CN1256849C - Method and apparatus for 1/4 pixel precision interpolation - Google Patents

Abstract

The present invention discloses a method and a device for 1/4 pixel accuracy interpolation. A method that different kinds of filter interpolation is used for sub-pixel points on different positions is adopted. Interpolation filters with different tapping numbers and different tapping coefficients are used in the interpolation process. Interpolation processing to sun-pixel positions is realized by a filter with less tapping coefficients, the number of pixel points of a whole pixel position taken in each time is reduced, and each time of transmission quantity of a data bus is decelerated to a higher degree, the data transmitting pressure of the data bus is reduced, and the capacity of a temporary storing device of data is reduced. Compared with the original background art, the performance of an interpolating result of the method used by the present invention is enhanced. The size of a residual error for motion compensation is decreased by the excellent interpolating result, code rates are reduced, and the efficiency of an image and video compression coding is enhanced.

Description

The method and apparatus of 1/4 pixel precision interpolation

Technical field

The present invention relates to electric numerical data processing technology field, particularly relate to a kind of method and apparatus of 1/4 pixel precision interpolation.

Background technology

H.261 traditional video encoding standard such as ITU formulate, H.263, H.26L, the MEPG-1 that organizes to set up of the MPEG of standard and ISO H.264, MPEG-2, MPEG-4 etc. are based on hybrid coding, promptly on the Hybrid Coding framework.So-called hybrid encoding frame is to take all factors into consideration prediction, and the coding framework of the method for conversion and entropy coding has following main feature:

1, utilize prediction to remove the redundancy of time-domain;

2, utilize conversion to remove the redundancy of spatial domain;

3, utilize entropy coding to remove statistical redundancy;

Above-mentioned video encoding standard all has intracoded frame, i.e. I frame, and inter-frame encoding frame, i.e. and P frame, I frame and P frame adopt different coding methods.Wherein the cataloged procedure of P frame is as follows: adopt estimation to obtain motion vector, adopt the inter prediction based on motion compensation then, then the residual block that inter prediction is obtained carries out two-dimensional transform, coefficient in transform domain is quantized again, and carries out entropy coding at last.

Because video data spatially has great correlation, adopting estimation to obtain in the process of motion vector, can utilize correlation between pixel to carry out the interpolation of sub-pixel location pixel value, to improve the accuracy of motion vector.And in the process of motion compensation, obtain less residual error data, thereby improve compression efficiency greatly.Sub-pixel interpolation is the key factor that improves coding gain in the motion compensation link.Therefore, sub-pixel interpolation is the important component part in the coding and decoding video.

The technology such as amplification that also usually relate to image in this external image processing techniques, remote sensing image processing for example, in the Medical Image Processing, it is exactly according to great correlation on the image data space that image amplifies, picture signal is carried out up-sampling, and promptly interpolation obtains the pixel value on the sub-pixel location.Interpolation method can be avoided aliasing occurring in the signal up-sampling process preferably, makes image still can keep The better resolution.Therefore, sub-pixel interpolation also is the important component part in the image processing.

Summary of the invention

The method and apparatus that the purpose of this invention is to provide a kind of 1/4 pixel precision interpolation, can utilize the correlation interpolation between the known pixels to go out half-pix and 1/4 pixel value, make that motion vector can be as accurate as 1/4 Pixel-level in the motion compensation, thereby improve precision of prediction, improve code efficiency.

In order to achieve the above object, the technical solution used in the present invention is as follows:

1, a kind of method of 1/4 pixel precision interpolation

1) put residing diverse location substep interpolation according to sub-pix and finish, 5 class sub-pix point positions are as follows:

Class half-pix point: with whole pixel at half-pix point with delegation or same row;

2 class half-pix points: other half-pix points except a class half-pix point;

One class, 1/4 pixel: with whole pixel at 1/4 pixel with delegation or same row;

Two classes, 1/4 pixel: simultaneously with a class half-pix point and 2 class half-pix points with delegation or while and a class half-pix point and 2 class half-pix points 1/4 pixel at same row;

Three classes, 1/4 pixel: 1/4 pixel except a class 1/4 pixel and two classes, 1/4 pixel;

2) put the filter interpolation of residing diverse location according to sub-pix with different tap number and different tap coefficients.

2, a class half-pix is used done 4 nearest in reference image vegetarian refreshments or the same row whole pixels with 4 nearest in the delegation whole pixels and do the reference image vegetarian refreshments and obtain by four tap balanced-filter interpolation with it with it.

3, a class 1/4 pixel is used being done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap dissymetrical filter interpolation, is mirror filter to the filter of a class 1/4 pixel interpolation that is positioned at adjacent class half-pix point left side or top with to the filter of a class 1/4 pixel interpolation that is positioned at an adjacent class half-pix point right side or below wherein.

4, two class half-pixs are used to select with 4 one nearest with it in delegation class half-pixs do 4 one nearest in reference image vegetarian refreshments or same row class half-pixs and select and do the reference image vegetarian refreshments and obtain by four tap balanced-filter interpolation with it; Wherein a class half-pix is used and is done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation.

5, two classes, 1/4 pixel is used to select with its 4 one nearest class half-pixs in the delegation doing 4 one nearest with it in reference image vegetarian refreshments or same row class half-pixs and select and do the reference image vegetarian refreshments and obtained by four tap dissymetrical filter interpolation, is mirror filter to the filter of two classes, the 1/4 pixel interpolation that is positioned at adjacent 2 class half-pix points left side or top with to the filter of two classes, the 1/4 pixel interpolation that is positioned at adjacent two class half-pix point right sides or below wherein; A class half-pix of doing reference is selected to use and is done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation.

6, three classes, 1/4 pixel o'clock is obtained by the symmetrical linear filter interpolation of two taps with being positioned at the adjacent whole pixel of its 45 degree and two class half-pixs; Wherein a class half-pix is used and is done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation; Two class half-pixs are used to select with 4 one nearest with it in delegation class half-pixs and are done 4 one nearest with it in reference image vegetarian refreshments or same row class half-pixs and select and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation.

7, three classes, 1/4 pixel o'clock is obtained by the symmetrical linear filter interpolation of two taps with being positioned at the adjacent whole pixel of its 45 degree and two class half-pixs; Wherein a class half-pix is used and is done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation; Two class half-pixs are used to select with 4 one nearest with it in delegation class half-pixs and are done 4 one nearest with it in reference image vegetarian refreshments or same row class half-pixs and select and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation.

8, three classes, 1/4 pixel is obtained by two taps symmetry linear filter interpolation with being positioned at its two one adjacent class 1/4 pixels; Or use with it and obtain by four tap dissymetrical filter interpolation with delegation's middle distance nearest four one class 1/4 pixels or nearest four one class 1/4 pixels of row middle distance same with it; Or with using four one class 1/4 pixels nearest to obtain with four tap balanced-filter interpolation with it; Wherein a class 1/4 pixel and whole pixel are at 1/4 pixel with delegation or same row, one class, 1/4 pixel is used and is done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap dissymetrical filter interpolation, is mirror filter to the filter of a class 1/4 pixel interpolation that is positioned at adjacent class half-pix point left side or top with to the filter of a class 1/4 pixel interpolation that is positioned at an adjacent class half-pix point right side or below wherein.

9, said filter is four tap balanced-filters, and filter form is: { Co ₁, Co ₂, Co ₂, Co ₁}/[2 * (Co ₁+ Co ₂)], Co _iBe filter tap coefficients, i=1...4 wherein, concrete tap coefficient be 3,19,19, and-3}/32 with 1,3,3,1}/8, the filtering formula is: x=(Co ₁* X ₁+ Co ₂* X ₂+ Co ₂* X ₃+ Co ₁* X ₄)/[(Co ₁+ Co ₂) * 2], X wherein _iBe the reference image vegetarian refreshments, i=1...4, wherein { 3,19,19 ,-3}/32 filters are used for a class half-pix and two class half-pix interpolation; { 1,3,3,1}/8 filters are used for three classes, 1/4 picture element interpolation.

10, said filter is four tap dissymetrical filters, and filter form is: { Co ₁, Co ₂, Co ₃, Co ₄}/(Co ₁+ Co ₂+ Co ₃+ Co ₄), Co _iBe filter tap coefficients, i=1...4 wherein, concrete tap coefficient be 5,59,13 ,-3}/64 or 3,29,7,1}/32, the filtering formula is: x=(Co ₁* X ₁+ Co ₂* X ₂+ Co ₂* X ₃+ Co ₁* X ₄)/[(Co ₁+ Co ₂) * 2], X wherein _iBe reference image vegetarian refreshments, i=1...4; This filter is used for the interpolation of a class 1/4 pixel, the interpolation of two classes, 1/4 pixel and the interpolation of three classes, 1/4 pixel.

Two, a kind of device of 1/4 pixel precision interpolation it comprise:

Be used for reading the input unit of whole pixel pixel value from primitive frame;

Be used for a class half-pix point interpolation processing unit to a class half-pix point interpolation

The apparatus for temporary storage that is used for a temporary class half-pix point interpolation result;

Be used for a class 1/4 pixel interpolation processor to a class 1/4 pixel interpolation;

The apparatus for temporary storage that is used for a temporary class 1/4 pixel interpolation result

Be used for two class half-pix point interpolation processing unit to two class half-pix point interpolations;

The apparatus for temporary storage that is used for temporary two class half-pix point interpolation results;

Be used for two classes, 1/4 pixel interpolation processor to two classes, 1/4 pixel interpolation;

Be used for three classes, 1/4 pixel interpolation processor to three classes, 1/4 pixel interpolation;

Be used to export the output device of whole sub-pix point pixel values;

Described:

Class half-pix point refers to and puts in order pixel at the half-pix point with delegation or same row;

2 class half-pix points refer to other half-pix points except a class half-pix point;

One class, 1/4 pixel refers to and puts in order pixel at 1/4 pixel with delegation or same row;

Two classes, 1/4 pixel refer to simultaneously with a class half-pix point and 2 class half-pix points with delegation or while and a class half-pix point and 2 class half-pix points 1/4 pixel at same row;

Three classes, 1/4 pixel refers to 1/4 pixel except a class 1/4 pixel and two classes, 1/4 pixel;

All the sub-pix point refers to above 5 class sub-pix points;

Whole pixel refers to each the whole pixel around the sub-pix;

Connect with data/address bus between an input unit and the class half-pix interpolation processor, transmit the pixel value of each whole location of pixels; Connect with data/address bus between input unit and three classes, the 1/4 pixel interpolation processor, transmit the pixel value of each whole location of pixels; Connect with data/address bus between input unit and a class 1/4 pixel interpolation processing device, transmit the pixel value of each whole location of pixels; Connect with data/address bus between one class half-pix interpolation processor and the data apparatus for temporary storage, transmit the pixel value of a class half-pix point position; Connect with data/address bus between one class, 1/4 pixel interpolation processing device and the data apparatus for temporary storage, transmit the pixel value of a class 1/4 pixel position; Connect the pixel value of the sub-pixel location that the transmission interpolation obtains between one class half-pix interpolation processor and the output device with data/address bus; Connect the pixel value of the sub-pixel location that the transmission interpolation obtains between one class, 1/4 pixel interpolation processing device and the output device with data/address bus; Connect the pixel value of each class half-pix point position of transmission between data apparatus for temporary storage and the two class half-pix point interpolation processing unit with data/address bus; Connect the pixel value of each class half-pix point position of transmission between data apparatus for temporary storage and two classes, the 1/4 pixel interpolation processor with data/address bus; Connect the pixel value of each class 1/4 pixel position of transmission between data apparatus for temporary storage and three classes, the 1/4 pixel interpolation processor with data/address bus; Connect with data/address bus between two class half-pix point interpolation processing unit and the data apparatus for temporary storage, transmit the pixel value of two class half-pix point positions; Connect the pixel value of the sub-pixel location that the transmission interpolation obtains between two class half-pix point interpolation processing unit and the output device with data/address bus; Connect the pixel value of the sub-pixel location that the transmission interpolation obtains between two classes, 1/4 pixel interpolation processor and the output device with data/address bus; Be connected with data/address bus between data apparatus for temporary storage and three classes, the 1/4 pixel interpolation processor, transmission is used for two class half-pix point pixel values of interpolation; Connect the sub-pixel location pixel value that the transmission interpolation obtains between three classes, 1/4 pixel interpolation processor and the output device with data/address bus.

The present invention compares with background technology, and the useful effect that has is:

The present invention partly provides sub-pixel interpolation method preferably at the motion compensated prediction of coding and decoding video ring.The first, with the filter of less tap coefficient sub-pixel location is realized the interpolation processing, reduced the number of the whole location of pixels pixel that at every turn is taken into, higher degree has slowed down the transmission quantity of each data/address bus, reduce the transmission data pressure of data/address bus, and reduced the capacity of data apparatus for temporary storage.The second, the method interpolation result of the present invention's use promotes to some extent than original background technology on performance.More excellent interpolation result has reduced the size of motion compensated residual, has reduced code check, has improved video compression coding efficient.The sub-pixel interpolation device is realized the interpolation method among the present invention fully in the coding and decoding video among the present invention.

Description of drawings

Fig. 1 is the position distribution schematic diagram of the sub-pix point that indicates of the whole pixel represented of capitalization and lowercase;

Fig. 2 utilizes whole pixel interpolation to obtain indicating with lowercase the distribution schematic diagram of half-pix point;

Fig. 3 utilizes the half-pix point interpolation that obtains to obtain indicating with lowercase the distribution schematic diagram of half-pix point;

Fig. 4 utilizes whole pixel interpolation to obtain indicating with lowercase the distribution schematic diagram of 1/4 pixel;

Fig. 5 utilizes the half-pix point interpolation that obtains to obtain indicating with lowercase the distribution schematic diagram of 1/4 pixel;

Fig. 6 utilizes whole pixel and two class rods to indicate the distribution schematic diagram of 1/4 pixel to what vegetarian refreshments was inserted out with lowercase;

Fig. 7 utilizes a class 1/4 pixel to indicate the distribution schematic diagram of 1/4 pixel in conjunction with what balanced-filter was inserted out with lowercase;

Fig. 8 utilizes a class 1/4 pixel to indicate the distribution schematic diagram of 1/4 pixel in conjunction with what dissymetrical filter was inserted out with lowercase;

Fig. 9 is the schematic diagram of 1/4 precision sub-pixel values interpolating apparatus;

Figure 10 is that four taps symmetry interpolation filter is formed schematic diagram;

Figure 11 is that the asymmetric interpolation filter of four taps is formed schematic diagram.

Embodiment

Earlier the location of pixels in the original image is called whole pixel in the method, for example the point that marks with capitalization A among Fig. 1.The location of pixels that needs interpolation to obtain is called the sub-pix point, for example all the other points that mark with lowercase among Fig. 1.The sub-pix point is divided into two classes, i.e. half-pix point and 1/4 pixel.Half-pix point is divided into two classes again, is called a class half-pix point and 2 class half-pix points.1/4 pixel is divided three classes, and is called a class 1/4 pixel, two classes, 1/4 pixel and three classes, 1/4 pixel.All kinds of sub-pix points are defined as follows respectively:

Class half-pix point: with whole pixel with the half-pix point of delegation or with the half-pix point of whole pixel at same row.The b among Fig. 1 for example ¹And b ²

2 class half-pix points: other half-pix points except a class half-pix point, for example among Fig. 1.The c among the figure one for example.

One class, 1/4 pixel: with whole pixel with 1/4 pixel of delegation or with 1/4 pixel of whole pixel at same row.The d among Fig. 1 for example ¹, d ², d ³, d ⁴

Two classes, 1/4 pixel: simultaneously with a class half-pix point and 2 class half-pix points with 1/4 pixel of delegation or with while and a class half-pix point and 2 class half-pix points 1/4 pixel at same row.The e among Fig. 1 for example ¹, e ², e ³, e ⁴

Three classes, 1/4 pixel: other 1/4 pixels except a class and two classes, 1/4 pixel.The f among Fig. 1 for example ¹, f ², f ³, f ⁴

The present invention is different according to the position classification of half-pix point and 1/4 pixel, uses different filter interpolations to go out the pixel value of this position at different time.Interpolating method is as follows:

1, is labeled as b ¹The pixel value of a class half-pix point obtain by following method:

Four integer pixel points that horizontal direction is nearest utilize 4 tap filter F ₁Carry out filtering, as shown in Figure 2, obtain interpolation result b ¹Wherein filter tap coefficients be 3,19,19 ,-3}.

b ¹＝((-3)×A ₁+19×A ₂+19×A ₃+(-3)×A ₄)/32

2, be labeled as b ²The pixel value of a class half-pix point obtain by following method:

Four integer pixel points that vertical direction is nearest utilize 4 tap filter F ₁Carry out filtering, as shown in Figure 2, obtain interpolation result b ²Wherein filter tap coefficients be 3,19,19 ,-3}.

b ²＝((-3)×A ₅+19×A ₂+19×A ₆+(-3)×A ₇)/32

3, be labeled as d ¹The pixel value of a class 1/4 pixel obtain by following method:

Four integer pixel points that horizontal direction is nearest utilize 4 tap filter F ₂Carry out filtering, as shown in Figure 4, obtain interpolation result d ¹Wherein filter tap coefficients be 3,29,7 ,-1}.

d ¹＝((-3)×A ₁+29×A ₂+7×A ₃+(-1)×A ₄)/32

4, be labeled as d ²The pixel value of a class 1/4 pixel obtain by following method:

Four integer pixel points that horizontal direction is nearest utilize 4 tap filter F ₃Carry out filtering, as shown in Figure 4, obtain interpolation result d ²Wherein filter tap coefficients be 1,7,29 ,-3}.

d ²＝((-1)×A ₁+7×A ₂+29×A ₃+(-3)×A ₄)/32

5, be labeled as d ³The pixel value of a class 1/4 pixel obtain by following method:

Four integer pixel points that vertical direction is nearest utilize 4 tap filter F ₂Carry out filtering, as shown in Figure 4, obtain interpolation result d ³Wherein filter tap coefficients be 3,29,7 ,-1}.

d ³＝((-3)×A ₅+29×A ₂+7×A ₆+(-1)×A ₇)/32

6, be labeled as d ⁴The pixel value of a class 1/4 pixel obtain by following method:

Four integer pixel points that vertical direction is nearest utilize 4 tap filter F ₃Carry out filtering, as shown in Figure 4, obtain interpolation result d ⁴Wherein filter tap coefficients be 1,7,29 ,-3}.

d ⁴＝((-1)×A ₅+7×A ₂+29×A ₆+(-3)×A ₇)/32

7, the pixel value that is labeled as the 2 class half-pix points of c is obtained by following method:

Level or four one nearest class half-pix points of vertical direction are utilized 4 tap filter F ₁Carry out filtering, as shown in Figure 3, obtain interpolation result c.Wherein filter tap coefficients be 3,19,19 ,-3}.

$c=((-3)\&times;{\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^2+19\&times;{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^2+19\&times;{\mathrm{<figure-callout\; id="3"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_3^2+(-3)\&times;{\mathrm{<figure-callout\; id="4"\; label="b"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_4^2)/32$

Perhaps

$c=((-3)\&times;{\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^1+19\&times;{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^1+19\&times;{\mathrm{<figure-callout\; id="3"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_3^1+(-3)\&times;{\mathrm{<figure-callout\; id="4"\; label="b"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_4^1)/32$

8, be labeled as e ¹The pixel value of two classes, 1/4 pixel obtain by following method:

Four the one class half-pix points that horizontal direction is nearest utilize 4 tap filter F ₂Carry out filtering, as shown in Figure 5, obtain interpolation result e ¹Wherein filter tap coefficients be 3,29,7 ,-1}.

$e^1=((-3)\&times;{\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^2+29\&times;{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^2+7\&times;{\mathrm{<figure-callout\; id="3"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_3^2+(-1)\&times;{\mathrm{<figure-callout\; id="4"\; label="b"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_4^2)/32$

9, be labeled as e ²The pixel value of two classes, 1/4 pixel obtain by following method:

Four the one class half-pix points that horizontal direction is nearest utilize 4 tap filter F ₃Carry out filtering, as shown in Figure 4, obtain interpolation result e ²Wherein filter tap coefficients be 1,7,29 ,-3}.

$e^2=((-1)\&times;{\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^2+7\&times;{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^2+29\&times;{\mathrm{<figure-callout\; id="3"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_3^2+(-3)\&times;{\mathrm{<figure-callout\; id="4"\; label="b"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_4^2)/32$

10, be labeled as e ³The pixel value of two classes, 1/4 pixel obtain by following method:

Four the one class half-pix points that vertical direction is nearest utilize 4 tap filter F ₂Carry out filtering, as shown in Figure 4, obtain interpolation result e ³Wherein filter tap coefficients be 3,29,7 ,-1}.

$e^3=((-3)\&times;{\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^1+29\&times;{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^1+7\&times;{\mathrm{<figure-callout\; id="3"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_3^1+(-1)\&times;{\mathrm{<figure-callout\; id="4"\; label="b"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_4^1)/32$

11, be labeled as e ⁴The pixel value of two classes, 1/4 pixel obtain by following method:

Four the one class half-pix points that vertical direction is nearest utilize 4 tap filter F ₃Carry out filtering, as shown in Figure 4, obtain interpolation result e ⁴Wherein filter tap coefficients be 1,7,29 ,-3}.

$e^4=((-1)\&times;{\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^1+7\&times;{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^1+29\&times;{\mathrm{<figure-callout\; id="3"\; label="b"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">b</figure-callout>}}_3^1+(-3)\&times;{\mathrm{<figure-callout\; id="4"\; label="b"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">b</figure-callout>}}_4^1)/32$

12, be labeled as f ₁, f ₂, f ₃Or f ₄The pixel value interpolation method one of three classes, 1/4 pixel: on average obtain by whole pixel and the two class half-pix point linearities adjacent with its 45 degree, as Fig. 6:

f ¹＝(A ₁+c)/2

f ²＝(A ₂+c)/2

f ³＝(A ₃+c)/2

f ⁴＝(A ₄+c)/2

Be labeled as f ₁, f ₂, f ₃Or f ₄The pixel value interpolation method two of three classes, 1/4 pixel: obtain by two taps symmetry linear filter interpolation with being positioned at its two one adjacent class 1/4 pixels, as Fig. 7:

f ¹＝(d ¹+d ³)/2

f ²＝(d ²+d ⁸)/2

f ³＝(d ⁴+d ⁵)/2

f ⁴＝(d ⁶+d ⁷)/2

Be labeled as f ₁, f ₂, f ₃Or f ₄The pixel value interpolation method three of three classes, 1/4 pixel: use with it to obtain by four tap dissymetrical filter interpolation with delegation's middle distance nearest four one class 1/4 pixels or nearest four one class 1/4 pixels of row middle distance same with it, wherein filter tap coefficients is { 1,7,29,-3}, as Fig. 8:

${\mathrm{<figure-callout\; id="1"\; label="f"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">f</figure-callout>}}_1=(-3\&times;{\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_1^3+29\&times;{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_2^3+7\&times;{\mathrm{<figure-callout\; id="3"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_3^3-{\mathrm{<figure-callout\; id="4"\; label="d"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_4^3)/32$

${\mathrm{<figure-callout\; id="2"\; label="f"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">f</figure-callout>}}_2=(-1\&times;{\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_1^3+7\&times;{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_2^3+29\&times;{\mathrm{<figure-callout\; id="3"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_3^3-3\&times;{\mathrm{<figure-callout\; id="4"\; label="d"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_4^3)/32$

${\mathrm{<figure-callout\; id="3"\; label="f"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">f</figure-callout>}}_3=(-3\&times;{\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_1^4+29\&times;{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_2^4+7\&times;{\mathrm{<figure-callout\; id="3"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_3^4-{\mathrm{<figure-callout\; id="4"\; label="d"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_4^4)/32$

${\mathrm{<figure-callout\; id="4"\; label="f"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">f</figure-callout>}}_4=(-1\&times;{\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_1^4+7\&times;{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_2^4+29\&times;{\mathrm{<figure-callout\; id="3"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_3^4-3\&times;{\mathrm{<figure-callout\; id="4"\; label="d"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_4^4)/32$ Or

${\mathrm{<figure-callout\; id="1"\; label="f"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">f</figure-callout>}}_1=(-3\&times;{\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_1^1+29\&times;{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_2^1+7\&times;{\mathrm{<figure-callout\; id="3"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_3^1-{\mathrm{<figure-callout\; id="4"\; label="d"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_4^1)/32$

${\mathrm{<figure-callout\; id="3"\; label="f"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">f</figure-callout>}}_3=(-1\&times;{\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_1^1+7\&times;{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_2^1+29\&times;{\mathrm{<figure-callout\; id="3"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_3^1-3\&times;{\mathrm{<figure-callout\; id="4"\; label="d"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_4^1)/32$

${\mathrm{<figure-callout\; id="2"\; label="f"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">f</figure-callout>}}_2=(-3\&times;{\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_1^2+29\&times;{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_2^2+7\&times;{\mathrm{<figure-callout\; id="3"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_3^2-{\mathrm{<figure-callout\; id="4"\; label="d"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_4^2)/32$

${\mathrm{<figure-callout\; id="4"\; label="f"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">f</figure-callout>}}_4=(-1\&times;{\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="C20031010846800172.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_1^2+7\&times;{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_2^2+29\&times;{\mathrm{<figure-callout\; id="3"\; label="d"\; filenames="C20031010846800182.png,C20031010846800193.png"\; state="\{\{state\}\}">d</figure-callout>}}_3^2-3\&times;{\mathrm{<figure-callout\; id="4"\; label="d"\; filenames="C20031010846800182.png,C20031010846800201.png"\; state="\{\{state\}\}">d</figure-callout>}}_4^2)/32$

Be labeled as f ₁, f ₂, f ₃Or f ₄The pixel value interpolation method four of three classes, 1/4 pixel: use four one class 1/4 pixels nearest to obtain with four tap balanced-filter interpolation with it, as Fig. 7:

f ¹＝(d ⁴+3×d ¹+3×d ³+d ²)/8

f ²＝(d ¹+3×d ²+3×d ⁸+d ⁷)/8

f ³＝(d ³+3×d ⁴+3×d ⁵+d ⁶)/8

f ⁴=(d ⁵+ 3 * d ⁶+ 3 * d ⁷+ d ⁸The interpolation of all sub-pixel location is so far finished step by step in)/8.

Use the digital filter F of 4 kinds of different coefficients and tap number in the interpolation process _x: Co1, and Co2, Co3, Co4}/(Co1+Co2+Co3+Co4).Its tap coefficient and tap number are specific as follows:

F ₁Be four tap filters of tap coefficient symmetry, follow according to its tap coefficient of different purposes to be: 3,19,19 ,-3}/32 or 1,3,3,1}/8.

f ₂Be asymmetric four tap filters of tap coefficient, its tap coefficient is: { m ₁, m ₂, m ₃, m ₄}/(m ₁+ m ₂+ m ₃,+m ₄), concrete coefficient can select 5,59,13 ,-3}/64 or 3,29,7, wherein one group of-1}/32.

F ₃Be asymmetric four tap filters of tap coefficient, its tap coefficient is F ₂Mirror image is: { m ₄, m ₃, m ₂, m ₁}/(m ₁+ m ₂+ m ₃,+m ₄), concrete coefficient is by F ₂The coefficient mirror image after obtain.

F ₄Be two tap linear interpolation filters of tap coefficient symmetry, its tap coefficient is: and 1,3,3,1}/8.

A kind of device that is used for the coding and decoding video sub-pixel interpolation comprises: input unit 210, a class half-pix point interpolation processing unit 220, data apparatus for temporary storage 230,250,270, a class 1/4 pixel interpolation processor 240, two class half-pix point interpolation processing unit 260, two classes, 1/4 pixel interpolation processor 280, three classes, 1/4 pixel interpolation processor 290 and output device 270, as shown in Figure 9:

Input unit 210 is used for taking out the piece of a size for (M+3) * (N+3) from primitive frame, and wherein M is for wanting the number of whole pixel in the every row of interpolation block, and N is for wanting the number of whole pixel in the every row of interpolation block;

One class half-pix point interpolation processing unit 220 utilizes whole pixel to a class half-pix point interpolation.Use filter F1, promptly 3,19,19 ,-3}/32.ALU is wherein realized interpolation process with multiplier, adder and shift unit;

One class, 1/4 pixel interpolation processor 240 utilizes whole pixel to a class 1/4 pixel interpolation.Use filter F2, F3.ALU is wherein realized interpolation process with multiplier, adder and shift unit;

Two class half-pix point interpolation processing unit 260 utilize a class half-pix o'clock to two class half-pix point interpolations.Use filter F1, promptly 3,19,19 ,-3}/32.ALU is wherein realized interpolation process with multiplier, adder and shift unit;

Two classes, 1/4 pixel interpolation processor 280 utilizes a class half-pix o'clock to two classes, 1/4 pixel interpolation.Use filter F2, F3.ALU is wherein realized interpolation process with multiplier, adder and shift unit;

Three classes, 1/4 pixel interpolation processor 290 utilizes whole pixel and 2 class half-pix points or utilizes a class 1/4 pixel to three classes, 1/4 pixel interpolation.Use two tap balanced-filters when utilizing whole phase vegetarian refreshments and two class half-pix point interpolations, use four tap balanced-filter F1 ({ 1 according to the difference of used reference point locations when utilizing a class 1/4 pixel interpolation, 3,3,1}/8), four tap dissymetrical filter F2, F3 or two tap balanced-filters.ALU is wherein realized interpolation process with multiplier, adder and shift unit;

The structure of four tap balanced-filter F1 as shown in figure 10 in the ALU.Co ₁' and Co ₂' be the filter coefficient after the normalization, promptly

Co ₁′＝Co ₁/[(Co ₁+Co ₂)×2]

Co ₂′＝Co ₂/[(Co ₁+Co ₂)×2]

The structure of four tap dissymetrical filter F2, F3 as shown in figure 11 in the ALU.Co ₁', Co ₂', Co ₃' and Co ₄' be the filter coefficient after the normalization, promptly

Co ₁′＝Co ₁/(Co ₁+Co ₂+Co ₃+Co ₄)

Co ₂′＝Co ₂/(Co ₁+Co ₂+Co ₃+Co ₄)

Co ₃′＝Co ₃/(Co ₁+Co ₂+Co ₃+Co ₄)

Co ₄′＝Co ₄/(Co ₁+Co ₂+Co ₃+Co ₄)

Data apparatus for temporary storage 230 between one class half-pix point interpolation processing unit 220 and the two class half-pix interpolation processors 260 is used to store the class half-pix point pixel value that interpolation produces, and these values are used for two class half-pix interpolation process;

Data apparatus for temporary storage 230 between one class half-pix point interpolation processing unit 220 and the two class half-pix interpolation processors 280 is used to store the class half-pix point pixel value that interpolation produces, and these values are used for two classes, 1/4 picture element interpolation process;

Data apparatus for temporary storage 250 between one class, 1/4 pixel interpolation processor 240 and three classes, the 1/4 pixel interpolation processing device 290 is used to store a class 1/4 pixel pixel value that interpolation produces, and these values are used for three classes, 1/4 picture element interpolation process;

Data apparatus for temporary storage 270 between two class half-pix point interpolation processing unit 260 and three classes, the 1/4 pixel interpolation processing device 290 is used to store the two class half-pix point pixel values that interpolation produces, and these values are used for three classes, 1/4 picture element interpolation process;

Output device 300 is used to export the pixel value of all sub-pixel location that arrive of interpolation.

Connect with data/address bus between an input unit 210 and the class half-pix interpolation processor 220, be used to transmit the pixel value of each whole location of pixels; Connect with data/address bus between input unit 210 and three classes, the 1/4 pixel interpolation processor 290, be used to transmit the pixel value of each whole location of pixels; Connect with data/address bus between input unit 210 and a class 1/4 pixel interpolation processing device 240, be used to transmit the pixel value of each whole location of pixels; Connect with data/address bus between one class half-pix interpolation processor 220 and the data apparatus for temporary storage 230, be used to transmit the pixel value of a class half-pix point position; Connect with data/address bus between one class, 1/4 pixel interpolation processing device 240 and the data apparatus for temporary storage 250, be used to transmit the pixel value of a class 1/4 pixel position; Connect with data/address bus between one class half-pix interpolation processor 220 and the output device 300, be used to transmit the pixel value of the sub-pixel location that interpolation obtains; Connect with data/address bus between one class, 1/4 pixel interpolation processing device 240 and the output device 300, be used to transmit the pixel value of the sub-pixel location that interpolation obtains; Connect with data/address bus between data apparatus for temporary storage 230 and the two class half-pix point interpolation processing unit 260, be used to transmit the pixel value of each class half-pix point position; Connect with data/address bus between data apparatus for temporary storage 230 and two classes, the 1/4 pixel interpolation processor 280, be used to transmit the pixel value of each class half-pix point position; Connect with data/address bus between data apparatus for temporary storage 250 and three classes, the 1/4 pixel interpolation processor 290, be used to transmit the pixel value of each class 1/4 pixel position; Connect with data/address bus between two class half-pix point interpolation processing unit 260 and the data apparatus for temporary storage 270, be used to transmit the pixel value of two class half-pix point positions; Connect with data/address bus between two class half-pix point interpolation processing unit 260 and the output device 300, be used to transmit the pixel value of the sub-pixel location that interpolation obtains; Connect with data/address bus between two classes, 1/4 pixel interpolation processor 280 and the output device 300, be used to transmit the pixel value of the sub-pixel location that interpolation obtains; Be connected with data/address bus between data apparatus for temporary storage 270 and three classes, the 1/4 pixel interpolation processor 290, transmission is used for two class half-pix point pixel values of interpolation; Connect with data/address bus between three classes, 1/4 pixel interpolation processor 290 and the output device 300, be used to transmit the sub-pixel location pixel value that interpolation obtains.

The sub-pixel interpolation device can be used processor system in the above-mentioned coding and decoding video, microcontroller, and programmable logic device or microprocessor are realized part or all of operation.Above-mentioned certain operations can realize that other operations simultaneously can realize with hardware with software.

For convenience's sake, these operations are described to the functional unit of different interconnection or different software modules.But this is dispensable.In some applications, these functional units or module can be integrated in single logical device, program or the operation, and do not have obvious limit.In any situation, the feature of functional unit and software module or description can independently realize, or reinstate hardware or software realization with other operations one.

Claims (11)

1, a kind of method of 1/4 pixel precision interpolation is characterized in that:

1) put residing diverse location substep interpolation according to sub-pix and finish, 5 class sub-pix point positions are as follows:

Class half-pix point: with whole pixel at half-pix point with delegation or same row;

2 class half-pix points: other half-pix points except a class half-pix point;

One class, 1/4 pixel: with whole pixel at 1/4 pixel with delegation or same row;

Two classes, 1/4 pixel: simultaneously with a class half-pix point and 2 class half-pix points with delegation or while and a class half-pix point and 2 class half-pix points 1/4 pixel at same row;

Three classes, 1/4 pixel: 1/4 pixel except a class 1/4 pixel and two classes, 1/4 pixel;

2) put the filter interpolation of residing diverse location according to sub-pix with different tap number and different tap coefficients.

2, the method for a kind of 1/4 pixel precision interpolation according to claim 1 is characterized in that: a class half-pix is used done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation.

3, the method for a kind of 1/4 pixel precision interpolation according to claim 1, it is characterized in that: a class 1/4 pixel being used done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap dissymetrical filter interpolation, is mirror filter to the filter of a class 1/4 pixel interpolation that is positioned at adjacent class half-pix point left side or top with to the filter of a class 1/4 pixel interpolation that is positioned at an adjacent class half-pix point right side or below wherein.

4, the method for a kind of 1/4 pixel precision interpolation according to claim 1 is characterized in that: two class half-pixs are used to select with 4 one nearest with it in delegation class half-pixs do 4 one nearest with it in reference image vegetarian refreshments or same row class half-pixs and select and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation; Wherein a class half-pix is used and is done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation.

5, the method for a kind of 1/4 pixel precision interpolation according to claim 1, it is characterized in that: two classes, 1/4 pixel being used to select with its 4 one nearest class half-pixs in the delegation do 4 one nearest with it in reference image vegetarian refreshments or same row class half-pixs and select and do the reference image vegetarian refreshments and obtained by four tap dissymetrical filter interpolation, is mirror filter to the filter of two classes, the 1/4 pixel interpolation that is positioned at adjacent 2 class half-pix points left side or top with to the filter of two classes, the 1/4 pixel interpolation that is positioned at adjacent two class half-pix point right sides or below wherein; A class half-pix of doing reference is selected to use and is done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation.

6, the method for a kind of 1/4 pixel precision interpolation according to claim 1 is characterized in that: three classes, 1/4 pixel o'clock is obtained by the symmetrical linear filter interpolation of two taps with being positioned at the adjacent whole pixel of its 45 degree and two class half-pixs; Wherein a class half-pix is used and is done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation; Two class half-pixs are used to select with 4 one nearest with it in delegation class half-pixs and are done 4 one nearest with it in reference image vegetarian refreshments or same row class half-pixs and select and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation.

7, the method for a kind of 1/4 pixel precision interpolation according to claim 4 is characterized in that: three classes, 1/4 pixel o'clock is obtained by the symmetrical linear filter interpolation of two taps with being positioned at the adjacent whole pixel of its 45 degree and two class half-pixs; Wherein a class half-pix is used and is done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation; Two class half-pixs are used to select with 4 one nearest with it in delegation class half-pixs and are done 4 one nearest with it in reference image vegetarian refreshments or same row class half-pixs and select and do the reference image vegetarian refreshments and obtained by four tap balanced-filter interpolation.

8, the method for a kind of 1/4 pixel precision interpolation according to claim 3 is characterized in that: three classes, 1/4 pixel is obtained by two taps symmetry linear filter interpolation with being positioned at its two one adjacent class 1/4 pixels; Or use with it and obtain by four tap dissymetrical filter interpolation with delegation's middle distance nearest four one class 1/4 pixels or nearest four one class 1/4 pixels of row middle distance same with it; Or with using four one class 1/4 pixels nearest to obtain with four tap balanced-filter interpolation with it; Wherein a class 1/4 pixel and whole pixel are at 1/4 pixel with delegation or same row, one class, 1/4 pixel is used and is done 4 nearest with it in reference image vegetarian refreshments or the same row whole pixels with 4 nearest with it in the delegation whole pixels and do the reference image vegetarian refreshments and obtained by four tap dissymetrical filter interpolation, is mirror filter to the filter of a class 1/4 pixel interpolation that is positioned at adjacent class half-pix point left side or top with to the filter of a class 1/4 pixel interpolation that is positioned at an adjacent class half-pix point right side or below wherein.

9, the method for a kind of 1/4 pixel precision interpolation according to claim 1 is characterized in that: said filter is four tap balanced-filters, and filter form is: { Co ₁, Co ₂, Co ₂, Co ₁}/[2 * (Co ₁+ Co ₂)], Co _iBe filter tap coefficients, wherein i=1 ... 4, concrete tap coefficient be 3,19,19, and-3}/32 with 1,3,3,1}/8, the filtering formula is: x=(Co ₁* X ₁+ Co ₂* X ₂+ Co ₂* X ₃+ Co ₁* X ₄)/[(Co ₁+ Co ₂) * 2], X wherein _iBe reference image vegetarian refreshments, i=1 ... 4, wherein { 3,19,19 ,-3}/32 filters are used for a class half-pix and two class half-pix interpolation; { 1,3,3,1}/8 filters are used for three classes, 1/4 picture element interpolation.

10, the method for a kind of 1/4 pixel precision interpolation according to claim 1 is characterized in that: said filter is four tap dissymetrical filters, and filter form is: { Co ₁, Co ₂, Co ₃, Co ₄}/(Co ₁+ Co ₂+ Co ₃+ Co ₄), Co _iBe filter tap coefficients, wherein i=1 ... 4, concrete tap coefficient be 5,59,13 ,-3}/64 or 3,29,7,1}/32, the filtering formula is: x=(Co ₁* X ₁+ Co ₂* X ₂+ Co ₂* X ₃+ Co ₁* X ₄)/[(Co ₁+ Co ₂) * 2], X wherein _iBe reference image vegetarian refreshments, i=1 ... 4; This filter is used for the interpolation of a class 1/4 pixel, the interpolation of two classes, 1/4 pixel and the interpolation of three classes, 1/4 pixel.

11, a kind of device of 1/4 pixel precision interpolation, it is characterized in that: it comprises:

Be used for reading the input unit (210) of whole pixel pixel value from primitive frame;

Be used for a class half-pix point interpolation processing unit (220) to a class half-pix point interpolation

The apparatus for temporary storage (230) that is used for a temporary class half-pix point interpolation result;

Be used for a class 1/4 pixel interpolation processor (240) to a class 1/4 pixel interpolation;

The apparatus for temporary storage (250) that is used for a temporary class 1/4 pixel interpolation result

Be used for two class half-pix point interpolation processing unit (260) to two class half-pix point interpolations;

The apparatus for temporary storage (270) that is used for temporary two class half-pix point interpolation results;

Be used for two classes, 1/4 pixel interpolation processor (280) to two classes, 1/4 pixel interpolation;

Be used for three classes, 1/4 pixel interpolation processor (290) to three classes, 1/4 pixel interpolation;

Be used to export the output device (300) of whole sub-pix point pixel values;

Described:

Class half-pix point refers to and puts in order pixel at the half-pix point with delegation or same row;

2 class half-pix points refer to other half-pix points except a class half-pix point;

One class, 1/4 pixel refers to and puts in order pixel at 1/4 pixel with delegation or same row;

Two classes, 1/4 pixel refer to simultaneously with a class half-pix point and 2 class half-pix points with delegation or while and a class half-pix point and 2 class half-pix points 1/4 pixel at same row;

Three classes, 1/4 pixel refers to 1/4 pixel except a class 1/4 pixel and two classes, 1/4 pixel;

All the sub-pix point refers to above 5 class sub-pix points;

Whole pixel refers to each the whole pixel around the sub-pix;

Connect with data/address bus between an input unit (210) and the class half-pix interpolation processor (220), transmit the pixel value of each whole location of pixels; Connect with data/address bus between input unit (210) and three classes, the 1/4 pixel interpolation processor (290), transmit the pixel value of each whole location of pixels; Connect with data/address bus between input unit (210) and a class 1/4 pixel interpolation processing device (240), transmit the pixel value of each whole location of pixels; Connect with data/address bus between one class half-pix interpolation processor (220) and the data apparatus for temporary storage (230), transmit a class

The pixel value of half-pix point position; Connect with data/address bus between one class, 1/4 pixel interpolation processing device (240) and the data apparatus for temporary storage (250), transmit the pixel value of a class 1/4 pixel position; Connect the pixel value of the sub-pixel location that the transmission interpolation obtains between one class half-pix interpolation processor (220) and the output device (300) with data/address bus; Connect the pixel value of the sub-pixel location that the transmission interpolation obtains between one class, 1/4 pixel interpolation processing device (240) and the output device (300) with data/address bus; Connect the pixel value of each class half-pix point position of transmission between data apparatus for temporary storage (230) and the two class half-pix point interpolation processing unit (260) with data/address bus; Connect the pixel value of each class half-pix point position of transmission between data apparatus for temporary storage (230) and two classes, the 1/4 pixel interpolation processor (280) with data/address bus; Connect the pixel value of each class 1/4 pixel position of transmission between data apparatus for temporary storage (250) and three classes, the 1/4 pixel interpolation processor (290) with data/address bus; Connect with data/address bus between two class half-pix point interpolation processing unit (260) and the data apparatus for temporary storage (270), transmit the pixel value of two class half-pix point positions; Connect the pixel value of the sub-pixel location that the transmission interpolation obtains between two class half-pix point interpolation processing unit (260) and the output device (300) with data/address bus; Connect the pixel value of the sub-pixel location that the transmission interpolation obtains between two classes, 1/4 pixel interpolation processor (280) and the output device (300) with data/address bus; Be connected the two class half-pix point pixel values that the transmission interpolation is used between data apparatus for temporary storage (270) and three classes, the 1/4 pixel interpolation processor (290) with data/address bus; Connect the sub-pixel location pixel value that the transmission interpolation obtains between three classes, 1/4 pixel interpolation processor (290) and the output device (300) with data/address bus.

Images