CN102521803A - Anti-aliasing method and device in image scaling - Google Patents

Abstract

The invention relates to the technical field of image processing and provides an anti-aliasing method in image scaling. The anti-aliasing method comprises the following steps of: A, according to the positions of points to be interpolated in an amplified target image, obtaining a brightness matrix corresponding to a preset neighborhood area in an original image; B, according to a brightness difference of the brightness matrix in the preset neighborhood area, judging whether the brightness difference of the points to be interpolated needs to be corrected, if so, carrying out step C, or finishing the process; and C, correcting the brightness difference of the points to be interpolated. The invention also provides an anti-aliasing device in image scaling. By application of the invention, the amplification of steps is restrained and reduced in the process of image amplification, so that an anti-aliasing effect is achieved.

Description

Anti-sawtooth distortion methods and device in the image zoom

Technical field

The present invention relates to the graph processing technique field, particularly a kind of anti-sawtooth distortion methods and device that is used for image zoom.

Background technology

At present, the type of display device is more and more abundanter, and various display device all have best separately demonstration size owing to own characteristic and application scenario is different.On the other hand, the size of importing the picture signal of these display device also has very wide transformation range.Based on this, must dwindle image, processing and amplifying.In addition, also all need carry out the convergent-divergent processing for making the user be absorbed in certain details of image or the whole general picture of acquisition image etc. to image.

It mainly is to realize through image interpolation that image amplifies.Image interpolation is exactly to derive and calculate new Pixel Information according to known image pixel information.Traditional interpolation method comprises nearest-neighbor interpolation, bilinear interpolation and bicubic interpolation.But; The subject matter that existing image magnification method faces comprises image blurring and two aspects of sawtooth distortion; Wherein image blurring to be meant that the details of amplifying the back image thickens unclear, and the sawtooth distortion is meant that enlarged image locates the jagged artificial trace that occurs on the edge of.

Because in digital picture; Except the linear edge of vertical and level, other edge is discontinuous, discrete, and promptly the edge in the image itself exists sawtooth and ladder; The method that image amplifies can make these little ladders become big ladder, thereby forms comparatively significantly sawtooth.If can in the image amplification process, suppress and reduce the amplification of ladder, will suppress the sawtooth distortion at edge.

So, how to suppress just to have become technical matters that needs to be resolved hurrily in the technical field of image processing with the amplification that reduces ladder.

Summary of the invention

The technical matters that (one) will solve

To above-mentioned shortcoming; The present invention is in order to solve the technical matters of the sawtooth distortion that exists in the conventional images amplifying technique; A kind of anti-sawtooth distortion methods and device that is used for image zoom is provided; It can suppress and reduce the amplification of ladder in the image amplification process, thereby reaches the effect of anti-sawtooth distortion.

(2) technical scheme

In order to solve the problems of the technologies described above, on the one hand, the invention provides the anti-sawtooth distortion methods in a kind of image zoom, comprise step:

A: the luminance matrix that obtains preset neighborhood in the corresponding original image according to interpolation point position in the target image after amplifying;

B: judge the whether needs correction of brightness value of said interpolation point according to the luminance difference of the luminance matrix of preset neighborhood, if then carry out step C; Otherwise, process ends;

C: the brightness value of proofreading and correct said interpolation point.

Wherein, said step B specifically comprises:

Step B1: the luminance difference threshold value of calculating the neighborhood far away of the luminance matrix of presetting neighborhood;

Step B2: the luminance difference threshold value of calculating the neighbour territory of the luminance matrix of presetting neighborhood;

Step B3: according to the whether needs correction of the brightness value of the said interpolation point of luminance difference threshold decision of the far and near neighborhood of the luminance matrix of preset neighborhood.

Wherein, said step C specifically comprises:

Step C1: the luminance difference threshold value of calculating said interpolation point;

Step C2: the brightness intermediate value of calculating luminance matrix;

Step C3: the brightness value and the output of proofreading and correct said interpolation point.

Wherein, said step C3 specifically comprises:

Step C31: calculate the brightness intermediate value of luminance matrix and the difference between this interpolation point brightness value behind the image zoom;

Step C32: with the positive negative value of the luminance difference threshold value of interpolation point respectively as the bound of difference;

Step C33: difference compensated on this interpolation point brightness value behind the image zoom export as image.

Wherein, said steps A specifically comprises:

(wherein M is an even number for fi, the neighborhood of the capable N row of M j) in original image, to get the interpolation point; N is an odd number, constitutes the luminance matrix of M * N with the brightness value of the capable N of this a M row neighborhood point, and wherein fi is definite according to formula ; Wherein h_src is the original graph image height, and h_dst is that target image is high, (m; J) be the position of interpolation point in target image; (fi j) is the position of interpolation point in original image

Or,

(wherein L is an odd number for i, the neighborhood of the capable P row of L fj) in original image, to get the interpolation point; P is an even number, constitutes the luminance matrix of L * P with the brightness value of the capable P of this a L row neighborhood point, and wherein fj is definite according to formula

; Wherein w_src is the original graph image width; W_dst is that target image is wide, and (i n) is the position of interpolation point in target image; (i fj) is the position of interpolation point in original image.

Wherein, said step B1 specifically comprises:

The luminance matrix middle one of calculating M * N lists preceding M/2 pixel and first row is gone up maximum brightness value top1_max and minimum luminance value top1_min in other pixels; And calculating middle one lists maximum brightness value bot1_max and the minimum luminance value bot1_min in other pixels on M/2 the pixel in back and last column; And according to formula diff1=max (top1_min-bot1_max; Bot1_min-top1_max) the luminance difference threshold value diff1 of the neighborhood far away of calculating luminance matrix; And when diff1＜0, make diff1=0; Or,

The luminance matrix middle row of calculating L * P P/2 the pixel and first of going forward lists maximum brightness value left1_max and minimum luminance value left1_min in other pixels; And P/2 the pixel in back lists maximum brightness value right1_max and minimum luminance value right1_min in other pixels with last on the calculating middle row; And according to formula diff1=max (left1_min-right1_max; Right1_min-left1_max) the luminance difference threshold value diff1 of the neighborhood far away of calculating luminance matrix; And when diff1＜0, make diff1=0.

Wherein, said step B2 specifically comprises:

Calculate in the middle of the luminance matrix of M * N one list before M/2 pixel and M/2 capable on maximum brightness value top2_max and minimum luminance value top2_min in other pixels; And in the middle of calculating one list M/2 the pixel in back and M/2+1 capable on maximum brightness value bot2_max and minimum luminance value bot2_min in other pixels; And according to formula diff2=max (top2_min-bot2_max; Bot2_min-top2_max) the luminance difference threshold value diff2 in the neighbour territory of calculating luminance matrix; And when diff2＜0, make diff2=0; Or,

Go forward P/2 pixel and P/2 of the luminance matrix middle row of calculating L * P lists maximum brightness value left2_max and minimum luminance value left2_min in other pixels; And P/2 the pixel in back and P/2+1 list maximum brightness value right2_max and minimum luminance value right2_min in other pixels on the calculating middle row; And according to formula diff2=max (left2_min-right2_max; Right2_min-left2_max) the luminance difference threshold value diff2 in the neighbour territory of calculating luminance matrix pixel2; And when diff2＜0, make diff2=0.

Wherein, said step B3 specifically comprises:

Obtain the luminance difference threshold value m_diff of mid point of the luminance matrix of preset neighborhood according to formula m_diff=diff1-diff2, when m_diff＞0, confirm that the brightness value of said interpolation point needs to proofread and correct, otherwise do not need correction.

Wherein, said step C1 specifically comprises:

Obtain the luminance difference threshold value threshold of interpolation point according to the luminance difference threshold value m_diff of offset distance Δ y and mid point longitudinally, computing formula is:

$\mathrm{Threshold}=\left\{\begin{array}{cc}\frac{\mathrm{\&Delta;\; y}}{0.5}\&times;m\_\mathrm{Diff}& 0\&le;\mathrm{\&Delta;\; y}\&le;0.5\\ \frac{1-\mathrm{\&Delta;\; y}}{0.5}\&times;m\_\mathrm{Diff}& 0.5<\mathrm{\&Delta;\; y}\&le;1\end{array}\right.;$ Or,

Obtain the luminance difference threshold value threshold of interpolation point according to the luminance difference threshold value m_diff of horizontal offset distance Δ x and mid point, computing formula is:

$\mathrm{threshold}=\left\{\begin{array}{cc}\frac{\mathrm{\&Delta;x}}{0.5}\&times;m\_\mathrm{diff}& 0\&le;\mathrm{\&Delta;x}\&le;0.5\\ \frac{1-\mathrm{\&Delta;x}}{0.5}\&times;m\_\mathrm{diff}& 0.5<\mathrm{\&Delta;x}\&le;1\end{array}\right..$

Wherein, said step C2 specifically comprises:

Brightness value to each pixel of the luminance matrix center of M * N two row sorts from low to high, gets the brightness value of the mean value avg of two brightness values in the middle of the ordering back as the luminance matrix mid point; Obtain the brightness intermediate value median of luminance matrix according to following formula,

$\mathrm{median}=\left\{\begin{array}{cc}2\&times;[(0.5-\mathrm{\&Delta;y})\&times;{\mathrm{pixel}}_{i,j}+\mathrm{\&Delta;y}\&times;\mathrm{avg}]& 0\&le;\mathrm{\&Delta;y}\&le;0.5\\ 2\&times;[(\mathrm{\&Delta;y}-0.5)\&times;{\mathrm{pixel}}_{i+1,j}+(1-\mathrm{\&Delta;y})\&times;\mathrm{avg}]& 0.5<\mathrm{\&Delta;y}\&le;1\end{array}\right.,$

Pixel wherein _{I, j}Expression point (i, brightness value j), pixel _{I+1, j}(i+1, brightness value j), i are following value of rounding of horizontal ordinate fi to the expression point; Or,

Brightness value to each pixel of the luminance matrix center of L * P two row sorts from low to high, gets the brightness value of the mean value avg of two brightness values in the middle of the ordering back as the luminance matrix mid point, obtains the brightness intermediate value median of luminance matrix according to following formula,

$\mathrm{median}=\left\{\begin{array}{cc}2\&times;[(0.5-\mathrm{\&Delta;x})\&times;{\mathrm{pixel}}_{i,j}+\mathrm{\&Delta;x}\&times;\mathrm{avg}]& 0\&le;\mathrm{\&Delta;x}\&le;0.5\\ 2\&times;[(\mathrm{\&Delta;x}-0.5)\&times;{\mathrm{pixel}}_{i,j+1}+(1-\mathrm{\&Delta;x})\&times;\mathrm{avg}]& 0.5<\mathrm{\&Delta;x}\&le;1\end{array}\right.,$

Pixel wherein _{I, j}Expression point (i, brightness value j), pixel _{I, j+1}(i, brightness value j+1), j are following value of rounding of ordinate fj to the expression point.

The invention also discloses the anti-sawtooth distortion device in a kind of image zoom, comprising:

The luminance matrix acquiring unit is used for obtaining presetting in the corresponding original image according to the target image interpolation point position after amplifying the luminance matrix of neighborhood;

Proofread and correct decision unit, be used for judging the whether needs correction of brightness value of said interpolation point according to the luminance difference of the luminance matrix of preset neighborhood;

The brightness value correcting unit is used to proofread and correct the brightness value of said interpolation point.

Wherein, said correction decision unit specifically comprises:

Neighborhood luminance difference threshold calculations subelement far away is used to calculate the luminance difference threshold value of neighborhood far away of the luminance matrix of preset neighborhood;

Neighbour territory luminance difference threshold calculations subelement is used to calculate the luminance difference threshold value in neighbour territory of the luminance matrix of preset neighborhood;

Proofread and correct the judgement subelement, be used for according to the whether needs correction of the brightness value of the said interpolation point of luminance difference threshold decision of the far and near neighborhood of the luminance matrix of preset neighborhood.

Wherein, said brightness value correcting unit specifically comprises:

Luminance difference threshold calculations subelement is used to calculate the luminance difference threshold value of said interpolation point;

Brightness median calculation subelement is used to calculate the brightness intermediate value of luminance matrix;

Brightness value syndrome unit, the brightness value and the output that are used to proofread and correct said interpolation point.

Wherein, said brightness value syndrome unit specifically comprises:

Difference calculating module is used to calculate the difference between the brightness intermediate value of luminance matrix and this interpolation point brightness value behind the image zoom;

The difference limit module is used for the positive negative value of the luminance difference threshold value of interpolation point respectively as the bound of difference;

The brightness value correction module is used for difference compensated on this interpolation point brightness value behind the image zoom and exports as image.

(3) beneficial effect

The invention provides anti-sawtooth distortion methods and device in a kind of image zoom; It obtains the luminance matrix of preset neighborhood in the corresponding original image according to interpolation point position in the target image after amplifying; Its transverse and longitudinal of considering image respectively amplifies the luminance matrix of choosing interpolation point preset neighborhood in corresponding original image; Avoided the interference of other factors, simple; Judge the whether needs correction of brightness value of said interpolation point according to the luminance difference of the luminance matrix of preset neighborhood; It is when choosing the luminance difference threshold value of far and near neighborhood; Except farthest or the row of nearest-neighbor or row; Other pixels of going or listing at interpolation point place have also been considered; Because the going or list the most obvious that the brightness value of pixel changes of interpolation point place, therefore the luminance difference threshold value of the feasible far and near neighborhood of choosing is more accurate, and the judgement of making thus of whether proofreading and correct meets reality more; Proofread and correct the brightness value of said interpolation point; With the positive negative value of the luminance difference threshold value of the said interpolation point that calculates bound as the difference between the brightness intermediate value of luminance matrix and this interpolation point brightness value behind the image zoom; With difference limit in a reasonable range; Again said difference is compensated on this interpolation point brightness value behind the image zoom and export as image; Make said interpolation point milder, thereby reach the effect of the sawtooth distortion that suppresses the edge with the transition of neighborhood point.

Description of drawings

Fig. 1 is the process flow diagram of the anti-sawtooth distortion methods in the image zoom of the present invention;

Fig. 2 is the process flow diagram of step B in the anti-sawtooth distortion methods in the image zoom of the present invention;

Fig. 3 is the process flow diagram of step C in the anti-sawtooth distortion methods in the image zoom of the present invention;

Fig. 4 is the process flow diagram of step C3 in the anti-sawtooth distortion methods in the image zoom of the present invention;

Fig. 5 (a)-Fig. 5 (b) is the synoptic diagram as a result of steps A in the anti-sawtooth distortion methods in the embodiment of the invention 1 described image zoom;

Fig. 6 (a) is the synoptic diagram as a result of step B1 in the anti-sawtooth distortion methods in the embodiment of the invention 1 described image zoom;

Fig. 6 (b) is the synoptic diagram as a result of step B2 in the anti-sawtooth distortion methods in the embodiment of the invention 1 described image zoom;

Fig. 7 (a)-Fig. 7 (b) is the synoptic diagram as a result of steps A in the anti-sawtooth distortion methods in the embodiment of the invention 2 described image zooms;

Fig. 8 (a) is the synoptic diagram as a result of step B1 in the anti-sawtooth distortion methods in the embodiment of the invention 2 described image zooms;

Fig. 8 (b) is the synoptic diagram as a result of step B2 in the anti-sawtooth distortion methods in the embodiment of the invention 2 described image zooms;

Fig. 9 (a)-Fig. 9 (d) is the effect contrast figure who goes to the sawtooth front and back of the anti-sawtooth distortion methods in the embodiment of the invention 1 described image zoom;

Figure 10 is the structural representation of the anti-sawtooth distortion device in the image zoom of the present invention.

Embodiment

To combine the accompanying drawing in the embodiment of the invention below, the technical scheme in the embodiment of the invention is carried out clear, intactly description, obviously, described embodiment is a part of embodiment of the present invention, rather than whole embodiment.Based on the embodiment among the present invention, the every other embodiment that those of ordinary skills are obtained under the prerequisite of not making creative work belongs to the scope that the present invention protects.

According to an aspect of the present invention, as shown in Figure 1, the anti-sawtooth distortion methods in the image zoom of the present invention comprises step:

A: the luminance matrix that obtains preset neighborhood in the corresponding original image according to interpolation point position in the target image after amplifying;

When only considering the vertical amplification of image, because the luminance difference between vertical point of transverse edge is bigger, vertical sawtooth of transverse edge also is exaggerated; In order to suppress by the sawtooth of vertical amplification, can to establish the original graph image height is h_src, and the target image height is h_dst; The position of interpolation point in target image be (m, j), then the corresponding position in original image of interpolation point is (fi; J), wherein:

$\mathrm{fi}=m\&times;\frac{h\_\mathrm{src}}{h\_\mathrm{dst}}---\left(1\right)$

(wherein M is an even number for fi, the neighborhood of the capable N row of M j) in original image, to get the interpolation point; N is an odd number, this moment the interpolation point (fi is j) between the middle one middle two rows pixel that lists; Interpolation point (fi; J) (i is Δ y at offset distance longitudinally j), and wherein i is following value of rounding of horizontal ordinate fi with point.Constitute the luminance matrix pixel1 of M * N with the brightness value of the capable N row of this a M neighborhood point.

When only considering the horizontal amplification of image, because the luminance difference between the crosswise spots of longitudinal edge is bigger, the horizontal sawtooth of longitudinal edge also is exaggerated; In order to suppress by the sawtooth of horizontal amplification, can to establish the original graph image width is w_src, and target image is wide to be w_dst; The position of interpolation point in target image be (i, n), then the corresponding position in original image of interpolation point is (i; Fj), wherein:

$\mathrm{fj}=n\&times;\frac{w\_\mathrm{src}}{w\_\mathrm{dst}}---\left(2\right)$

(wherein L is an odd number for i, the neighborhood of the capable P row of L fj) in original image, to get the interpolation point; P is an even number, and (i is fj) between the two row pixels of the centre on the middle row for the interpolation point at this moment; Interpolation point (i; Fj) (i is Δ x at horizontal offset distance j), and wherein j is following value of rounding of ordinate fj with point.Constitute the luminance matrix pixel2 of L * P with the brightness value of the capable P row of this a L neighborhood point.

If both considered the vertical sawtooth when image vertically amplifies, consider the horizontal sawtooth when image laterally amplifies again, then can any order obtain the luminance matrix pixel1 and the luminance matrix pixel2 of preset neighborhood in the corresponding original image according to the method described above respectively.

In this step, consider that respectively the transverse and longitudinal of image amplifies the luminance matrix of choosing interpolation point preset neighborhood in corresponding original image, avoided the interference of other factors, simple.

B: judge the whether needs correction of brightness value of said interpolation point according to the luminance difference of the luminance matrix of preset neighborhood, if then carry out step C; Otherwise, process ends;

Promptly in this step, judge according to the luminance difference of the luminance matrix of presetting neighborhood whether interpolation point is positioned at edge of image.

As shown in Figure 2, this step specifically comprises:

Step B1: the luminance difference threshold value of calculating the neighborhood far away of the luminance matrix of presetting neighborhood;

If the luminance matrix pixel1 of M * N; In the middle of then calculating one list before last maximum brightness value top1_max and the minimum luminance value top1_min in other pixels of M/2 pixel and first row; And one list maximum brightness value bot1_max and the minimum luminance value bot1_min in other pixels on M/2 the pixel in back and last column in the middle of calculating, and according to the luminance difference threshold value diff1 of the neighborhood far away of computes luminance matrix pixel1

diff1＝max(top1_min-bot1_max，bot1_min-top1_max) (3)

Because when the interpolation point was in the transverse edge of image, edge minimum value on one side also was not less than the maximal value of another side, so diff1 >=0.Otherwise, when interpolation point is not the transverse edge of image, for guaranteeing the brightness value behind the image zoom not to be proofreaied and correct, when diff1＜0, make diff1=0.

If the luminance matrix pixel2 of L * P; Then calculate middle row P/2 the pixel and first of going forward and list maximum brightness value left1_max and minimum luminance value left1_min in other pixels; And P/2 the pixel in back and last list maximum brightness value right1_max and minimum luminance value right1_min in other pixels on the calculating middle row, and according to the luminance difference threshold value diff1 of the neighborhood far away of computes luminance matrix pixel2

diff1＝max(left1_min-right1_max，right1_min-left1_max) (4)

Because when the interpolation point was in the longitudinal edge of image, edge minimum value on one side also was not less than the maximal value of another side, so diff1 >=0.Otherwise, when interpolation point is not the longitudinal edge of image, for guaranteeing the brightness value behind the image zoom not to be proofreaied and correct, when diff1＜0, make diff1=0.

Step B2: the luminance difference threshold value of calculating the neighbour territory of the luminance matrix of presetting neighborhood;

If the luminance matrix pixel1 of M * N; In the middle of then calculating one list before M/2 pixel and M/2 capable on maximum brightness value top2_max and minimum luminance value top2_min in other pixels; And in the middle of calculating one list M/2 the pixel in back and M/2+1 capable on maximum brightness value bot2_max and minimum luminance value bot2_min in other pixels, and according to the luminance difference threshold value diff2 in the neighbour territory of computes luminance matrix pixel1

diff2＝max(top2_min-bot2_max，bot2_min-top2_max) (5)

Because when the interpolation point was in the transverse edge of image, edge minimum value on one side also was not less than the maximal value of another side, so diff2 >=0.Otherwise, when interpolation point is not the transverse edge of image, for guaranteeing the brightness value behind the image zoom not to be proofreaied and correct, when diff2＜0, make diff2=0.

If the luminance matrix pixel2 of L * P; Then calculate go forward P/2 pixel and P/2 of middle row and list maximum brightness value left2_max and minimum luminance value left2_min in other pixels; And P/2 the pixel in back and P/2+1 list maximum brightness value right2_max and minimum luminance value right2_min in other pixels on the calculating middle row, and according to the luminance difference threshold value diff2 in the neighbour territory of computes luminance matrix pixel2

diff2＝max(left2_min-right2_max，right2_min-left2_max) (6)

Because when the interpolation point was in the longitudinal edge of image, edge minimum value on one side also was not less than the maximal value of another side, so diff2 >=0.Otherwise, when interpolation point is not the longitudinal edge of image, for guaranteeing the brightness value behind the image zoom not to be proofreaied and correct, when diff2＜0, make diff2=0.

The execution of above-mentioned steps B1 and B2 does not have sequencing, can carry out simultaneously, can any order successively carry out yet.

Step B3: according to the whether needs correction of the brightness value of the said interpolation point of luminance difference threshold decision of the far and near neighborhood of the luminance matrix of preset neighborhood.

In this step, the luminance difference threshold value m_diff of the mid point of the luminance matrix of preset neighborhood is:

m_diff＝diff1-diff2 (7)

Owing in edge of image, comprise transverse edge and/or longitudinal edge, generally the difference than neighbour territory pixel is big for the difference of neighborhood territory pixel far away, so m_diff >=0.For guaranteeing not not being that the brightness value of image border is proofreaied and correct, when m_diff＜0, make m_diff=0.

Promptly when m_diff=0; Represent that then this interpolation point is not in edge of image; Comprise transverse edge and/or longitudinal edge; Therefore image is exported this interpolation point brightness value scl_pix after result equals image zoom, promptly need not proofread and correct this interpolation point, otherwise need proofread and correct this interpolation point.

In this step; When choosing the luminance difference threshold value of far and near neighborhood; Except farthest or the row of nearest-neighbor or row, other pixels of going or listing at interpolation point place have also been considered, because the going or list brightness value variation the most obvious of pixel of interpolation point place; Therefore the luminance difference threshold value of the feasible far and near neighborhood of choosing is more accurate, and the judgement of making thus of whether proofreading and correct meets reality more.

C: the brightness value of proofreading and correct said interpolation point.

As shown in Figure 3, this step specifically comprises:

Step C1: the luminance difference threshold value threshold that calculates said interpolation point;

If the luminance matrix pixel1 of M * N; The distance of then establishing between original pixels is 1; Offset distance 0≤Δ y≤1 longitudinally then obtains the luminance difference threshold value threshold of interpolation point according to the luminance difference threshold value m_diff of offset distance Δ y and mid point longitudinally, and computing formula is:

$\mathrm{threshold}=\left\{\begin{array}{cc}\frac{\mathrm{\&Delta;y}}{0.5}\&times;m\_\mathrm{diff}& 0\&le;\mathrm{\&Delta;y}\&le;0.5\\ \frac{1-\mathrm{\&Delta;y}}{0.5}\&times;m\_\mathrm{diff}& 0.5<\mathrm{\&Delta;y}\&le;1\end{array}\right.---\left(8\right)$

If the luminance matrix pixel2 of L * P; The distance of then establishing between original pixels is 1; Then horizontal offset distance 0≤Δ x≤1 obtains the luminance difference threshold value threshold of interpolation point according to the luminance difference threshold value m_diff of horizontal offset distance Δ x and mid point, and computing formula is:

$\mathrm{threshold}=\left\{\begin{array}{cc}\frac{\mathrm{\&Delta;x}}{0.5}\&times;m\_\mathrm{diff}& 0\&le;\mathrm{\&Delta;x}\&le;0.5\\ \frac{1-\mathrm{\&Delta;x}}{0.5}\&times;m\_\mathrm{diff}& 0.5<\mathrm{\&Delta;x}\&le;1\end{array}\right.---\left(9\right)$

Step C2: the brightness intermediate value median that calculates luminance matrix;

If the luminance matrix pixel1 of M * N, then the brightness value to each pixel of its center two row sorts from low to high, gets the brightness value of the mean value avg of two brightness values in the middle of the ordering back as the luminance matrix mid point;

If the distance between original pixels is 1, offset distance 0≤Δ y≤1 longitudinally then.If 0≤Δ y≤0.5, then (i carries out linear interpolation between j), obtains the brightness intermediate value median of luminance matrix at luminance matrix mid point and point; If 0.5＜Δ y≤1, then (i+1 carries out linear interpolation between j), obtains the brightness intermediate value median of luminance matrix at luminance matrix mid point and point.Computing formula is:

$\mathrm{median}=\left\{\begin{array}{cc}2\&times;[(0.5-\mathrm{\&Delta;y})\&times;{\mathrm{pixel}}_{i,j}+\mathrm{\&Delta;y}\&times;\mathrm{avg}]& 0\&le;\mathrm{\&Delta;y}\&le;0.5\\ 2\&times;[(\mathrm{\&Delta;y}-0.5)\&times;{\mathrm{pixel}}_{i+1,j}+(1-\mathrm{\&Delta;y})\&times;\mathrm{avg}]& 0.5<\mathrm{\&Delta;y}\&le;1\end{array}\right.---\left(10\right)$

Pixel wherein _{I, j}Expression point (i, brightness value j), pixel _{I+1, j}Expression point (i+1, brightness value j).

If the luminance matrix pixel2 of L * P, then the brightness value to each pixel of its center two row sorts from low to high, gets the brightness value of the mean value avg of two brightness values in the middle of the ordering back as the luminance matrix mid point;

If the distance between original pixels is 1, then horizontal offset distance 0≤Δ x≤1.If 0≤Δ x≤0.5, then (i carries out linear interpolation between j), obtains intermediate value median at luminance matrix mid point and point; If 0.5＜Δ x≤1, then (i carries out linear interpolation between j+1), obtains intermediate value median at luminance matrix mid point and point.Computing formula is:

$\mathrm{median}=\left\{\begin{array}{cc}2\&times;[(0.5-\mathrm{\&Delta;x})\&times;{\mathrm{pixel}}_{i,j}+\mathrm{\&Delta;x}\&times;\mathrm{avg}]& 0\&le;\mathrm{\&Delta;x}\&le;0.5\\ 2\&times;[(\mathrm{\&Delta;x}-0.5)\&times;{\mathrm{pixel}}_{i,j+1}+(1-\mathrm{\&Delta;x})\&times;\mathrm{avg}]& 0.5<\mathrm{\&Delta;x}\&le;1\end{array}\right.---\left(11\right)$

Pixel wherein _{I, j}Expression point (i, brightness value j), pixel _{I, j+1}Expression point (i, brightness value j+1).

Step C3: the brightness value and the output of proofreading and correct said interpolation point.

As shown in Figure 4, this step specifically comprises:

Step C31: calculate the brightness intermediate value median of luminance matrix and the difference diff between this interpolation point brightness value scl_pix behind the image zoom:

diff＝median-scl_pix (12)

Step C32: with the positive negative value of the luminance difference threshold value threshold of interpolation point respectively as the bound of difference diff;

Utilize following formula that diff is limited in the reasonable range, that is:

$\mathrm{diff}=\left\{\begin{array}{cc}-\mathrm{threshold}& \mathrm{diff}<-\mathrm{threshold}\\ \mathrm{diff}& -\mathrm{threshold}\&le;\mathrm{diff}\&le;+\mathrm{threshold}\\ +\mathrm{threshold}& \mathrm{diff}>+\mathrm{threshold}\end{array}\right.---\left(13\right)$

Step C33: this interpolation point brightness value scl_pix that difference diff is compensated to behind the image zoom goes up as image output result:

result＝scl_pix+diff (14)

Therefore; When the interpolation point is in the edge in the image; Comprise transverse edge and/or longitudinal edge, image output result is the result after limiting according to the luminance difference threshold value threshold of interpolation point between the brightness intermediate value median of luminance matrix and the brightness value scl_pix behind the image zoom;

When interpolation point is not in the edge in the image, comprise transverse edge and/or longitudinal edge, image output result is the brightness value scl_pix behind the image zoom.

In this step; With the positive negative value of the luminance difference threshold value of the said interpolation point that calculates bound as the difference between the brightness intermediate value of luminance matrix and this interpolation point brightness value behind the image zoom; With difference limit in a reasonable range; Again said difference is compensated on this interpolation point brightness value behind the image zoom as image output, make said interpolation point milder, thereby reach the effect of the sawtooth distortion at inhibition edge with the transition of neighborhood point.

Embodiment 1:

Shown in Fig. 5-6, in the present embodiment, vertically amplifying original image and vertical sawtooth is suppressed with the bicubic interpolation method is that example describes.

Shown in Fig. 5 (a), (then (fi, j) between the 2nd pixel of middle row and the 3rd pixel, the M value is 4 consistent with vertical 4 points of bicubic interpolation needs to the interpolation point for i, 4 * 3 neighborhoods j) in steps A, to get point.

Shown in Fig. 6 (a); Coordinate is the maximum brightness value top1_max and the minimum luminance value top1_min of 00,01,02 and 11 4 point in 4 * 3 the luminance matrix pixel1 that need to calculate among the step B1 shown in Fig. 5 (b), and calculates that coordinate is the maximum brightness value bot1_max and the minimum luminance value bot1_min of 21,30,31 and 32 4 points among 4 * 3 the luminance matrix pixel1.

Shown in Fig. 6 (b); Need to calculate in step B2 that coordinate is the maximum brightness value top2_max and the minimum luminance value top2_min of 01,10,11 and 12 4 point among 4 * 3 the luminance matrix pixel1, and calculate that coordinate is the maximum brightness value bot2_max and the minimum luminance value bot2_min of 20,21,22 and 31 4 points among 4 * 3 the luminance matrix pixel1.

The brightness value that need be 10,11,12,20,21 and 22 6 points in step C2 to the coordinate among 4 * 3 the luminance matrix pixel1 sorts from low to high, gets the brightness value of the mean value avg of middle two brightness values in ordering back as the luminance matrix mid point.

Embodiment 2:

Shown in Fig. 7-8, in the present embodiment, laterally amplifying original image and horizontal sawtooth is suppressed with the bicubic interpolation method is that example describes.

Shown in Fig. 7 (a), (then (i, fj) between the 2nd pixel of middle row and the 3rd pixel, the P value is 4 to need vertical 4 points consistent with bicubic interpolation to the interpolation point for i, 3 * 4 neighborhoods j) in steps A, to get mid point.

Shown in Fig. 8 (a); Coordinate is the maximum brightness value left1_max and the minimum luminance value left1_min of 00,10,20 and 11 4 point in 3 * 4 the luminance matrix pixel2 that need to calculate among the step B1 shown in Fig. 7 (b), and calculates that coordinate is the maximum brightness value right1_max and the minimum luminance value right1_min of 03,13,23 and 12 4 point among 3 * 4 the luminance matrix pixel2.

Shown in Fig. 8 (b); Need to calculate in step B2 that coordinate is the maximum brightness value left2_max and the minimum luminance value left2_min of 10,01,11 and 21 4 points among 3 * 4 the luminance matrix pixel2, and calculate that coordinate is the maximum brightness value right2_max and the minimum luminance value right2_min of 02,12,22 and 13 4 point among 3 * 4 the luminance matrix pixel2.

The brightness value that need be 01,11,21,02,12 and 22 6 point in step C2 to the coordinate among 3 * 4 the luminance matrix pixel2 sorts from low to high, gets the brightness value of the mean value avg of middle two brightness values in ordering back as the luminance matrix mid point.

Embodiment 3:

In the present embodiment, amplifying original image and the transverse and longitudinal sawtooth is all suppressed with bicubic interpolation method transverse and longitudinal is that example describes.

Promptly carry out embodiment 1 and embodiment 2 respectively with random order.

Effect comparison is as shown in Figure 9 before and after being used for described in the embodiment of the invention 1 removing sawtooth in the method for the anti-sawtooth distortion of image zoom; Wherein Fig. 9 (a) is the original image before the convergent-divergent; Fig. 9 (b) is the part sectional drawing of Fig. 9 (a); Can find that the edge of circular arc itself just exists little ladder in the original image.Fig. 9 (c) becomes the part sectional drawing after 2 * 2 times for Fig. 9 (a) is carried out bicubic interpolation, can find, the little ladder at the edge of circular arc also has been exaggerated, and has produced the sawtooth distortion.Fig. 9 (d) can find that for Fig. 9 (c) vertically being removed the part sectional drawing behind the sawtooth the vertical sawtooth among Fig. 9 (c) has obtained effective inhibition in Fig. 9 (d).

Shown in figure 10, according to a further aspect in the invention, the present invention also provides the anti-sawtooth distortion device in a kind of image zoom to comprise simultaneously:

The luminance matrix acquiring unit is used for obtaining presetting in the corresponding original image according to the target image interpolation point position after amplifying the luminance matrix of neighborhood;

Proofread and correct decision unit, be used for judging the whether needs correction of brightness value of said interpolation point according to the luminance difference of the luminance matrix of preset neighborhood;

The brightness value correcting unit is used to proofread and correct the brightness value of said interpolation point.

Said correction decision unit specifically comprises:

Neighborhood luminance difference threshold calculations subelement far away is used to calculate the luminance difference threshold value of neighborhood far away of the luminance matrix of preset neighborhood;

Neighbour territory luminance difference threshold calculations subelement is used to calculate the luminance difference threshold value in neighbour territory of the luminance matrix of preset neighborhood;

Proofread and correct the judgement subelement, be used for according to the whether needs correction of the brightness value of the said interpolation point of luminance difference threshold decision of the far and near neighborhood of the luminance matrix of preset neighborhood.

Said brightness value correcting unit specifically comprises:

Luminance difference threshold calculations subelement is used to calculate the luminance difference threshold value of said interpolation point;

Brightness median calculation subelement is used to calculate the brightness intermediate value of luminance matrix;

Brightness value syndrome unit, the brightness value and the output that are used to proofread and correct said interpolation point.

Said brightness value syndrome unit specifically comprises:

Difference calculating module is used to calculate the difference between the brightness intermediate value of luminance matrix and this interpolation point brightness value behind the image zoom;

The difference limit module is used for the positive negative value of the luminance difference threshold value of interpolation point respectively as the bound of difference;

The brightness value correction module is used for difference compensated on this interpolation point brightness value behind the image zoom and exports as image.

In sum; The invention provides anti-sawtooth distortion methods and device in a kind of image zoom; It obtains the luminance matrix of preset neighborhood in the corresponding original image according to interpolation point position in the target image after amplifying; Its transverse and longitudinal of considering image respectively amplifies the luminance matrix of choosing interpolation point preset neighborhood in corresponding original image, has avoided the interference of other factors, and is simple; Judge the whether needs correction of brightness value of said interpolation point according to the luminance difference of the luminance matrix of preset neighborhood; It is when choosing the luminance difference threshold value of far and near neighborhood; Except farthest or the row of nearest-neighbor or row; Other pixels of going or listing at interpolation point place have also been considered; Because the going or list the most obvious that the brightness value of pixel changes of interpolation point place, therefore the luminance difference threshold value of the feasible far and near neighborhood of choosing is more accurate, and the judgement of making thus of whether proofreading and correct meets reality more; Proofread and correct the brightness value of said interpolation point; With the positive negative value of the luminance difference threshold value of the said interpolation point that calculates bound as the difference between the brightness intermediate value of luminance matrix and this interpolation point brightness value behind the image zoom; With difference limit in a reasonable range; Again said difference is compensated on this interpolation point brightness value behind the image zoom and export as image; Make said interpolation point milder, thereby reach the effect of the sawtooth distortion that suppresses the edge with the transition of neighborhood point.

Above embodiment only is used to explain the present invention; And be not limitation of the present invention; The those of ordinary skill in relevant technologies field under the situation that does not break away from the spirit and scope of the present invention, can also be made various variations and modification; Therefore all technical schemes that are equal to also belong to category of the present invention, and real protection scope of the present invention should be defined by the claims.

Claims (14)

1. the anti-sawtooth distortion methods in the image zoom is characterized in that, comprises step:

A: the luminance matrix that obtains preset neighborhood in the corresponding original image according to interpolation point position in the target image after amplifying;

B: judge the whether needs correction of brightness value of said interpolation point according to the luminance difference of the luminance matrix of preset neighborhood, if then carry out step C; Otherwise, process ends;

C: the brightness value of proofreading and correct said interpolation point.

2. anti-sawtooth distortion methods according to claim 1 is characterized in that said step B specifically comprises:

Step B1: the luminance difference threshold value of calculating the neighborhood far away of the luminance matrix of presetting neighborhood;

Step B2: the luminance difference threshold value of calculating the neighbour territory of the luminance matrix of presetting neighborhood;

Step B3: according to the whether needs correction of the brightness value of the said interpolation point of luminance difference threshold decision of the far and near neighborhood of the luminance matrix of preset neighborhood.

3. anti-sawtooth distortion methods according to claim 2 is characterized in that said step C specifically comprises:

Step C1: the luminance difference threshold value of calculating said interpolation point;

Step C2: the brightness intermediate value of calculating luminance matrix;

Step C3: the brightness value and the output of proofreading and correct said interpolation point.

4. anti-sawtooth distortion methods according to claim 3 is characterized in that said step C3 specifically comprises:

Step C31: calculate the brightness intermediate value of luminance matrix and the difference between this interpolation point brightness value behind the image zoom;

Step C32: with the positive negative value of the luminance difference threshold value of interpolation point respectively as the bound of difference;

Step C33: difference compensated on this interpolation point brightness value behind the image zoom export as image.

5. anti-sawtooth distortion methods according to claim 4 is characterized in that said steps A specifically comprises:

(wherein M is an even number for fi, the neighborhood of the capable N row of M j) in original image, to get the interpolation point; N is an odd number, constitutes the luminance matrix of M * N with the brightness value of the capable N of this a M row neighborhood point, and wherein fi is definite according to formula

; Wherein h_src is the original graph image height, and h_dst is that target image is high, (m; J) be the position of interpolation point in target image; (fi j) is the position of interpolation point in original image

Or,

(wherein L is an odd number for i, the neighborhood of the capable P row of L fj) in original image, to get the interpolation point; P is an even number, constitutes the luminance matrix of L * P with the brightness value of the capable P of this a L row neighborhood point, and wherein fj is definite according to formula ; Wherein w_src is the original graph image width; W_dst is that target image is wide, and (i n) is the position of interpolation point in target image; (i fj) is the position of interpolation point in original image.

6. anti-sawtooth distortion methods according to claim 5 is characterized in that said step B1 specifically comprises:

The luminance matrix middle one of calculating M * N lists preceding M/2 pixel and first row is gone up maximum brightness value top1_max and minimum luminance value top1_min in other pixels; And calculating middle one lists maximum brightness value bot1_max and the minimum luminance value bot1_min in other pixels on M/2 the pixel in back and last column; And according to formula diff1=max (top1_min-bot1_max; Bot1_min-top1_max) the luminance difference threshold value diff1 of the neighborhood far away of calculating luminance matrix; And when diff1＜0, make diff1=0; Or,

The luminance matrix middle row of calculating L * P P/2 the pixel and first of going forward lists maximum brightness value left1_max and minimum luminance value left1_min in other pixels; And P/2 the pixel in back lists maximum brightness value right1_max and minimum luminance value right1_min in other pixels with last on the calculating middle row; And according to formula diff1=max (left1_min-right1_max; Right1_min-left1_max) the luminance difference threshold value diff1 of the neighborhood far away of calculating luminance matrix; And when diff1＜0, make diff1=0.

7. anti-sawtooth distortion methods according to claim 6 is characterized in that said step B2 specifically comprises:

Calculate in the middle of the luminance matrix of M * N one list before M/2 pixel and M/2 capable on maximum brightness value top2_max and minimum luminance value top2_min in other pixels; And in the middle of calculating one list M/2 the pixel in back and M/2+1 capable on maximum brightness value bot2_max and minimum luminance value bot2_min in other pixels; And according to formula diff2=max (top2_min-bot2_max; Bot2_min-top2_max) the luminance difference threshold value diff2 in the neighbour territory of calculating luminance matrix; And when diff2＜0, make diff2=0; Or,

Go forward P/2 pixel and P/2 of the luminance matrix middle row of calculating L * P lists maximum brightness value left2_max and minimum luminance value left2_min in other pixels; And P/2 the pixel in back and P/2+1 list maximum brightness value right2_max and minimum luminance value right2_min in other pixels on the calculating middle row; And according to formula diff2=max (left2_min-right2_max; Right2_min-left2_max) the luminance difference threshold value diff2 in the neighbour territory of calculating luminance matrix pixel2; And when diff2＜0, make diff2=0.

8. anti-sawtooth distortion methods according to claim 7 is characterized in that said step B3 specifically comprises:

Obtain the luminance difference threshold value m_diff of mid point of the luminance matrix of preset neighborhood according to formula m_diff=diff1-diff2, when m_diff＞0, confirm that the brightness value of said interpolation point needs to proofread and correct, otherwise do not need correction.

9. anti-sawtooth distortion methods according to claim 8 is characterized in that said step C1 specifically comprises:

Obtain the luminance difference threshold value threshold of interpolation point according to the luminance difference threshold value m_diff of offset distance Δ y and mid point longitudinally, computing formula is:

$\mathrm{Threshold}=\left\{\begin{array}{cc}\frac{\mathrm{\&Delta;\; y}}{0.5}\&times;m\_\mathrm{Diff}& 0\&le;\mathrm{\&Delta;\; y}\&le;0.5\\ \frac{1-\mathrm{\&Delta;\; y}}{0.5}\&times;m\_\mathrm{Diff}& 0.5<\mathrm{\&Delta;\; y}\&le;1\end{array}\right.;$ Or,

Obtain the luminance difference threshold value threshold of interpolation point according to the luminance difference threshold value m_diff of horizontal offset distance Δ x and mid point, computing formula is:

$\mathrm{threshold}=\left\{\begin{array}{cc}\frac{\mathrm{\&Delta;x}}{0.5}\&times;m\_\mathrm{diff}& 0\&le;\mathrm{\&Delta;x}\&le;0.5\\ \frac{1-\mathrm{\&Delta;x}}{0.5}\&times;m\_\mathrm{diff}& 0.5<\mathrm{\&Delta;x}\&le;1\end{array}\right..$

10. anti-sawtooth distortion methods according to claim 9 is characterized in that said step C2 specifically comprises:

Brightness value to each pixel of the luminance matrix center of M * N two row sorts from low to high, gets the brightness value of the mean value avg of two brightness values in the middle of the ordering back as the luminance matrix mid point; Obtain the brightness intermediate value median of luminance matrix according to following formula,

$\mathrm{median}=\left\{\begin{array}{cc}2\&times;[(0.5-\mathrm{\&Delta;y})\&times;{\mathrm{pixel}}_{i,j}+\mathrm{\&Delta;y}\&times;\mathrm{avg}]& 0\&le;\mathrm{\&Delta;y}\&le;0.5\\ 2\&times;[(\mathrm{\&Delta;y}-0.5)\&times;{\mathrm{pixel}}_{i+1,j}+(1-\mathrm{\&Delta;y})\&times;\mathrm{avg}]& 0.5<\mathrm{\&Delta;y}\&le;1\end{array}\right.,$

Pixel wherein _{I, j}Expression point (i, brightness value j), pixel _{I+1, j}(i+1, brightness value j), i are following value of rounding of horizontal ordinate fi to the expression point; Or,

Brightness value to each pixel of the luminance matrix center of L * P two row sorts from low to high, gets the brightness value of the mean value avg of two brightness values in the middle of the ordering back as the luminance matrix mid point, obtains the brightness intermediate value median of luminance matrix according to following formula,

$\mathrm{median}=\left\{\begin{array}{cc}2\&times;[(0.5-\mathrm{\&Delta;x})\&times;{\mathrm{pixel}}_{i,j}+\mathrm{\&Delta;x}\&times;\mathrm{avg}]& 0\&le;\mathrm{\&Delta;x}\&le;0.5\\ 2\&times;[(\mathrm{\&Delta;x}-0.5)\&times;{\mathrm{pixel}}_{i,j+1}+(1-\mathrm{\&Delta;x})\&times;\mathrm{avg}]& 0.5<\mathrm{\&Delta;x}\&le;1\end{array}\right.,$

Pixel wherein _{I, j}Expression point (i, brightness value j), pixel _{I, j+1}(i, brightness value j+1), j are following value of rounding of ordinate fj to the expression point.

11. the anti-sawtooth distortion device in the image zoom is characterized in that, comprising:

The luminance matrix acquiring unit is used for obtaining presetting in the corresponding original image according to the target image interpolation point position after amplifying the luminance matrix of neighborhood;

Proofread and correct decision unit, be used for judging the whether needs correction of brightness value of said interpolation point according to the luminance difference of the luminance matrix of preset neighborhood;

The brightness value correcting unit is used to proofread and correct the brightness value of said interpolation point.

12. anti-sawtooth distortion device according to claim 11 is characterized in that said correction decision unit specifically comprises:

Neighborhood luminance difference threshold calculations subelement far away is used to calculate the luminance difference threshold value of neighborhood far away of the luminance matrix of preset neighborhood;

Neighbour territory luminance difference threshold calculations subelement is used to calculate the luminance difference threshold value in neighbour territory of the luminance matrix of preset neighborhood;

Proofread and correct the judgement subelement, be used for according to the whether needs correction of the brightness value of the said interpolation point of luminance difference threshold decision of the far and near neighborhood of the luminance matrix of preset neighborhood.

13. anti-sawtooth distortion device according to claim 12 is characterized in that said brightness value correcting unit specifically comprises:

Luminance difference threshold calculations subelement is used to calculate the luminance difference threshold value of said interpolation point;

Brightness median calculation subelement is used to calculate the brightness intermediate value of luminance matrix;

Brightness value syndrome unit, the brightness value and the output that are used to proofread and correct said interpolation point.

14. anti-sawtooth distortion device according to claim 13 is characterized in that, said brightness value syndrome unit specifically comprises:

Difference calculating module is used to calculate the difference between the brightness intermediate value of luminance matrix and this interpolation point brightness value behind the image zoom;

The difference limit module is used for the positive negative value of the luminance difference threshold value of interpolation point respectively as the bound of difference;

The brightness value correction module is used for difference compensated on this interpolation point brightness value behind the image zoom and exports as image.

Images