CN106920216A - A kind of method and device for eliminating picture noise - Google Patents

Abstract

A kind of method and device for eliminating picture noise, described image includes p image block, and the size of described image block is odd number for M × N, M and N, and methods described includes：Receive the image block on Bayer domain；By with the pixel of center pixel same channels centered on, take the subimage block of Sm × Sn sizes, wherein：The center pixel is the center pixel of the described image block for receiving, and Sm and Sn is odd number；The filtering weighting coefficient of each subimage block is obtained, described image block is filtered according to the filtering weighting coefficient, the true aberration of the center pixel is obtained, as the first aberration；According to first aberration and default denoising adjustment factor, the pixel value of the center pixel after removal noise is calculated.The amount of calculation of noise processed while denoising sound effective value is ensured, can be reduced using such scheme.

Description

A kind of method and device for eliminating picture noise

Technical field

The present invention relates to image processing field, more particularly to a kind of method and device for eliminating picture noise.

Background technology

Each camera lens on digital camera carries an optical sensor, is used to measure the bright journey of light Degree, but to obtain a width full color image, generally required three optical sensors obtain respectively it is red, green, Blue three primary colours information, and in order to reduce the cost and volume of digital camera, production firm would generally be using electricity Lotus coupled apparatus (Charge-coupled Device, CCD) or complementary metal oxide semiconductors (CMOS) CMOS (Complementary Metal Oxide Semiconductor) imageing sensor, coordinates its surface to cover Color filter matrix (Color Filter Array, CFA) so that each pixel only allows a kind of primary colours By reaching the location of pixels, the noise between primary color channels is different to light, and just due to this Inconsistency, can cause the noise for occurring obvious color on image.

Image processing process, generally includes：The image on Bayer (Bayer) domain is obtained, to described image Color interpolation treatment is carried out, the image after color interpolation treatment is then converted to the figure on RGB domains Picture, then the image on the RGB domains is converted to the image on YUV domains again.

At present, it is the color noise on removal image, to YUV after the image on YUV domains are obtained Image on domain carries out elimination noise processed.

But, noise processed is carried out with the above method, noise processed amount of calculation is very big.

The content of the invention

The problem that the present invention is solved is how while denoising sound effective value is ensured, to reduce the meter of noise processed Calculation amount.

To solve the above problems, the present invention provides a kind of method for eliminating picture noise, and described image includes p Individual image block, the size of described image block is odd number for M × N, M and N, and methods described includes：

Receive the image block on Bayer domain；

By with the pixel of center pixel same channels centered on, take the subimage block of Sm × Sn sizes, wherein： The center pixel is the center pixel of the described image block for receiving, and Sm and Sn is odd number；

The filtering weighting coefficient of each subimage block is obtained, according to the filtering weighting coefficient to described Image block is filtered, and the true aberration of the center pixel is obtained, as the first aberration；

According to first aberration and default denoising adjustment factor, the center after removal noise is calculated The pixel value of pixel.

Alternatively, the filtering weighting coefficient includes filtering aberration weight coefficient, and the acquisition is described in each The filtering weighting coefficient of subimage block, including：

When the pixel non-green, the value of chromatism of each subimage block is calculated；

According to the value of chromatism of each subimage block, the reference aberration of the center pixel is calculated, as Second aberration；

Value of chromatism and second aberration according to each subimage block, are calculated all subgraphs The average color difference of block, as the 3rd aberration；

The mapping relations of the filtering aberration weight coefficient and aberration difference corresponding to the 3rd aberration are selected, The aberration difference is the absolute value of the value of chromatism with the difference of second aberration of each subimage block；

The interval of aberration difference according to each subimage block in the mapping relations, it is determined that with it is every The individual subimage block is corresponding to filter aberration weight coefficient.

Alternatively, the filtering weighting coefficient also includes filtering distance weighting coefficient, also includes：

The mapping relations of the filtering distance weighting coefficient and distance corresponding to the 3rd aberration are selected, it is described Distance is the distance between each described subimage block subimage block corresponding with the center pixel；

The interval of distance according to each subimage block in the mapping relations, it is determined that with the son The corresponding filtering distance weighting coefficient of image block.

Alternatively, the filtering weighting coefficient for obtaining each subimage block, also includes：

When the pixel is for green, calculate described average with the pixel of the center pixel same channels Energy；

The mapping relations of the filtering aberration weight coefficient and aberration difference corresponding to the average energy are selected, The aberration difference is the absolute value of the value of chromatism with the difference of second aberration of each subimage block；

The interval of aberration difference according to each subimage block in the mapping relations, it is determined that and institute State the corresponding filtering aberration weight coefficient of subimage block.

Alternatively, the filtering weighting coefficient also includes filtering distance weighting coefficient, and methods described also includes：

The mapping relations of the filtering distance weighting coefficient and distance corresponding to the average energy are selected, it is described Distance is the distance between each described subimage block subimage block corresponding with the center pixel；

The interval of distance value according to each subimage block in the mapping relations, it is determined that with it is described The corresponding filtering distance weighting coefficient of subimage block.

Alternatively, the value of chromatism for calculating each subimage block, including：

R statistical values, G statistical values and the B statistical values of each subimage block are calculated respectively；

When the center pixel is for red, by the R statistical values of each subimage block and G statistical values Difference, as the value of chromatism of subimage block each described；

When the center pixel is for blueness, by the B statistical values of each subimage block and G statistical values Difference, as the value of chromatism of subimage block each described.

Alternatively, the value of chromatism according to each subimage block, calculates the ginseng of the center pixel Aberration is examined, as the second aberration, including：

Ask for what is selected from the value of chromatism of subimage block each described, the intermediate value of the value of chromatism of predetermined number；

Calculate the intermediate value and the subimage block corresponding to the center pixel of the value of chromatism of the predetermined number The absolute value of the difference of value of chromatism；

When the color of intermediate value and the subimage block corresponding to the center pixel of the value of chromatism of the predetermined number When the absolute value of the difference of difference is more than or equal to default threshold value, by the subgraph corresponding to the center pixel As the value of chromatism of block is used as second aberration；

When the color of intermediate value and the subimage block corresponding to the center pixel of the value of chromatism of the predetermined number When the absolute value of the difference of difference is less than default threshold value, the intermediate value of the value of chromatism of the predetermined number is made It is second aberration.

Alternatively, the predetermined number is 5, and the subimage block of the selection is corresponding to the center pixel Subimage block subimage block adjacent up and down.

Alternatively, using equation below, R statistical values, the G systems of each subimage block are calculated respectively Evaluation and B statistical values：

R_{Sm/2, Sn/2}=median (Total_R)；

G_Sm/2,Sn/2=median (Total_G)；

B_Sm/2,Sn/2=median (Total_B)；

Wherein, R_{Sm/2, Sn/2}It is the R statistical values of subimage block each described, G_Sm/2,Sn/2It is son each described The G statistical values of image block, B_Sm/2,Sn/2It is the B statistical values of subimage block each described, median () Expression takes intermediate value to the parameter in bracket.

Alternatively, the average color difference of all subimage blocks is calculated using equation below, as the 3rd color Difference：

cd_diff_mn=abs (cd_cen-cd_mn)；

Wherein, cd_cen is second aberration, cd_mnIt is the value of chromatism of subimage block each described, NUM= Fix (M/2) × fix (N/2), is the number of the subimage block, and cd_diff is the 3rd aberration, abs () represents and the parameter in bracket is taken absolute value.

Alternatively, using equation below, according to first aberration and default denoising adjustment factor, meter Calculate the pixel value of the center pixel after removal noise：

Residual=G_Sm/2,Sn/2+cd_output；

output_pixel^·=output_pixel+residual*ratio；

Wherein, G_Sm/2,Sn/2It is the G statistical values of subimage block each described, cd_output is described first Aberration, ratio be the default denoising adjustment factor, output_pixel be remove noise after it is described in The pixel value of imago element, output_pixel is the pixel value for removing the center pixel before noise.

A kind of device for eliminating picture noise is the embodiment of the invention provides, described image includes p image Block, the size of described image block is odd number for M × N, M and N, and described device includes：

Receiving unit, is suitable to receive the image block on Bayer domain；

Choose unit, be suitable to by with the pixel of center pixel same channels centered on, take Sm × Sn sizes Subimage block, wherein：The center pixel is the center pixel of the described image block for receiving, Sm and Sn It is odd number；

Coefficient acquiring unit, is suitable to obtain the filtering weighting coefficient of each subimage block；

Filter unit, is suitable to be filtered described image block according to the filtering weighting coefficient, obtains institute The true aberration of center pixel is stated, as the first aberration；

Pixel calculation, is suitable to according to first aberration and default denoising adjustment factor, and calculating is gone The pixel value of the center pixel after except noise.

Alternatively, the filtering weighting coefficient include filtering aberration weight coefficient, the coefficient acquiring unit, Including：

First computation subunit, is suitable to, when the pixel non-green, calculate each subimage block Value of chromatism；

Second computation subunit, is suitable to the value of chromatism according to each subimage block, calculates the center The reference aberration of pixel, as the second aberration；

3rd computation subunit, is suitable to the value of chromatism and second aberration according to each subimage block, The average color difference of all subimage blocks is calculated, as the 3rd aberration；

First choice subelement, be suitably selected for filtering aberration weight coefficient corresponding to the 3rd aberration with The mapping relations of aberration difference, the aberration difference is the value of chromatism and second color of each subimage block The absolute value of poor difference；

First coefficient obtains subelement, is suitable to be reflected described according to the aberration difference of each subimage block The interval penetrated in relation, it is determined that filtering aberration weight coefficient corresponding with subimage block each described.

Alternatively, the filtering weighting coefficient also includes filtering distance weighting coefficient, and the coefficient obtains single Unit also includes：

Second selection subelement, be suitably selected for filtering distance weighting coefficient corresponding to the 3rd aberration with The mapping relations of distance, the distance is each described subimage block subgraph corresponding with the center pixel As the distance between block；

Second coefficient obtains subelement, is suitable to be closed in the mapping according to the distance of each subimage block Interval in system, it is determined that filtering distance weighting coefficient corresponding with the subimage block.

Alternatively, the coefficient acquiring unit, also includes：

4th computation subunit, is suitable to when the pixel is for green, with the center pixel described in calculating The average energy of the pixel of same channels；

3rd selection subelement, be suitably selected for filtering aberration weight coefficient corresponding to the average energy with The mapping relations of aberration difference, the aberration difference is the value of chromatism and described of each subimage block The absolute value of the difference of two aberration；

3rd coefficient obtains subelement, is suitable to be reflected described according to the aberration difference of each subimage block The interval penetrated in relation, it is determined that the filtering aberration weight coefficient corresponding with the subimage block.

Alternatively, the filtering weighting coefficient also includes filtering distance weighting coefficient, and the coefficient obtains single Unit, also includes：

4th selection subelement, be suitably selected for filtering distance weighting coefficient corresponding to the average energy with The mapping relations of distance, the distance is each described subimage block subgraph corresponding with the center pixel As the distance between block；

4th coefficient obtains subelement, is suitable to the distance value according to each subimage block in the mapping Interval in relation, it is determined that filtering distance weighting coefficient corresponding with the subimage block.

Alternatively, first computation subunit, is suitable to calculate respectively the R systems of each subimage block Evaluation, G statistical values and B statistical values, when the center pixel is for red, by each subgraph The R statistical values of block and the difference of G statistical values, as the value of chromatism of subimage block each described, when described When center pixel is for blueness, the B statistical values of each subimage block and the difference of G statistical values are made It is the value of chromatism of subimage block each described.

Alternatively, second computation subunit, is suitable to ask for the value of chromatism from subimage block each described Middle selection, the intermediate value of the value of chromatism of predetermined number, calculate the predetermined number value of chromatism intermediate value with The absolute value of the difference of the value of chromatism of the subimage block corresponding to the center pixel, when the predetermined number Value of chromatism intermediate value and the difference of the value of chromatism of the subimage block corresponding to the center pixel absolute value During more than or equal to default threshold value, using the value of chromatism of the subimage block corresponding to the center pixel as institute The second aberration is stated, when the intermediate value and the subgraph corresponding to the center pixel of the value of chromatism of the predetermined number When being less than default threshold value as the absolute value of the difference of the value of chromatism of block, by the value of chromatism of the predetermined number Intermediate value as second aberration.

Alternatively, the predetermined number is 5, and the subimage block of the selection is corresponding to the center pixel Subimage block subimage block adjacent up and down.

Alternatively, first computation subunit, is suitable for use with equation below, calculates respectively described in each The R statistical values of subimage block, G statistical values and B statistical values：

R_{Sm/2, Sn/2}=median (Total_R)；

G_Sm/2,Sn/2=median (Total_G)；

B_Sm/2,Sn/2=median (Total_B)；

Wherein, R_{Sm/2, Sn/2}It is the R statistical values of subimage block each described, G_Sm/2,Sn/2It is son each described The G statistical values of image block, B_Sm/2,Sn/2It is the B statistical values of subimage block each described, median () Expression takes intermediate value to the parameter in bracket.

Alternatively, the 3rd computation subunit, is suitable for use with equation below and is calculated all subgraphs The average color difference of block, as the 3rd aberration：

cd_diff_mn=abs (cd_cen-cd_mn)；

Wherein, cd_cen is second aberration, cd_mnIt is the value of chromatism of subimage block each described, NUM= Fix (M/2) × fix (N/2), is the number of the subimage block, and cd_diff is the 3rd aberration, abs () represents and the parameter in bracket is taken absolute value.

Alternatively, the pixel calculation, is suitable for use with equation below, according to first aberration and Default denoising adjustment factor, calculates the pixel value of the center pixel after removal noise：

Residual=G_Sm/2,Sn/2+cd_output；

output_pixel^·=output_pixel+residual*ratio；

Wherein, G_Sm/2,Sn/2It is the G statistical values of subimage block each described, cd_output is described first Aberration, ratio be the default denoising adjustment factor, output_pixel be remove noise after it is described in The pixel value of imago element, output_pixel is the pixel value for removing the center pixel before noise.

Compared with prior art, technical scheme has advantages below：

By receiving the image block on Bayer domain, the center of the described image block after removal noise is then calculated The pixel value of pixel, is that the treatment for eliminating noise is done to the image block on the Bayer domain, and Bayer domain On image procossing color interpolation treatment before, can avoid carry noise image by color interpolation The noise of script is exaggerated after treatment, the noise can be eliminated by repeatedly treatment on YUV domains, from And the amount of calculation of noise processed while denoising sound effective value is ensured, can be reduced.

Further, by asking for what is selected from the value of chromatism of each subimage block, the aberration of predetermined number The intermediate value of value, then calculates the intermediate value and the subgraph corresponding to center pixel of the value of chromatism of the predetermined number As the absolute value of the difference of the value of chromatism of block, then by between relatively more described absolute value and predetermined threshold value Relation, determines the second aberration, is the side that image block can be avoided damage to by the method for medium filtering Edge information.

Brief description of the drawings

Fig. 1 is a kind of schematic flow sheet of the method for the elimination picture noise in the embodiment of the present invention；

Fig. 2 is the schematic flow sheet of another method for eliminating picture noise in the embodiment of the present invention；

Fig. 3 is the image block on a kind of Bayer domain in the embodiment of the present invention；

Fig. 4 is a kind of aberration value array of the subimage block in the embodiment of the present invention；

Fig. 5 is a kind of structural representation of the device of the elimination picture noise in the embodiment of the present invention；

Fig. 6 is a kind of structural representation of the coefficient acquiring unit in the embodiment of the present invention；

Fig. 7 is the structural representation of another coefficient acquiring unit in the embodiment of the present invention.

Specific embodiment

Image processing process, generally includes：The image on Bayer (Bayer) domain is obtained, to described image Color interpolation treatment is carried out, the image after color interpolation treatment is then converted to the figure on RGB domains Picture, then the image on the RGB domains is converted to the image on YUV domains again.

At present, it is the color noise on removal image, to YUV after the image on YUV domains are obtained Image on domain carries out elimination noise processed.

But, noise processed is carried out with the above method, noise processed amount of calculation is very big.

Be solve problems described above, the embodiment of the present invention by receiving the image block on Bayer domain, then The pixel value of the center pixel of the described image block after removal noise is calculated, is on the Bayer domain Image block does the treatment for eliminating noise, and image procossing on Bayer domain is before color interpolation treatment, can To avoid the image for carrying noise that the noise of script is exaggerated after color interpolation is processed, on YUV domains The noise can be eliminated by repeatedly treatment, such that it is able to while denoising sound effective value is ensured, reduce The amount of calculation of noise processed.

It is understandable to enable the above objects, features and advantages of the present invention to become apparent, below in conjunction with the accompanying drawings Specific embodiment of the invention is described in detail.

The method that shown below a kind of elimination picture noise in the embodiment of the present invention is specific such as Fig. 1 institutes Show, the specific steps that methods described includes have：

S11：Receive the image block on Bayer domain.

It should be noted that any one in p image block of the described image block included by whole image It is individual, any particular/special requirement is had no, for ease of follow-up explanation, the size of described image block can be M × N, M and N are odd number.

It should be noted that M and N and do not exist absolute magnitude relationship, such as M can be more than N, M might be less that N, and M can also be equal with N.

In specific implementation, the image block on Bayer domain can be received, then to the figure on the Bayer domain As block eliminate the treatment of noise, such that it is able to avoid the noise reduction subsequent color interpolation processing Accuracy.

S12：By with the pixel of center pixel same channels centered on, take the subimage block of Sm × Sn sizes.

In an embodiment of the present invention, the center pixel is the center pixel of the described image block for receiving. It is understood that following all for the treatment of is directed to the center pixel and carries out de-noise operation, Because for each pixel, presence can make the image block of center pixel, as long as therefore it is described in Imago element has this corresponding relation with described image block.

In specific implementation, because the pixel with the center pixel same channels is to the center pixel Pixel value can be impacted, thus can by with the pixel of center pixel same channels centered on, take Sm × Sn The subimage block of size, so as to subsequent treatment.

It should be noted that Sm and Sn is odd number, and Sm and Sn does not have the limitation of magnitude relationship, Such as Sm can be more than Sn, and Sm might be less that Sn, and Sm can also be equal with Sn.

S13：The filtering weighting coefficient of each subimage block is obtained, according to the filtering weighting coefficient pair Described image block is filtered, and the true aberration of the center pixel is obtained, as the first aberration.

In specific implementation, the influence size due to each subimage block to the pixel value of the center pixel Difference, therefore the filtering weighting coefficient of each subimage block can be obtained, according to the filtering weighting system It is several that described image block is filtered, the true aberration of the center pixel can be obtained, for ease of follow-up Illustrate, the true aberration can be referred to as the first aberration.

Perception due to human eye to green be different from it is red with it is blue, and green can reflect object in itself More information, therefore for the image on Bayer domains, quantity and red pixel and the blueness of green pixel The quantity of pixel is different, therefore the specific method for calculating filtering weighting coefficient is also different.

Thus in specific implementation, the color of the center pixel can be first determined whether, when the pixel is non- When green, can be by calculating the value of chromatism of each subimage block first, then according to each The value of chromatism of subimage block, calculates the reference aberration of the center pixel, as the second aberration, Jin Ergen According to the value of chromatism and second aberration of subimage block each described, the flat of all subimage blocks is calculated Equal aberration, as the 3rd aberration, then selects the filtering aberration weight coefficient corresponding to the 3rd aberration With the mapping relations of aberration difference, the aberration difference is the value of chromatism and described second of each subimage block The absolute value of the difference of aberration, the aberration difference finally according to subimage block each described is closed in the mapping Interval in system, it is determined that filtering aberration weight coefficient corresponding with subimage block each described.

In an embodiment of the present invention, the value of chromatism of each subimage block can be calculated by following step： Calculate R statistical values, G statistical values and the B statistical values of each subimage block, Ran Hougen respectively first Color according to the center pixel is different, and the value of chromatism is calculated using distinct methods, specifically, When the center pixel is for red, G statistical values are subtracted by the R statistical values of subimage block each described, Using difference as subimage block each described value of chromatism；And when the center pixel is for blueness, will be every The B statistical values of the individual subimage block and the difference of G statistical values, as the color of subimage block each described Difference.

It is corresponding with above-described embodiment, in an embodiment of the present invention, can calculate every using formula (1) The R statistical values of the individual subimage block：

R_{Sm/2, Sn/2}=median (Total_R) (1)

The G statistical values of each subimage block can be calculated using formula (2)：

G_Sm/2,Sn/2=median (Total_G) (2)

The B statistical values of each subimage block can be calculated using formula (3)：

B_Sm/2,Sn/2=median (Total_B) (3)

Wherein, R_{Sm/2, Sn/2}It is the R statistical values of subimage block each described, G_Sm/2,Sn/2It is son each described The G statistical values of image block, B_Sm/2,Sn/2It is the B statistical values of subimage block each described, median () Expression takes intermediate value to the parameter in bracket, and Total_R represents the picture with passage with each and the center pixel All of red pixel in the corresponding subimage block of element, Total_G is represented and each and the center pixel With all of green pixel in the corresponding subimage block of the pixel of passage, Total_B is represented and each and institute Center pixel is stated with all of blue pixel in the corresponding subimage block of pixel of passage.

In specific implementation, due to can be to the center pixel with the pixel of passage with the center pixel Pixel value and noise are impacted, and consider the factor of amount of calculation, therefore can be asked for from each institute The intermediate value of the value of chromatism of the predetermined number selected in the value of chromatism of subimage block is stated, then calculates described default The absolute value of the difference of the value of chromatism of the subimage block corresponding to the intermediate value and center pixel of the value of chromatism of number, If the aberration of the intermediate value of the value of chromatism of the predetermined number and the subimage block corresponding to the center pixel When the absolute value of the difference of value is more than or equal to default threshold value, by the subgraph corresponding to the center pixel The value of chromatism of block is used as second aberration；If the intermediate value of the value of chromatism of the predetermined number with it is described in When the absolute value of the difference of the value of chromatism of the subimage block corresponding to imago element is less than default threshold value, by institute The intermediate value of value of chromatism of predetermined number is stated as second aberration.

In an embodiment of the present invention, the predetermined number is 5, during the subimage block of the selection is described The subimage block adjacent up and down of the subimage block corresponding to imago element.Certainly, the predetermined number Can also be other numerical value, those skilled in the art can according to actual needs choose the subgraph of suitable number Second aberration is calculated as the aberration of block.

With corresponding to above-described embodiment, in an embodiment of the present invention, can be calculated using formula (4) Selected from the value of chromatism of subimage block each described, the intermediate value of the value of chromatism of predetermined number：

Med_cd=median (cd_Sm/2-1,Sn/2,cd_sm/2+1,Sn/2,cd_Sm/2,Sn/2,cd_Sm/2,Sn/2-1,cd_Sm/2,Sn/2+1) (4)

Wherein, med_cd is the intermediate value of the value of chromatism of the predetermined number.

And perform formula (5) and the judgement described in formula (6)：

If cd_cen=med_cd abs (med_cd-cd_sm/2,Sn/2)<med_thr (5)

Cd_cen=cd_sm/2,Sn/2If abs (med_cd-cd_sm/2,Sn/2)>=med_thr (6)

I.e.：If the intermediate value med_cd of the value of chromatism of the predetermined number with corresponding to the center pixel The value of chromatism cd of subimage block_sm/2,Sn/2Difference absolute value be more than or equal to default threshold value med_thr when, By the value of chromatism cd of the subimage block corresponding to the center pixel_sm/2,Sn/2As second aberration cd_cen；If the intermediate value med_cd of the value of chromatism of the predetermined number with corresponding to the center pixel The value of chromatism cd of subimage block_sm/2,Sn/2Difference absolute value be less than default threshold value med_thr when, by institute The intermediate value med_cd of value of chromatism of predetermined number is stated as the second aberration cd_cen.

In an embodiment of the present invention, each subimage block can be calculated relative to described using formula (7) The deviation of the second aberration：

cd_diff_mn=abs (cd_cen-cd_mn) (7)

Then the average color difference of all subimage blocks is calculated using formula (8), as the 3rd aberration：

Wherein, cd_cen is second aberration, cd_mnIt is the value of chromatism of subimage block each described, NUM= Fix (M/2) × fix (N/2), is the number of the subimage block, and cd_diff is the 3rd aberration, abs () represents and the parameter in bracket is taken absolute value.

Had shown that by lot of experiments and research, pixel value of each described subimage block to the center pixel And the influence of noise figure lay also with the subimage block in center pixel and the center pixel distance It is relevant, therefore in an embodiment of the present invention, it is also an option that the filtering distance corresponding to the 3rd aberration The mapping relations of weight coefficient and distance, the distance is each described subimage block and the center pixel The distance between corresponding subimage block, then the distance value according to each subimage block reflected described The interval penetrated in relation, it is determined that filtering distance weighting coefficient corresponding with the subimage block.

In specific implementation, if the pixel is for green, can calculate described with the center pixel The average energy of the pixel of same channels, and then select the filtering aberration weight corresponding to the average energy The mapping relations of coefficient and aberration difference, the aberration difference be each subimage block value of chromatism with The absolute value of the difference of second aberration, then the aberration difference according to each subimage block is in institute The interval in mapping relations is stated, it is determined that filtering aberration weight coefficient corresponding with the subimage block.

Similarly, no matter what the color of the center pixel is, each subimage block is to the middle imago The influence of element is also and distance dependent, therefore in an embodiment of the present invention, can select the average energy Corresponding filtering distance weighting coefficient and the mapping relations of distance, the distance are each described subgraph The distance between block subimage block corresponding with the center pixel, and then according to each subimage block Interval of the distance value in the mapping relations, it is determined that filtering distance power corresponding with the subimage block Weight coefficient.

It should be noted that for color be green the center pixel for, can directly by with institute The green statistics value of the corresponding subimage block of center pixel is stated as the reference aberration of the center pixel, i.e., Second aberration.

S14：According to first aberration and default denoising adjustment factor, the institute after removal noise is calculated State the pixel value of center pixel.

In specific implementation, can be according to the current Quality of image and the relativeness of aimed quality, in advance Rational denoising adjustment factor is set, and the true aberration for then being calculated according to S13 and the denoising are adjusted Coefficient, calculates the pixel value of the center pixel after removal noise, so as to complete at the elimination noise The step of reason.

In an embodiment of the present invention, formula (9) can be used, according to first aberration and default Denoising adjustment factor, calculates the pixel value of the center pixel after removal noise：

Residual=G_Sm/2,Sn/2+cd_output (9)

output_pixel^·=output_pixel+residual*ratio (10)

Wherein, G_Sm/2,Sn/2It is the G statistical values of subimage block each described, cd_output is described first Aberration, ratio is the default denoising adjustment factor, output_pixel^·For removal noise after it is described in The pixel value of imago element, output_pixel is the pixel value for removing the center pixel before noise.Need Illustrate, those skilled in the art can according to actual needs set the denoising adjustment factor.

To cause that those skilled in the art more fully understand and realize the present invention, it is also provided below another Specific steps included by methods described are introduced by the method for eliminating picture noise below with reference to Fig. 2：

S21：Judge whether center pixel is green.

In an embodiment of the present invention, the image block on the Bayer domain of acquisition is as shown in figure 3, described image The size of block is 11 × 11, B₅₅As described center pixel, R, G, B correspondence three kinds of face of red, green, blue Color pixel, the subscript of the pixel represents pixel present position in image block.0~a represents 0 to 10. , it is worthwhile to note that the size 11 × 11 of described image block is not necessary condition, those skilled in the art can be with Described image block is zoomed in and out according to the actual requirements, for example, described image block can be contracted to 7 × 7, also may be used 15 × 15 are extended to by described image block.

When the center pixel non-green, S22 is performed, when the center pixel is for green, perform S24.

S22：Colorimetry.

In an embodiment of the present invention, in image block it is all be currently needed for treatment, that is to say institute The pixel of center pixel B55 same channels is stated, around it 3 × 3 image block can be taken as subimage block, In such as Fig. 3, the gray area centered on pixel B 11, mark is centered on pixel B 11 3 × 3 subimage block blocks, and the gray area centered on pixel B 55, mark is with pixel B 55 Centered on 3 × 3 subimage blocks, be one of subimage block.It is 11 × 11 for the size shown in Fig. 3 Image block, can have the subimage block of 25 3 × 3.

In order to know the colour difference information of each subimage block, can to the R of each 3 × 3 subimage block, G, B statistical values are individually calculated, and obtain one group of R_mn, G_mn, B_mnValue, wherein subscript m and n are rope Draw value, because the image block shown in Fig. 3 can have 25 subimage blocks, therefore the value model of m and n Enclose is [0,4].For ease of understanding, to illustrate as a example by 3 × 3 subimage blocks that can be centered on B55 The calculation procedure of R, G, B statistical value is stated, it is specific as follows：

R₂₂=median (R44, R46, R64, R66) (11)

G₂₂=median (G45, G54, G56, G65) (12)

B₂₂=B55 (13)

Wherein median () is median filter, that is, represent and take intermediate value to the parameter inside bracket.Can be with Understand, the computational methods of R, G, B statistical value of remaining each 3 × 3 subimage block can be with such Push away, will not be repeated here.

In order to know the chromaticity distortion with the pixel of the center pixel same channels, then can calculate every The value of chromatism of individual 3 × 3 subimage block, when the center pixel of the subimage block is for red, can use Equation below calculates the value of chromatism of each 3 × 3 subimage block：

Cdmn=Rmn-Gmn (14)

When the subimage block center pixel for it is blue when, can using equation below calculate each 3 × 3 The value of chromatism of subimage block：

Cdmn=Bmn-Gmn (15)

It is B for the center pixel described in Fig. 3₅₅Image block, each can be calculated using formula (15) The value of chromatism of the subimage block.So, it is possible to obtain 25 cd values.

S23：Filtering weighting coefficient is asked for, bilateral filtering is carried out to center pixel.

25 value of chromatism cdmn of the subimage block are constituted the array of new 5 × 5, specifically may be used With as shown in Figure 4.In order to further eliminate picture noise, retain the marginal information of image and consider The factor of amount of calculation, can carry out cross medium filtering to the center pixel, that is, take the middle imago The value of chromatism cd of plain adjacent subimage block up and down₁₂,cd₂₁,cd₂₃,cd₃₂Intermediate value, it is specific such as formula (16) shown in：

Med_cd=median (cd₁₂,cd₂₁,cd₂₂,cd₂₃,cd₃₂) (16)

Then judged by a predetermined threshold value med_thr, if made using the intermediate value aberration med_cd It is the center pixel cd₂₂Reference aberration cd_cen, i.e. the second aberration.Specifically, as abs (med_cd -cd₂₂)<During med_thr, the relation shown in formula (17) can be obtained：

Cd_cen=med_cd (17)

As abs (med_cd-cd₂₂)>During=med_thr, the relation shown in formula (18) can be obtained：

Cd_cen=cd₂₂ (18)

If that is, the difference of the intermediate value aberration med_cd and aberration cd22 of the center pixel Absolute value be less than predetermined threshold value med_thr when, using the intermediate value med_cd as second aberration cd_cen；Conversely, using the aberration cd22 of the center pixel as the second aberration cd_cen.

In order to more accurately eliminate the aberration of the center pixel, the center pixel can be carried out bilateral Filtering, the bilateral filtering considers the spatial domain of the center pixel and the difference of codomain simultaneously, correspondence Weight can be utilized respectively filtering distance weighting coefficient S igma_dis and filtering aberration weight coefficient Sigma_range represents that the specific calculating process of two above weight coefficient is as follows：

Each subimage block is calculated with the deviation of the second aberration cd_cen first with formula (19) Absolute value：

cd_diff_mn=abs (cd_cen-cd_mn) (19)

Then the average color difference of all subimage blocks is calculated using formula (20), as the 3rd aberration：

And then the difference cd_diff of aberration can be divided into by scope not of uniform size by threshold value, to every One scope carries out bilateral filtering using a set of Sigma_dis and Sigma_range weight coefficients for presetting. That is, the filtering aberration weight coefficient corresponding to the 3rd aberration cd_diff can be selected poor with aberration The mapping relations of value, filtering distance weighting coefficient and distance, wherein：The aberration difference is each subgraph The value of chromatism of picture block and the absolute value of the difference of second aberration, the distance is each described subgraph The distance between block subimage block corresponding with the center pixel.

When bilateral filtering is carried out, for each subimage block, can be by calculating and the center Aberration cd₂₂Between distance, filtering distance weighting coefficient S igma_dis is selected in the mapping relations；Use abs(cd_mn- cen_cd) calculate aberration difference degree, in the mapping relations select filtering aberration weight Coefficient S igma_range, bilateral filtering is carried out followed by the filtering weighting coefficient to the center pixel, The true aberration cd_output of the center pixel is finally given, for convenience of description, can be referred to as One aberration.

S24：Filtering weighting coefficient is asked for, bilateral filtering is carried out to center pixel.

In specific implementation, if the center pixel is green, directly the center pixel can be entered The treatment of row bilateral filtering.First with described in S22, the R of each 3 × 3 sub-block can be respectively calculated_mn, G_mn, B_mnValue, obtains 25 R, G, B statistical values.

But because human eye is to the perception characteristic of green pixel, adopted different from red and blue bilateral filtering Obtained with cross medium filtering and refer to aberration cd_cen, for the pixel that center pixel is green, can be direct By the green statistics value G of the center pixel_2,2As the reference aberration cd_cen.

Then, it is each and the center pixel phase by the average energy mean_g of green component With the pixel selection weight of passage, the average energy mean_g can be calculated using formula (21):

That is, can select filtering aberration weight coefficient corresponding to the average energy mean_g with The mapping relations of aberration difference, filtering distance weighting coefficient and distance, wherein：The aberration difference is every The absolute value of the value of chromatism of individual subimage block and the difference of second aberration, the distance is for described in each The distance between subimage block subimage block corresponding with the center pixel.

Then mean_g can be divided into by different scopes by threshold value, each scope is used a set of pre- If good filtering distance weighting coefficient S igma_dis and filtering aberration weight coefficient Sigma_range, is carried out double Side filters.When bilateral filtering is carried out, by the distance between calculating with center pixel, selection filtering away from From weight coefficient Sigma_dis；Use abs (g_cen-G_mn) calculate aberration difference degree, selection filtering Aberration weight coefficient Sigma_range, obtains final aberration output_pixel.

S25：Using first aberration and default denoising adjustment factor, calculate described after removal noise The pixel value of center pixel.

In an embodiment of the present invention, it is possible to use formula (22) and (23) remove noise to calculate The pixel value output_pixel of the center pixel afterwards^·：

Residual=G₂₂+cd_output (22)

output_pixel^·=output_pixel+residual*ratio (23)

Wherein：output_pixel^·For removal ratio is default denoising adjustment factor, cd_output is institute State the first aberration, G₂₂It is the green statistics value of the corresponding subimage block of the center pixel.

To cause that those skilled in the art more fully understand and realize the present invention, it is also provided below can be real The device of the method for existing above-mentioned elimination picture noise, as shown in Figure 5.For convenience of description, if described image Including p image block, the size of each image block can be odd number, the dress for M × N, M and N Putting can include receiving unit 51, choose unit 52, coefficient acquiring unit 53, filter unit 54 and picture Plain computing unit 55, wherein：

The receiving unit 51, is suitable to receive the image block on Bayer domain；

It is described selection unit 52, be suitable to by with the pixel of center pixel same channels centered on, take Sm × Sn The subimage block of size, wherein：The center pixel is the center pixel of the described image block for receiving, Sm Odd number is with Sn；

The coefficient acquiring unit 53, is suitable to obtain the filtering weighting coefficient of each subimage block；

The filter unit 54, is suitable to be filtered described image block according to the filtering weighting coefficient, The true aberration of the center pixel is obtained, as the first aberration；

The pixel calculation 55, is suitable to according to first aberration and default denoising adjustment factor, Calculate the pixel value of the center pixel after removal noise.

To cause that those skilled in the art more fully understand and realize the present invention, the present invention also provided below A kind of structural representation of the coefficient acquiring unit in embodiment, specifically may be referred to Fig. 6, the filtering power Weight coefficient includes filtering aberration weight coefficient, and the coefficient acquiring unit 54 can include that the first calculating is single First 61, second computation subunit 62, the 3rd computation subunit 63, first choice subelement 64 and first Coefficient obtains subelement 65, wherein：

First computation subunit 61, is suitable to, when the pixel non-green, calculate each described subgraph As the value of chromatism of block；

Second computation subunit 62, is suitable to the value of chromatism according to each subimage block, calculates institute The reference aberration of center pixel is stated, as the second aberration；

3rd computation subunit 63, is suitable to the value of chromatism and described according to each subimage block Two aberration, are calculated the average color difference of all subimage blocks, used as the 3rd aberration；

The first choice subelement 64, is suitably selected for the filtering aberration weight corresponding to the 3rd aberration The mapping relations of coefficient and aberration difference, the aberration difference be the value of chromatism of each subimage block with it is described The absolute value of the difference of the second aberration；

First coefficient obtains subelement 65, is suitable to be existed according to the aberration difference of each subimage block Interval in the mapping relations, it is determined that filtering aberration weight coefficient corresponding with subimage block each described.

In specific implementation, the filtering weighting coefficient also includes filtering distance weighting coefficient, the coefficient Acquiring unit 54 can also include that the second selection subelement 66 and second coefficient obtains subelement 67, wherein：

The second selection subelement 66, is suitably selected for the filtering distance weighting corresponding to the 3rd aberration The mapping relations of coefficient and distance, the distance is that each described subimage block is corresponding with the center pixel The distance between subimage block；

Second coefficient obtains subelement 67, is suitable to the distance according to each subimage block described Interval in mapping relations, it is determined that filtering distance weighting coefficient corresponding with the subimage block.

In specific implementation, first computation subunit 61 is suitable to calculate each described subgraph respectively The R statistical values of block, G statistical values and B statistical values, when the center pixel for it is red when, by each institute The R statistical values of subimage block and the difference of G statistical values are stated, as the value of chromatism of subimage block each described, When the center pixel is for blueness, by the B statistical values of each subimage block and the difference of G statistical values Value, as the value of chromatism of subimage block each described.

In specific implementation, second computation subunit 62 is suitable to ask for from subimage block each described Value of chromatism in select, the intermediate value of the value of chromatism of predetermined number calculates the value of chromatism of the predetermined number Intermediate value and the difference of the value of chromatism of the subimage block corresponding to the center pixel absolute value, when described The difference of the intermediate value of the value of chromatism of predetermined number and the value of chromatism of the subimage block corresponding to the center pixel Absolute value be more than or equal to default threshold value when, by the aberration of the subimage block corresponding to the center pixel It is worth as second aberration, when the intermediate value of the value of chromatism of the predetermined number is right with center pixel institute When the absolute value of the difference of the value of chromatism of the subimage block answered is less than default threshold value, by the predetermined number Value of chromatism intermediate value as second aberration.

In specific implementation, the predetermined number is 5, and the subimage block of the selection is the center pixel The subimage block adjacent up and down of corresponding subimage block.

In specific implementation, first computation subunit 61 is suitable for use with equation below, calculates respectively The R statistical values of each subimage block, G statistical values and B statistical values：

R_{Sm/2, Sn/2}=median (Total_R)；

G_Sm/2,Sn/2=median (Total_G)；

B_Sm/2,Sn/2=median (Total_B)；

Wherein, R_{Sm/2, Sn/2}It is the R statistical values of subimage block each described, G_Sm/2,Sn/2It is son each described The G statistical values of image block, B_Sm/2,Sn/2It is the B statistical values of subimage block each described, median () Expression takes intermediate value to the parameter in bracket.

In specific implementation, the 3rd computation subunit 63 is suitable for use with equation below and is calculated institute There is the average color difference of subimage block, as the 3rd aberration：

cd_diff_mn=abs (cd_cen-cd_mn)；

Wherein, cd_cen is second aberration, cd_mnIt is the value of chromatism of subimage block each described, NUM= Fix (M/2) × fix (N/2), is the number of the subimage block, and cd_diff is the 3rd aberration, abs () represents and the parameter in bracket is taken absolute value.

In specific implementation, the pixel calculation is suitable for use with equation below, according to described first Aberration and default denoising adjustment factor, calculate the pixel value of the center pixel after removal noise：

Residual=G_Sm/2,Sn/2+cd_output；

output_pixel^·=output_pixel+residual*ratio；

Wherein, G_Sm/2,Sn/2It is the G statistical values of subimage block each described, cd_output is described first Aberration, ratio be the default denoising adjustment factor, output_pixel be remove noise after it is described in The pixel value of imago element, output_pixel is the pixel value for removing the center pixel before noise.

To cause that those skilled in the art more fully understand and realize the present invention, the present invention also provided below Another coefficient acquiring unit in embodiment, may be referred to Fig. 7, and the filtering weighting coefficient includes filtering color Difference weight coefficient, the coefficient acquiring unit 54 can include that the 4th computation subunit the 71, the 3rd selects son The coefficient of unit 72 and the 3rd obtains subelement 73, wherein：

4th computation subunit 71, is suitable to when the pixel is for green, calculate it is described with it is described in The average energy of the pixel of imago element same channels；

The 3rd selection subelement 72, is suitably selected for the filtering aberration weight corresponding to the average energy The mapping relations of coefficient and aberration difference, the aberration difference be each subimage block value of chromatism with The absolute value of the difference of second aberration；

3rd coefficient obtains subelement 73, is suitable to be existed according to the aberration difference of each subimage block Interval in the mapping relations, it is determined that the filtering aberration weight coefficient corresponding with the subimage block.

In specific implementation, the filtering weighting coefficient also includes filtering distance weighting coefficient, correspondingly, The coefficient acquiring unit 54, can also include that the 4th selection subelement 74 and the 4th coefficient obtains subelement 75, wherein：

The 4th selection subelement 74, is suitably selected for the filtering distance weighting corresponding to the average energy The mapping relations of coefficient and distance, the distance is that each described subimage block is corresponding with the center pixel The distance between subimage block；

4th coefficient obtains subelement 75, is suitable to the distance value according to each subimage block in institute The interval in mapping relations is stated, it is determined that filtering distance weighting coefficient corresponding with the subimage block.

One of ordinary skill in the art will appreciate that all or part of step in the various methods of above-described embodiment Rapid to can be by program to instruct the hardware of correlation to complete, the program can be stored in can with computer Read in storage medium, storage medium can include：ROM, RAM, disk or CD etc..

Although present disclosure is as above, the present invention is not limited to this.Any those skilled in the art, Without departing from the spirit and scope of the present invention, can make various changes or modifications, therefore guarantor of the invention Shield scope should be defined by claim limited range.

Claims (22)

1. it is a kind of eliminate picture noise method, described image include p image block, the size of described image block It is that M × N, M and N are odd number, it is characterised in that methods described includes：

Receive the image block on Bayer domain；

By with the pixel of center pixel same channels centered on, take the subimage block of Sm × Sn sizes, wherein： The center pixel is the center pixel of the described image block for receiving, and Sm and Sn is odd number；

The filtering weighting coefficient of each subimage block is obtained, according to the filtering weighting coefficient to the figure As block is filtered, the true aberration of the center pixel is obtained, as the first aberration；

According to first aberration and default denoising adjustment factor, the described middle imago after removal noise is calculated The pixel value of element.

2. it is according to claim 1 eliminate picture noise method, it is characterised in that the filtering weighting Coefficient includes filtering aberration weight coefficient, the filtering weighting coefficient of the acquisition each subimage block, Including：

When the pixel non-green, the value of chromatism of each subimage block is calculated；

According to the value of chromatism of each subimage block, the reference aberration of the center pixel is calculated, as the Two aberration；

Value of chromatism and second aberration according to each subimage block, are calculated all subgraphs As the average color difference of block, as the 3rd aberration；

Select the mapping relations of the filtering aberration weight coefficient and aberration difference corresponding to the 3rd aberration, institute State the absolute value of value of chromatism that aberration difference is each subimage block and the difference of second aberration；

The interval of aberration difference according to each subimage block in the mapping relations, it is determined that and each The subimage block is corresponding to filter aberration weight coefficient.

3. it is according to claim 2 eliminate picture noise method, it is characterised in that the filtering weighting Coefficient also includes filtering distance weighting coefficient, also includes：

Select the mapping relations of filtering distance weighting coefficient and the distance corresponding to the 3rd aberration, it is described away from From being the distance between subimage block subimage block corresponding with the center pixel each described；

The interval of distance according to each subimage block in the mapping relations, it is determined that with the subgraph As the corresponding filtering distance weighting coefficient of block.

4. it is according to claim 2 eliminate picture noise method, it is characterised in that the acquisition each The filtering weighting coefficient of the subimage block, also includes：

When the pixel is for green, the average energy of the pixel with center pixel same channels is calculated；

Select the mapping relations of the filtering aberration weight coefficient and aberration difference corresponding to the average energy, institute State the absolute value of value of chromatism that aberration difference is each subimage block and the difference of second aberration；

The interval of aberration difference according to each subimage block in the mapping relations, it is determined that with it is described The corresponding filtering aberration weight coefficient of subimage block.

5. it is according to claim 4 eliminate picture noise method, it is characterised in that the filtering weighting Coefficient also includes filtering distance weighting coefficient, and methods described also includes：

Select the mapping relations of filtering distance weighting coefficient and the distance corresponding to the average energy, it is described away from From being the distance between subimage block subimage block corresponding with the center pixel each described；

The interval of distance value according to each subimage block in the mapping relations, it is determined that with the son The corresponding filtering distance weighting coefficient of image block.

6. it is according to claim 2 eliminate picture noise method, it is characterised in that the calculating each The value of chromatism of subimage block, including：

R statistical values, G statistical values and the B statistical values of each subimage block are calculated respectively；

When the center pixel is for red, by the R statistical values of each subimage block and G statistical values Difference, as the value of chromatism of subimage block each described；

When the center pixel is for blueness, by the B statistical values of each subimage block and G statistical values Difference, as the value of chromatism of subimage block each described.

7. it is according to claim 6 eliminate picture noise method, it is characterised in that it is described according to each The value of chromatism of the subimage block, calculates the reference aberration of the center pixel, as the second aberration, Including：

Ask for the intermediate value of the value of chromatism of the predetermined number selected from the value of chromatism of subimage block each described；

Calculate the color of intermediate value and the subimage block corresponding to the center pixel of the value of chromatism of the predetermined number The absolute value of the difference of difference；

When the aberration of intermediate value and the subimage block corresponding to the center pixel of the value of chromatism of the predetermined number When the absolute value of the difference of value is more than or equal to default threshold value, by the subgraph corresponding to the center pixel As the value of chromatism of block is used as second aberration；

When the aberration of intermediate value and the subimage block corresponding to the center pixel of the value of chromatism of the predetermined number When the absolute value of the difference of value is less than default threshold value, the intermediate value of the value of chromatism of the predetermined number is made It is second aberration.

8. it is according to claim 7 eliminate picture noise method, it is characterised in that the predetermined number Be 5, the subimage block of the selection be subimage block corresponding to the center pixel up and down Adjacent subimage block.

9. it is according to claim 7 eliminate picture noise method, it is characterised in that use equation below, R statistical values, G statistical values and the B statistical values of each subimage block are calculated respectively：

R_{Sm/2, Sn/2}=median (Total_R)；

G_Sm/2,Sn/2=median (Total_G)；

B_Sm/2,Sn/2=median (Total_B)；

Wherein, R_{Sm/2, Sn/2}It is the R statistical values of subimage block each described, G_Sm/2,Sn/2It is son each described The G statistical values of image block, B_Sm/2,Sn/2It is the B statistical values of subimage block each described, median () Expression takes intermediate value to the parameter in bracket.

10. it is according to claim 2 eliminate picture noise method, it is characterised in that use equation below The average color difference of all subimage blocks is calculated, as the 3rd aberration：

cd_diff_mn=abs (cd_cen-cd_mn)；

$cd\_diff=\frac{1}{NUM}{\&Sigma;}_{m=0}^{Sm}{\&Sigma;}_{n=0}^{Sn}cd\_{\mathrm{diff}}_{mn};$

Wherein, cd_cen is second aberration, cd_mnIt is the value of chromatism of subimage block each described, NUM= Fix (M/2) × fix (N/2), is the number of the subimage block, and cd_diff is the 3rd aberration, Abs () is represented and the parameter in bracket is taken absolute value.

11. methods for eliminating picture noise according to claim 1, it is characterised in that use equation below, According to first aberration and default denoising adjustment factor, the described middle imago after removal noise is calculated The pixel value of element：

Residual=G_Sm/2,Sn/2+cd_output；

Output_pixel=output_pixel+residual*ratio；

Wherein, G_Sm/2,Sn/2It is the G statistical values of subimage block each described, cd_output is first aberration, Ratio is the default denoising adjustment factor, output_pixel^·It is the described middle imago after removal noise The pixel value of element, output_pixel is the pixel value for removing the center pixel before noise.

A kind of 12. devices for eliminating picture noise, described image includes p image block, and the size of described image block is M × N, M and N are odd number, it is characterised in that including：

Receiving unit, is suitable to receive the image block on Bayer domain；

Choose unit, be suitable to by with the pixel of center pixel same channels centered on, take the son of Sm × Sn sizes Image block, wherein：The center pixel is the center pixel of the described image block for receiving, Sm and Sn It is odd number；

Coefficient acquiring unit, is suitable to obtain the filtering weighting coefficient of each subimage block；

Filter unit, is suitable to be filtered described image block according to the filtering weighting coefficient, obtains described The true aberration of center pixel, as the first aberration；

Pixel calculation, is suitable to, according to first aberration and default denoising adjustment factor, calculate removal The pixel value of the center pixel after noise.

13. devices for eliminating picture noise according to claim 12, it is characterised in that the filtering weighting Coefficient include filtering aberration weight coefficient, the coefficient acquiring unit, including：

First computation subunit, is suitable to, when the pixel non-green, calculate the color of each subimage block Difference；

Second computation subunit, is suitable to the value of chromatism according to each subimage block, calculates the middle imago The reference aberration of element, as the second aberration；

3rd computation subunit, is suitable to the value of chromatism and second aberration according to each subimage block, The average color difference of all subimage blocks is calculated, as the 3rd aberration；

First choice subelement, is suitably selected for filtering aberration weight coefficient and color corresponding to the 3rd aberration The mapping relations of difference difference, the aberration difference is the value of chromatism and second color of each subimage block The absolute value of poor difference；

First coefficient obtains subelement, is suitable to the aberration difference according to each subimage block in the mapping Interval in relation, it is determined that filtering aberration weight coefficient corresponding with subimage block each described.

14. devices for eliminating picture noise according to claim 13, it is characterised in that the filtering weighting Coefficient also includes filtering distance weighting coefficient, and the coefficient acquiring unit also includes：

Second selection subelement, be suitably selected for filtering distance weighting coefficient corresponding to the 3rd aberration with away from From mapping relations, the distance be each described subimage block subgraph corresponding with the center pixel As the distance between block；

Second coefficient obtains subelement, is suitable to the distance according to each subimage block in the mapping relations In interval, it is determined that it is corresponding with the subimage block filtering distance weighting coefficient.

15. devices for eliminating picture noise according to claim 13, it is characterised in that the coefficient is obtained Unit, also includes：

4th computation subunit, is suitable to when the pixel is for green, with the center pixel phase described in calculating With the average energy of the pixel of passage；

3rd selection subelement, is suitably selected for filtering aberration weight coefficient and color corresponding to the average energy The mapping relations of difference difference, the aberration difference is the value of chromatism and described the of each subimage block The absolute value of the difference of two aberration；

3rd coefficient obtains subelement, is suitable to the aberration difference according to each subimage block in the mapping Interval in relation, it is determined that the filtering aberration weight coefficient corresponding with the subimage block.

16. devices for eliminating picture noise according to claim 15, it is characterised in that the filtering weighting Coefficient also includes filtering distance weighting coefficient, and the coefficient acquiring unit also includes：

4th selection subelement, be suitably selected for filtering distance weighting coefficient corresponding to the average energy with away from From mapping relations, the distance be each described subimage block subgraph corresponding with the center pixel As the distance between block；

4th coefficient obtains subelement, is suitable to be closed in the mapping according to the distance value of each subimage block Interval in system, it is determined that filtering distance weighting coefficient corresponding with the subimage block.

17. devices for eliminating picture noise according to claim 13, it is characterised in that described first calculates Subelement, is suitable to calculate the R statistical values of each subimage block, G statistical values and B statistics respectively Value, when the center pixel is for red, the R statistical values of each subimage block and G is counted The difference of value, as the value of chromatism of subimage block each described, when the center pixel is for blueness, By the B statistical values of each subimage block and the difference of G statistical values, as subgraph each described The value of chromatism of block.

18. devices for eliminating picture noise according to claim 17, it is characterised in that described second calculates Subelement, is suitable to ask for what is selected from the value of chromatism of subimage block each described, the color of predetermined number The intermediate value of difference, calculate the predetermined number value of chromatism intermediate value with corresponding to the center pixel The absolute value of the difference of the value of chromatism of subimage block, intermediate value and institute when the value of chromatism of the predetermined number The absolute value for stating the difference of the value of chromatism of the subimage block corresponding to center pixel is more than or equal to default threshold During value, using the value of chromatism of the subimage block corresponding to the center pixel as second aberration, when The value of chromatism of the intermediate value of the value of chromatism of the predetermined number and the subimage block corresponding to the center pixel Difference absolute value be less than default threshold value when, using the intermediate value of the value of chromatism of the predetermined number as Second aberration.

19. devices for eliminating picture noise according to claim 18, it is characterised in that the predetermined number Be 5, the subimage block of the selection be subimage block corresponding to the center pixel up and down Adjacent subimage block.

20. devices for eliminating picture noise according to claim 18, it is characterised in that described first calculates Subelement, is suitable for use with equation below, and R statistical values, the G of each subimage block are calculated respectively Statistical value and B statistical values：

R_{Sm/2, Sn/2}=median (Total_R)；

G_Sm/2,Sn/2=median (Total_G)；

B_Sm/2,Sn/2=median (Total_B)；

Wherein, R_{Sm/2, Sn/2}It is the R statistical values of subimage block each described, G_Sm/2,Sn/2It is son each described The G statistical values of image block, B_Sm/2,Sn/2It is the B statistical values of subimage block each described, median () Expression takes intermediate value to the parameter in bracket.

21. devices for eliminating picture noise according to claim 14, it is characterised in that the described 3rd calculates Subelement, is suitable for use with the average color difference that equation below is calculated all subimage blocks, as the 3rd Aberration：

cd_diff_mn=abs (cd_cen-cd_mn)；

$cd\_diff=\frac{1}{NUM}{\&Sigma;}_{m=0}^{Sm}{\&Sigma;}_{n=0}^{Sn}cd\_{\mathrm{diff}}_{mn};$

Wherein, cd_cen is second aberration, cd_mnIt is the value of chromatism of subimage block each described, NUM= Fix (M/2) × fix (N/2), is the number of the subimage block, and cd_diff is the 3rd aberration, Abs () is represented and the parameter in bracket is taken absolute value.

22. devices for eliminating picture noise according to claim 13, it is characterised in that the pixel is calculated Unit, is suitable for use with equation below, according to first aberration and default denoising adjustment factor, meter Calculate the pixel value of the center pixel after removal noise：

Residual=G_Sm/2,Sn/2+cd_output；

Output_pixel=output_pixel+residual*ratio；

Wherein, G_Sm/2,Sn/2It is the G statistical values of subimage block each described, cd_output is first aberration, Ratio is the default denoising adjustment factor, output_pixel^·It is the described middle imago after removal noise The pixel value of element, output_pixel is the pixel value for removing the center pixel before noise.