, the pixel value of preceding field pixel corresponding with next interpolation pixel with subsequent fields
The motion weighted value of element calculates the motion weighted value of the interpolation pixel；
Inter-field interpolation computing module, calculates the inter-field interpolation of the interpolation pixel；
Field interpolation computing module, calculates the field interpolation of the interpolation pixel；
Computing module is mixed, using the motion weighted value of the inter-field interpolation, the field interpolation and the interpolation pixel
Calculate the original pixel values of the interpolation pixel；
Standard of compensation computing module, according to the mathematic interpolation current field between the pixel value of the pixel in current field in multiple directions
The correlation of multiple directions, selects multiple pixels and preceding field on the direction of correlation maximum corresponding with the interpolation pixel
Pixel pixel value carry out medium filtering, calculate compensated pixel value；
Determination module is compensated, is treated according to being calculated the original pixel values, the compensated pixel value and the field interpolation
Interpolating pixel pixel value really, chooses between the original pixel values, the compensated pixel value and the field interpolation
One pixel value is used as interpolation pixel pixel value really.
8. de-interlacing apparatus according to claim 7, it is characterised in that：
The pixel value of motion detection block application preceding field pixel corresponding with the interpolation pixel with subsequent fields it
Between the state of motion value of mathematic interpolation first, using preceding field pixel corresponding with the next interpolation pixel with subsequent fields
Pixel value between the state of motion value of mathematic interpolation second, in first state of motion value, second state of motion value
During with there is two to be expressed as motion in the motion weighted value of a upper interpolating pixel, by the motion weight of the interpolation pixel
Value is set to motion value, otherwise, and the motion weighted value of the interpolation pixel is set into static value.
9. de-interlacing apparatus according to claim 7, it is characterised in that：
The pixel value of motion detection block application preceding field pixel corresponding with the interpolation pixel with subsequent fields it
Between the state of motion value of mathematic interpolation first, using preceding field pixel corresponding with the next interpolation pixel with subsequent fields
Pixel value between the state of motion value of mathematic interpolation second, the motion weighted value of the interpolation pixel is the described first motion
State value, second state of motion value and the half of the motion weighted value sum of a upper interpolating pixel.
10. the de-interlacing apparatus according to any one of claim 7 to 9, it is characterised in that：
The compensation determination module carries out intermediate value filter to the original pixel values, the compensated pixel value and the field interpolation
Ripple.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410515279.7A CN104202555B (en) | 2014-09-29 | 2014-09-29 | Interlace-removing method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410515279.7A CN104202555B (en) | 2014-09-29 | 2014-09-29 | Interlace-removing method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104202555A CN104202555A (en) | 2014-12-10 |
| CN104202555B true CN104202555B (en) | 2017-10-20 |
Family
ID=52087775
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410515279.7A Active CN104202555B (en) | 2014-09-29 | 2014-09-29 | Interlace-removing method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104202555B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105744201B (en) * | 2014-12-31 | 2019-01-18 | 展讯通信（上海）有限公司 | The de-interlaced treating method and apparatus of TV signal |
| CN105898179A (en) * | 2015-12-14 | 2016-08-24 | 乐视云计算有限公司 | Deinterlace method and device for interlaced video |
| CN105704415B (en) * | 2016-02-02 | 2018-10-02 | 福州瑞芯微电子股份有限公司 | A kind of video deinterleaves treating method and apparatus |
| CN113261276B (en) * | 2019-01-09 | 2023-08-22 | 西安诺瓦星云科技股份有限公司 | De-interlacing interpolation method, de-interlacing interpolation device, de-interlacing interpolation system, video processing method and storage medium |
| CN113852830B (en) * | 2021-09-23 | 2023-12-29 | 杭州国芯科技股份有限公司 | Median filtering video de-interlacing method |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1422074A (en) * | 2001-11-30 | 2003-06-04 | 三星电子株式会社 | Interpolating picture element data selection for motion compensation and its method |
| CN201222771Y (en) * | 2008-05-30 | 2009-04-15 | 深圳艾科创新微电子有限公司 | High speed edge self-adapting de-interlaced interpolation device |
| CN101510985A (en) * | 2009-02-27 | 2009-08-19 | 西安交通大学 | Self-adapting de-interleave method for movement compensation accessory movement |
| CN102148953A (en) * | 2011-04-20 | 2011-08-10 | 广东威创视讯科技股份有限公司 | Method and device for detecting three-field motion of de-interlacing processing and de-interlacing system |
| CN102186045A (en) * | 2011-04-20 | 2011-09-14 | 广东威创视讯科技股份有限公司 | Three-field motion detection method and device for deinterlacing processing and deinterlacing system |
| CN102215368A (en) * | 2011-06-02 | 2011-10-12 | 中山大学 | Motion self-adaptive de-interlacing method based on visual characteristics |
| CN102497524A (en) * | 2011-12-05 | 2012-06-13 | 四川虹微技术有限公司 | Edge adaptive de-interlacing interpolation method |
| CN102497523A (en) * | 2011-12-05 | 2012-06-13 | 四川虹微技术有限公司 | Median-filtering de-interlacing method |
| CN103369208A (en) * | 2013-07-15 | 2013-10-23 | 青岛海信信芯科技有限公司 | Self-adaptive de-interlacing method and device |
-
2014
- 2014-09-29 CN CN201410515279.7A patent/CN104202555B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1422074A (en) * | 2001-11-30 | 2003-06-04 | 三星电子株式会社 | Interpolating picture element data selection for motion compensation and its method |
| CN201222771Y (en) * | 2008-05-30 | 2009-04-15 | 深圳艾科创新微电子有限公司 | High speed edge self-adapting de-interlaced interpolation device |
| CN101510985A (en) * | 2009-02-27 | 2009-08-19 | 西安交通大学 | Self-adapting de-interleave method for movement compensation accessory movement |
| CN102148953A (en) * | 2011-04-20 | 2011-08-10 | 广东威创视讯科技股份有限公司 | Method and device for detecting three-field motion of de-interlacing processing and de-interlacing system |
| CN102186045A (en) * | 2011-04-20 | 2011-09-14 | 广东威创视讯科技股份有限公司 | Three-field motion detection method and device for deinterlacing processing and deinterlacing system |
| CN102215368A (en) * | 2011-06-02 | 2011-10-12 | 中山大学 | Motion self-adaptive de-interlacing method based on visual characteristics |
| CN102497524A (en) * | 2011-12-05 | 2012-06-13 | 四川虹微技术有限公司 | Edge adaptive de-interlacing interpolation method |
| CN102497523A (en) * | 2011-12-05 | 2012-06-13 | 四川虹微技术有限公司 | Median-filtering de-interlacing method |
| CN103369208A (en) * | 2013-07-15 | 2013-10-23 | 青岛海信信芯科技有限公司 | Self-adaptive de-interlacing method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104202555A (en) | 2014-12-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104202555B (en) | Interlace-removing method and device | |
| US8749703B2 (en) | Method and system for selecting interpolation as a means of trading off judder against interpolation artifacts | |
| EP0830018A2 (en) | Method and system for motion detection in a video image | |
| CN1783995A (en) | Frame rate conversion device, image display apparatus, and method of converting frame rate | |
| CN101600061A (en) | De-interlaced method of video motion-adaptive and device | |
| CN101663885A (en) | Image processing device and method, and image display device and method | |
| JP5363656B2 (en) | Method and apparatus for calculating video image blur | |
| CN1422074A (en) | Interpolating picture element data selection for motion compensation and its method | |
| CN101207707A (en) | System and method for advancing frame frequency based on motion compensation | |
| EP1761045A3 (en) | Image signal processing apparatus and interlace-to-progressive conversion method | |
| CN102447870A (en) | Detection method for static objects and motion compensation device | |
| JP4774051B2 (en) | Preprocessing device and method before encoding video image sequence | |
| CN101110966A (en) | Image correction circuit, image correction method and image display | |
| JP4374047B2 (en) | Video signal processing apparatus and video signal processing method | |
| JP5616014B2 (en) | Method for generating a distance representing the direction of an edge in a video picture, corresponding device, and use of the method for deinterlacing or format conversion | |
| CN101232570B (en) | Apparatus and method of motion adaptive image processing | |
| CN101232568B (en) | Method for image deinterlacing and related device | |
| CN102497492B (en) | Detection method for subtitle moving in screen | |
| JP2011223086A (en) | Resolution converting device and method, scanning line interpolating device and method, and video display device and method | |
| JPS58205377A (en) | Adaptive type time spatial filter | |
| JP2014187690A (en) | Video signal processing device and method | |
| CN103686043B (en) | The method and apparatus that reciprocal cross mistake is carried out to video | |
| US20100079668A1 (en) | Video de-interlacing | |
| JP4681341B2 (en) | Sequential scan converter | |
| US8244072B2 (en) | System and method for determining color value of pixel |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220125 Address after: Rooms 1306-1309, 13 / F, 19 science Avenue West, Hong Kong Science Park, Shatin, New Territories, China Patentee after: BUILDWIN INTERNATIONAL (ZHUHAI) LTD. Address before: 519015 3rd Floor, Stereo Science and Technology Building, 184 Bailian Road, Jida, Zhuhai City, Guangdong Province Patentee before: BUILDWIN INTERNATIONAL (ZHUHAI) Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220317 Address after: Rooms 1306-1309, 13 / F, 19 science Avenue West, Hong Kong Science Park, Shatin, New Territories, China Patentee after: BUILDWIN INTERNATIONAL (ZHUHAI) LTD. Address before: 519015 3rd Floor, Stereo Science and Technology Building, 184 Bailian Road, Jida, Zhuhai City, Guangdong Province Patentee before: BUILDWIN INTERNATIONAL (ZHUHAI) Ltd. |