CN101271196A - Lens shade correction index confirming method, lens shade emendation method and device - Google Patents

Abstract

The invention discloses a lens shadow correction coefficient confirming method, a lens shadow correcting method and a device thereof, which comprise: the reference correction coefficient of each pixel point is obtained according to a training image; the reference correction coefficient forms a two-dimensional correction curved surface at a pixel point coordinate plane; the binary high-ordered power multinomial of the pixel point coordinate is used for fitting the two-dimensional correction curved surface; the lens shadow correction coefficient corresponding to each pixel point coordinate is calculated according to the multinomial, and an image shot by the lens is processed for lens shadow correction disposal according to the confirmed lens shadow correction coefficient. The method of the invention is suitable for the lens shadow correction for the image shot by various types of lenses, which has good image correction effect and is simple to be realized.

Description

Camera lens shadow correction coefficient is determined method, camera lens shadow correction method and device

Technical field

The present invention relates to Flame Image Process, relate in particular in the image camera lens shadow correction coefficient and determine method, camera lens shadow correction method and device.

Background technology

The camera lens shade is because the inconsistent imaging plane inconsistent phenomenon of exposing that causes of camera lens light transmission, is embodied in the middle of the image that obtains dark around bright.In the existing method, the general bosom of all supposing camera lens is bright spot, darkness symmetry around the camera lens, when carrying out the camera lens Shadows Processing, think that the ratio of image brightness all around and center brightness meets the relation of cosine 4 powers of angle between optical axis and imaging point and the optical center line, carries out the camera lens shadow correction according to this shadow model to image.Adopt above-mentioned hypothesis and shadow model that image is carried out the camera lens shadow correction and handle, have certain effect for the captured image of the smaller camera lens of camera lens shade.But treatment effect from some real images, a lot of camera lenses and not in full conformity with this hypothesis, therefore be that a class shadow model of representative does not have versatility with cosine 4 powers, this model for some hatching effects seriously and bright spot be not camera lens poor effect at the imaging plane center.

Summary of the invention

The invention provides a kind of be applicable to all kinds camera lens, calibration result preferably camera lens shadow correction coefficient determine method, camera lens shadow correction method and device.

Camera lens shadow correction coefficient provided by the invention is determined method, comprising:

Obtain the reference correction coefficient of each pixel according to training image;

Generate 2DC curved surface on the pixel coordinate plane by described with reference to correction coefficient;

Adopt the described 2DC curved surface of binary high order power fitting of a polynomial of pixel coordinate;

Go out the camera lens shadow correction coefficient corresponding according to described polynomial computation with each pixel coordinate.

Describedly obtain the reference correction coefficient of each pixel, comprising according to training image:

Obtain the standard pixel value of each pixel of training image; And

Obtain true pixel values with each pixel of the photographic images that obtains behind the described training image of lens shooting;

Calculate the ratio of true pixel values and its standard pixel value of each pixel, obtain the reference correction coefficient of each pixel.

The described 2DC curved surface of binary high order power fitting of a polynomial of described employing pixel coordinate comprises:

Adopt the described 2DC curved surface of 2-10 power fitting of a polynomial of pixel horizontal ordinate and ordinate.

The described 2DC curved surface of binary high order power fitting of a polynomial of described employing pixel coordinate comprises:

Adopt the described 2DC curved surface of 6 power fitting of a polynomials of pixel horizontal ordinate and ordinate.

The present invention also provides a kind of camera lens shadow correction method, comprising:

Import image to be corrected;

According to pixels point coordinate obtains the pixel value of the corresponding pixel points of current pixel point coordinate in proper order, and obtains the camera lens shadow correction coefficient corresponding with the current pixel point coordinate; Wherein, definite method of described camera lens shadow correction coefficient is: the reference correction coefficient that obtains each pixel according to training image, generate 2DC curved surface on the pixel coordinate plane by described with reference to correction coefficient, adopt the described 2DC curved surface of binary high order power fitting of a polynomial of pixel coordinate, go out the camera lens shadow correction coefficient corresponding with each pixel coordinate according to described polynomial computation;

Multiply by the described camera lens shadow correction coefficient that obtains with the described pixel value that obtains, obtain proofreading and correct back image and output.

Above-mentioned camera lens shadow correction method of the present invention also comprises: from the camera lens shadow correction coefficient of each pixel coordinate correspondence of calculating, extract the camera lens shadow correction coefficient and the storage of part pixel coordinate correspondence by the decimation rule of setting in advance;

Describedly obtain the camera lens shadow correction coefficient corresponding, comprising with the current pixel point coordinate:

According to the camera lens shadow correction coefficient of storage and the interpolation algorithm of setting, calculate the camera lens shadow correction coefficient of current pixel point coordinate correspondence.

Described camera lens shadow correction coefficient and storage by the decimation rule extraction part pixel coordinate correspondence of setting, comprising: every interval plurality of pixels point extracts the camera lens shadow correction coefficient of a pixel coordinate correspondence and stores;

The interpolation algorithm of described setting is the cube interpolation algorithm.

The present invention also provides a kind of camera lens shadow correction coefficient to determine device, comprising:

With reference to the correction coefficient acquiring unit, be used for obtaining the reference correction coefficient and the output of each pixel according to training image;

2DC curved surface generation unit is used to receive described described with reference to correction coefficient with reference to the output of correction coefficient acquiring unit, generates the 2DC curved surface on the pixel coordinate plane;

The match unit, the 2DC curved surface that is used to adopt the described 2DC curved surface of the binary high order power fitting of a polynomial generation unit of pixel coordinate to generate;

Camera lens shadow correction coefficient calculation unit is used for the polynomial expression determined according to described match unit, calculates the camera lens shadow correction coefficient corresponding with each pixel coordinate.

Described with reference to the correction coefficient acquiring unit, comprising:

Obtain subelement, be used to obtain the standard pixel value of each pixel of training image, and with the true pixel values of each pixel of the photographic images that obtains behind the described training image of lens shooting;

Computation subunit is used to calculate the ratio of true pixel values and its standard pixel value of each pixel, obtains the reference correction coefficient and the output of each pixel.

The present invention also provides a kind of camera lens shadow correction device, comprising:

The image input block is used to receive image to be corrected;

The pixel value acquiring unit is used for the image to be corrected according to described image input block reception, and according to pixels point coordinate obtains the pixel value and the output of the corresponding pixel points of current pixel point coordinate in proper order;

Camera lens shadow correction coefficient determining unit, be used for obtaining the reference correction coefficient of each pixel according to training image, generate 2DC curved surface on the pixel coordinate plane by described with reference to correction coefficient, adopt the described 2DC curved surface of binary high order power fitting of a polynomial of pixel coordinate, go out the camera lens shadow correction coefficient corresponding with each pixel coordinate according to described polynomial computation;

Camera lens shadow correction coefficient output unit is used for the camera lens shadow correction coefficient corresponding with each pixel coordinate that calculate according to described camera lens shadow correction coefficient determining unit, the camera lens shadow correction coefficient of output current pixel point coordinate correspondence;

Correcting unit is used for the described pixel value with described pixel value acquiring unit output, multiply by the camera lens shadow correction coefficient of the respective pixel point coordinate of described camera lens shadow correction coefficient output unit output, generates to proofread and correct the back image;

The image output unit is used for image behind the output calibration.

Described camera lens shadow correction coefficient output unit comprises:

Extract storing sub-units, be used for the camera lens shadow correction coefficient of each pixel coordinate correspondence of calculating from described camera lens shadow correction coefficient determining unit, extract the camera lens shadow correction coefficient and the storage of part pixel coordinate correspondence by the decimation rule of setting;

Interpolation calculation output subelement is used for according to the camera lens shadow correction coefficient of storage and the interpolation algorithm of setting, calculates the camera lens shadow correction coefficient and the output of current pixel point coordinate correspondence.

Beneficial effect of the present invention is as follows:

Adopt the present invention, to all determining its corresponding camera lens shadow correction coefficient by each pixel of the image of lens shooting, that is: at first according to training image and with the photographic images that obtains behind the lens shooting training image, determine the reference correction coefficient of each pixel, by the 2DC curved surface on the reference correction coefficient generation pixel coordinate plane of each pixel, adopt the binary high order power fitting of a polynomial 2DC curved surface of pixel coordinate, go out the camera lens shadow correction coefficient corresponding according to the binary high order power polynomial computation of determining again with each pixel coordinate.Because the 2DC curved surface that generates with reference to correction coefficient is for carrying noisy two-dimentional curved surface, it or not the true reflection of camera lens shade, the present invention is when definite camera lens shadow correction coefficient, the two-dimentional curved surface that the binary high order power fitting of a polynomial of employing pixel coordinate obtains, obtain two-dimentional smooth surface, make the camera lens shadow correction coefficient of determining near true.And, make that the image rectification effect is better owing to a well-determined corresponding camera lens shadow correction coefficient is all arranged at each pixel coordinate.The inventive method is applicable to dissimilar camera lenses, in the time need carrying out the camera lens Shadows Processing to the image of certain type lens shooting, only need use this lens shooting training image in advance, adopt method provided by the invention can determine camera lens shadow correction coefficient at the captured image of this camera lens, again according to the camera lens shadow correction coefficient of determining to carrying out treatment for correcting by the captured image of this camera lens.

Description of drawings

Fig. 1 determines the method step process flow diagram for the camera lens shadow correction coefficient that the embodiment of the invention provides;

The camera lens shadow correction method flow diagram that Fig. 2 provides for the embodiment of the invention;

Fig. 3 determines the apparatus structure synoptic diagram for the camera lens shadow correction coefficient that the embodiment of the invention provides;

Fig. 4 determines in the device with reference to correction coefficient acquiring unit structural representation for the camera lens shadow correction coefficient of Fig. 3;

The camera lens shadow correction apparatus structure synoptic diagram that Fig. 5 provides for the embodiment of the invention;

Fig. 6 is a camera lens shadow correction coefficient output unit structural representation in the camera lens shadow correction device of Fig. 5.

Embodiment

Below in conjunction with accompanying drawing, the inventive method and device are described in detail with specific embodiment.

Embodiment one:

Referring to Fig. 1, for providing a kind of camera lens shadow correction coefficient, the embodiment of the invention one determines method, comprise the steps:

Step S101, obtain the reference correction coefficient of each pixel according to training image.

Be specially:

Adopt lens shooting one width of cloth training image, obtain the photographic images of this training image; Wherein, the standard pixel value of each pixel is known in the training image.In the reality, for for simplicity, can use uniform blank sheet of paper of quality of lens shooting, for example obtain width of cloth size and be the photographic images of 640*480, the standard pixel value of supposing each pixel in this training image is 255, because the existence of camera lens shade, the pixel value of each pixel all might change (true pixel values that is each pixel might be not equal to 255) in the photographic images, for this reason, obtain the true pixel values (concrete acquisition methods is existing common practise, does not launch to describe) of each pixel in this photographic images.

In getting access to, after the true pixel values of each pixel, calculate the ratio of true pixel values and its standard pixel value (255) of each pixel, obtain the reference correction coefficient of each pixel at the corresponding photographic images of training image.As above-mentioned, when the image size is 640*480, obtain 640*480 with reference to correction coefficient.

Step S102, generate 2DC curved surface on the pixel coordinate plane by the reference correction coefficient that obtains.

Be specially:

Owing to each is corresponding with a pixel with reference to correction coefficient, and each pixel has the pixel coordinate of a unique correspondence, promptly presses from top to bottom in the image, series arrangement pixel from left to right, is example with the 640*480 image still, wherein:

The pixel coordinate of the 1st pixel correspondence is (1,1);

The pixel coordinate of the 2nd pixel correspondence is (2,1);

The pixel coordinate of the 3rd pixel correspondence is (3,1);

......

The pixel coordinate of the 640th pixel correspondence is (640,1);

The pixel coordinate of the 641st pixel correspondence is (1,2);

The pixel coordinate of the 642nd pixel correspondence is (2,2);

......

The pixel coordinate that the 307200th pixel (last pixel of this width of cloth image) is corresponding is (640,480);

The horizontal ordinate of supposing pixel is X, ordinate is Y, can generate a horizontal ordinate by the respective pixel point coordinate of 640*480 image is that X, ordinate are the pixel coordinate plane of Y, since the reference correction coefficient that obtains be with this pixel coordinate plane on each pixel one to one, therefore, can generate a 2DC curved surface by the 640*480 that obtains with reference to correction coefficient perpendicular to this pixel coordinate plane.

The hatching effect that the 2DC curved surface that generates among this step S102 is characterized is not the true hatching effect of camera lens, wherein contain noise, need carry out denoising, obtain characterizing a two-dimentional smooth surface of the true hatching effect of camera lens, for this reason, need to continue the following step.

The described 2DC curved surface of binary high order power fitting of a polynomial of step S103, employing pixel coordinate.

Be specially:

Setting is the binary high order power polynomial expression of variable with pixel horizontal stroke, ordinate, supposes that pixel horizontal ordinate variable is x, and the ordinate variable is y, is example with 6 powers, and wherein each polynomial weight coefficient is respectively β ₀₀, β ₀₁..., β ₆₆, specifically be calculated as:

$A=\left[\begin{array}{c}a\left(\mathrm{1,1}\right)\\ a\left(\mathrm{1,2}\right)\\ .\\ .\\ .\\ a(x,y)\\ .\\ .\\ .\\ a(M,N)\end{array}\right]=\left[\begin{array}{c}{\mathrm{\φ}}_{00}\left(\mathrm{1,1}\right){\mathrm{\φ}}_{01}\left(\mathrm{1,1}\right)...{\mathrm{\φ}}_{\mathrm{ij}}\left(\mathrm{1,1}\right)...{\mathrm{\φ}}_{66}\left(\mathrm{1,1}\right)\\ {\mathrm{\φ}}_{00}\left(\mathrm{1,2}\right){\mathrm{\φ}}_{01}\left(\mathrm{1,2}\right)...{\mathrm{\φ}}_{\mathrm{ij}}\left(\mathrm{1,2}\right)...{\mathrm{\φ}}_{66}\left(\mathrm{1,2}\right)\\ .\\ .\\ .\\ {\mathrm{\φ}}_{00}(x,y){\mathrm{\φ}}_{01}(x,y)...{\mathrm{\φ}}_{\mathrm{ij}}(x,y)...{\mathrm{\φ}}_{66}(x,y)\\ .\\ .\\ .\\ {\mathrm{\φ}}_{00}(M,N){\mathrm{\φ}}_{01}(M,N)...{\mathrm{\φ}}_{\mathrm{ij}}(M,N)...{\mathrm{\φ}}_{66}(M,N)\end{array}\right]\left[\begin{array}{c}{\mathrm{\β}}_{00}\\ {\mathrm{\β}}_{01}\\ .\\ .\\ .\\ {\mathrm{\β}}_{\mathrm{ij}}\\ .\\ .\\ .\\ {\mathrm{\β}}_{66}\end{array}\right]=\mathrm{\ψ\β}$

Wherein,

$\mathrm{\ψ}=\left[\begin{array}{c}{\mathrm{\φ}}_{00}\left(\mathrm{1,1}\right){\mathrm{\φ}}_{01}\left(\mathrm{1,1}\right)...{\mathrm{\φ}}_{\mathrm{ij}}\left(\mathrm{1,1}\right)...{\mathrm{\φ}}_{66}\left(\mathrm{1,1}\right)\\ {\mathrm{\φ}}_{00}\left(\mathrm{1,2}\right){\mathrm{\φ}}_{01}\left(\mathrm{1,2}\right)...{\mathrm{\φ}}_{\mathrm{ij}}\left(\mathrm{1,2}\right)...{\mathrm{\φ}}_{66}\left(\mathrm{1,2}\right)\\ .\\ .\\ .\\ {\mathrm{\φ}}_{00}(x,y){\mathrm{\φ}}_{01}(x,y)...{\mathrm{\φ}}_{\mathrm{ij}}(x,y)...{\mathrm{\φ}}_{66}(x,y)\\ .\\ .\\ .\\ {\mathrm{\φ}}_{00}(M,N){\mathrm{\φ}}_{01}(M,N)...{\mathrm{\φ}}_{\mathrm{ij}}(M,N)...{\mathrm{\φ}}_{66}(M,N)\end{array}\right]$

$\mathrm{\β}=\left[\begin{array}{c}{\mathrm{\β}}_{00}\\ {\mathrm{\β}}_{01}\\ .\\ .\\ .\\ {\mathrm{\β}}_{\mathrm{ij}}\\ .\\ .\\ .\\ {\mathrm{\β}}_{66}\end{array}\right]$

In the following formula, suppose that the image size is M*N, φ _Ij(x, y)=x ⁱy ^jMatrix A each pixel corresponding reference correction coefficient for obtaining by training image is considered noise effect, and according to the least square rule, the computing formula that obtains β is as follows:

β＝(ψ ^Tψ) ^-1ψ ^TA

After determining β, promptly determined match 2DC curved surface binary high order power polynomial expression Im_shading (x, y), that is:

Im_shading(x，y)＝β ₀₀+β ₀₁x+β ₀₂x ²+β ₀₃x ³+β ₀₄x ⁴+β ₀₅x ⁵+β ₀₆x ⁶

+β ₁₀y+β ₁₁xy+β ₁₂x ²y+β ₁₃x ³y+β ₁₄x ⁴y+β ₁₅x ⁵y+β ₁₆x ⁶y

+β ₂₀y ²+β ₂₁xy ²+β ₂₂x ²y ²+β ₂₃x ³y ²+β ₂₄x ⁴y ²+β ₂₅x ⁵y ²+β ₂₆x ⁶y ²

+β ₃₀y ³+β ₃₁xy ³+β ₃₂x ²y ³+β ₃₃x ³y ³+β ₃₄x ⁴y ³+β ₃₅x ⁵y ³+β ₃₆x ⁶y ³

+β ₄₀y ⁴+β ₄₁xy ⁴+β ₄₂x ²y ⁴+β ₄₃x ³y ⁴+β ₄₄x ⁴y ⁴+β ₄₅x ⁵y ⁴+β ₄₆x ⁶y ⁴

+β ₅₀y ⁵+β ₅₁xy ⁵+β ₅₂x ²y ⁵+β ₅₃x ³y ⁵+β ₅₄x ⁴y ⁵+β ₅₅x ⁵y ⁵+β ₅₆x ⁶y ⁵

+β ₆₀y ⁶+β ₆₁xy ⁶+β ₆₂x ²y ⁶+β ₆₃x ³y ⁶+β ₆₄x ⁴y ⁶+β ₆₅x ⁵y ⁶+β ₆₆x ⁶y ⁶

Among the step S103, be to be example with 6 power polynomial expressions of binary, the 2DC curved surface that generates is carried out match, in the practical application, can adopt 2-10 power fitting of a polynomial of pixel horizontal ordinate and ordinate to carry out the match of 2DC curved surface according to different situations.

The binary high order power polynomial expression that step S104, basis are determined calculates the camera lens shadow correction coefficient corresponding with each pixel coordinate.

Be specially:

With each pixel coordinate (x, the binary high order power polynomial expression Im_shading that y) substitution step S103 determines (x, y) in, promptly calculate the camera lens shadow correction coefficient corresponding with each pixel coordinate.

The camera lens shadow correction coefficient that adopts the foregoing description one step to obtain, be to adopt the image of a certain resolution and fixed at one type of camera lens, promptly at this type camera lens of taking training image, adopt and take other image that the identical resolution of training image is taken, can adopt the camera lens shadow correction coefficient of determining that other image of taking is carried out the camera lens shadow correction and handle.

If to the camera lens of other type, or carry out the camera lens shadow correction at the image of other resolution and handle, can adopt the process step of embodiment one to carry out same processing, promptly the camera lens with this type adopts given resolution to take training image, obtains camera lens shadow correction coefficient corresponding with each pixel coordinate in the image of camera lens under this resolution of this type.

After having obtained camera lens shadow correction coefficient, just can utilize this camera lens shadow correction coefficient that the successive image that is obtained by this camera lens is carried out treatment for correcting, to remove the camera lens shade in the image.

Embodiment two:

The embodiment of the invention two provides a kind of camera lens shadow correction method, and idiographic flow comprises as shown in Figure 2:

Step S201, import image to be corrected.

Step S202, according to pixels point coordinate obtains the pixel value of the corresponding pixel points of current pixel point coordinate in proper order, and obtains the camera lens shadow correction coefficient corresponding with the current pixel point coordinate.

Be specially: according in the image from top to bottom, from left to right pixel order, obtain the pixel value of corresponding pixel points successively; According to the pixel coordinate of pixel correspondence, obtain corresponding camera lens shadow correction coefficient again.Wherein, the camera lens shadow correction coefficient of each pixel coordinate correspondence adopts embodiment one described method to determine.

The pixel value that step S203, usefulness are obtained multiply by camera lens shadow correction coefficient, obtains proofreading and correct back image and output.

Be specially: the pixel value of each pixel in the image is proofreaied and correct with its corresponding camera lens shadow correction coefficient, that is: the pixel value with each pixel in the image to be corrected multiply by corresponding camera lens shadow correction coefficient, pixel value after obtaining proofreading and correct, export the pixel value after each pixel is proofreaied and correct, obtain carrying out the output image behind the camera lens shadow correction.

In the step S202 of embodiment two, obtain the concrete grammar of the camera lens shadow correction coefficient corresponding with the current pixel point coordinate, can adopt following two kinds of implementations at least:

Mode one: adopt the mode of software to preserve the camera lens shadow correction coefficient of each pixel coordinate correspondence; According to pixels point coordinate obtains the camera lens shadow correction coefficient corresponding with each pixel coordinate in proper order successively.

For example can adopt the mode of software control, according to pixels the point coordinate order is exported the camera lens shadow correction coefficient of each pixel coordinate correspondence successively.

Mode two: adopt hardware mode to preserve camera lens shadow correction coefficient, for example adopt storer (or storage unit), in order to save the storage space of hardware, earlier from the camera lens shadow correction coefficient of each pixel coordinate correspondence of calculating, extract the camera lens shadow correction coefficient and the storage of part pixel coordinate correspondence by the decimation rule of setting.For example, the camera lens shadow correction coefficient and the storage of a pixel coordinate correspondence of every interval plurality of pixels point extraction; A kind of preferable extraction result is: extract (2 ⁿ+ 1) * (2 ⁿ+ 1) individual pixel is as extracting 17*17 pixel and storage.

When employing mode two is preserved camera lens shadow correction coefficient, before execution in step S203, need to adopt interpolation algorithm to calculate and the corresponding camera lens shadow correction of each pixel coordinate coefficient and output, concrete interpolation algorithm can adopt various interpolation algorithm of the prior art theoretically, in order to guarantee the degree of accuracy of interpolation, generally do not consider to adopt linear interpolation, a kind of preferable interpolation algorithm is for adopting cube (cubic) interpolation algorithm.

Mode two is a kind of preferable implementation, realizes camera lens shadow correction, stable performance by adopting hardware; Adopt the mode of dot interlace extraction, saved the storage space of hardware, realize that cost is low; Adopt cube interpolation scheduling algorithm to export the corresponding camera lens shadow correction of each pixel coordinate coefficient synchronously, computational complexity is low, and control is simple.

Embodiment three:

Determine method according to the camera lens shadow correction coefficient that the above embodiment of the present invention one provides, present embodiment three provides a kind of corresponding camera lens shadow correction coefficient to determine device, its structural representation comprises as shown in Figure 3: with reference to correction coefficient acquiring unit 31,2DC curved surface generation unit 32, match unit 33 and camera lens shadow correction coefficient calculation unit 34.Wherein:

With reference to correction coefficient acquiring unit 31, be used for obtaining the reference correction coefficient and the output of each pixel according to training image;

2DC curved surface generation unit 32 is used to receive the reference correction coefficient with reference to 31 outputs of correction coefficient acquiring unit, generates the 2DC curved surface on the pixel coordinate plane;

Match unit 33, the 2DC curved surface that is used to adopt the binary high order power fitting of a polynomial 2DC curved surface generation unit 32 of pixel coordinate to generate;

Camera lens shadow correction coefficient calculation unit 34 is used for the polynomial expression determined according to match unit 33, calculates the camera lens shadow correction coefficient corresponding with each pixel coordinate.

In a kind of preferred implementation, with reference to the structural representation of correction coefficient acquiring unit 31 as shown in Figure 4, comprising:

Obtain subelement 311, be used to obtain the standard pixel value of each pixel of training image, and with the true pixel values of each pixel of the photographic images that obtains behind this training image of lens shooting;

Computation subunit 312 is used to calculate the ratio of true pixel values and its standard pixel value of each pixel, obtains the reference correction coefficient and the output of each pixel.

Embodiment four:

According to the camera lens shadow correction method that the above embodiment of the present invention two provides, present embodiment four provides a kind of corresponding camera lens shadow correction device, and its structural representation comprises as shown in Figure 5:

Image input block 41 is used to receive image to be corrected;

Pixel value acquiring unit 42 is used for the image to be corrected according to 41 receptions of image input block, and according to pixels point coordinate obtains the pixel value and the output of the corresponding pixel points of current pixel point coordinate in proper order;

Camera lens shadow correction coefficient determining unit 43, be used for obtaining the reference correction coefficient of each pixel according to training image, by the 2DC curved surface on the reference correction coefficient generation pixel coordinate plane, adopt the 2DC curved surface of the binary high order power fitting of a polynomial generation of pixel coordinate, go out the camera lens shadow correction coefficient corresponding with each pixel coordinate according to the polynomial computation of determining;

Camera lens shadow correction coefficient output unit 44 is used for according to the camera lens shadow correction coefficient determining unit 43 camera lens shadow correction coefficients corresponding with each pixel coordinate that calculate, the camera lens shadow correction coefficient of output current pixel point coordinate correspondence;

Correcting unit 45 is used for the pixel value with 42 outputs of pixel value acquiring unit, multiply by the camera lens shadow correction coefficient of the respective pixel point coordinate of camera lens shadow correction coefficient output unit 44 outputs, generates to proofread and correct the back image;

Image output unit 46 is used for image behind the output calibration.

In a kind of better embodiment, the structural representation of camera lens shadow correction coefficient output unit 44 comprises as shown in Figure 6:

Extract storing sub-units 441, be used for the camera lens shadow correction coefficient of each pixel coordinate correspondence of calculating from camera lens shadow correction coefficient determining unit 43, extract the camera lens shadow correction coefficient and the storage of part pixel coordinate correspondence by the decimation rule of setting;

Interpolation calculation output subelement 442 is used for calculating the camera lens shadow correction coefficient of current pixel point coordinate correspondence and exporting to correcting unit 45 according to the camera lens shadow correction coefficient that extracts storing sub-units 441 storages and the interpolation algorithm of setting.

According to the above embodiment of the present invention, at first according to training image and with the photographic images that obtains behind the lens shooting training image, determine the reference correction coefficient of each pixel, by the 2DC curved surface on the reference correction coefficient generation pixel coordinate plane of each pixel, because the 2DC curved surface is for carrying noisy two-dimentional curved surface, it or not the true reflection of camera lens shade, the present invention is by adopting the binary high order power fitting of a polynomial 2DC curved surface of pixel coordinate, determine each polynomial weight coefficient, calculate the camera lens shadow correction coefficient corresponding, thereby make the camera lens shadow correction coefficient of determining near true with each pixel coordinate.Use camera lens shadow correction coefficient that the inventive method determines better to the effect that the image of the follow-up shooting of this camera lens carries out the camera lens shadow correction.And the inventive method is common to dissimilar camera lenses, in the time need carrying out the camera lens Shadows Processing to the image of certain type lens shooting, only need use this lens shooting training image in advance, adopt method provided by the invention can determine camera lens shadow correction coefficient at the captured image of this camera lens, again according to the camera lens shadow correction coefficient of determining to carrying out treatment for correcting by the captured image of this camera lens.

Obviously, those skilled in the art can carry out various changes and modification to the present invention and not break away from the spirit and scope of the present invention.Like this, if of the present invention these are revised and modification belongs within the scope of claim of the present invention and equivalent technologies thereof, then the present invention also is intended to comprise these changes and modification interior.

Claims (11)

1, a kind of camera lens shadow correction coefficient is determined method, it is characterized in that, comprising:

Obtain the reference correction coefficient of each pixel according to training image;

Generate 2DC curved surface on the pixel coordinate plane by described with reference to correction coefficient;

Adopt the described 2DC curved surface of binary high order power fitting of a polynomial of pixel coordinate;

Go out the camera lens shadow correction coefficient corresponding according to described polynomial computation with each pixel coordinate.

2, the method for claim 1 is characterized in that, describedly obtains the reference correction coefficient of each pixel according to training image, comprising:

Obtain the standard pixel value of each pixel of training image; And

Obtain true pixel values with each pixel of the photographic images that obtains behind the described training image of lens shooting;

Calculate the ratio of true pixel values and its standard pixel value of each pixel, obtain the reference correction coefficient of each pixel.

3, method as claimed in claim 1 or 2 is characterized in that, the described 2DC curved surface of binary high order power fitting of a polynomial of described employing pixel coordinate comprises:

Adopt the described 2DC curved surface of 2-10 power fitting of a polynomial of pixel horizontal ordinate and ordinate.

4, method as claimed in claim 3 is characterized in that, the described 2DC curved surface of binary high order power fitting of a polynomial of described employing pixel coordinate comprises:

Adopt the described 2DC curved surface of 6 power fitting of a polynomials of pixel horizontal ordinate and ordinate.

5, a kind of camera lens shadow correction method is characterized in that, comprising:

Import image to be corrected;

According to pixels point coordinate obtains the pixel value of the corresponding pixel points of current pixel point coordinate in proper order, and obtains the camera lens shadow correction coefficient corresponding with the current pixel point coordinate; Wherein, definite method of described camera lens shadow correction coefficient is: the reference correction coefficient that obtains each pixel according to training image, generate 2DC curved surface on the pixel coordinate plane by described with reference to correction coefficient, adopt the described 2DC curved surface of binary high order power fitting of a polynomial of pixel coordinate, go out the camera lens shadow correction coefficient corresponding with each pixel coordinate according to described polynomial computation;

Multiply by the described camera lens shadow correction coefficient that obtains with the described pixel value that obtains, obtain proofreading and correct back image and output.

6, method as claimed in claim 5, it is characterized in that, also comprise: from the camera lens shadow correction coefficient of each pixel coordinate correspondence of calculating, extract the camera lens shadow correction coefficient and the storage of part pixel coordinate correspondence by the decimation rule of setting in advance;

Describedly obtain the camera lens shadow correction coefficient corresponding, comprising with the current pixel point coordinate:

According to the camera lens shadow correction coefficient of storage and the interpolation algorithm of setting, calculate the camera lens shadow correction coefficient of current pixel point coordinate correspondence.

7, method as claimed in claim 6, it is characterized in that, described camera lens shadow correction coefficient and storage by the decimation rule extraction part pixel coordinate correspondence of setting, comprising: every interval plurality of pixels point extracts the camera lens shadow correction coefficient of a pixel coordinate correspondence and stores;

The interpolation algorithm of described setting is the cube interpolation algorithm.

8, a kind of camera lens shadow correction coefficient is determined device, it is characterized in that, comprising:

With reference to the correction coefficient acquiring unit, be used for obtaining the reference correction coefficient and the output of each pixel according to training image;

2DC curved surface generation unit is used to receive described described with reference to correction coefficient with reference to the output of correction coefficient acquiring unit, generates the 2DC curved surface on the pixel coordinate plane;

The match unit, the 2DC curved surface that is used to adopt the described 2DC curved surface of the binary high order power fitting of a polynomial generation unit of pixel coordinate to generate;

Camera lens shadow correction coefficient calculation unit is used for the polynomial expression determined according to described match unit, calculates the camera lens shadow correction coefficient corresponding with each pixel coordinate.

9, device as claimed in claim 8 is characterized in that, and is described with reference to the correction coefficient acquiring unit, comprising:

Obtain subelement, be used to obtain the standard pixel value of each pixel of training image, and with the true pixel values of each pixel of the photographic images that obtains behind the described training image of lens shooting;

Computation subunit is used to calculate the ratio of true pixel values and its standard pixel value of each pixel, obtains the reference correction coefficient and the output of each pixel.

10, a kind of camera lens shadow correction device is characterized in that, comprising:

The image input block is used to receive image to be corrected;

The pixel value acquiring unit is used for the image to be corrected according to described image input block reception, and according to pixels point coordinate obtains the pixel value and the output of the corresponding pixel points of current pixel point coordinate in proper order;

Camera lens shadow correction coefficient determining unit, be used for obtaining the reference correction coefficient of each pixel according to training image, generate 2DC curved surface on the pixel coordinate plane by described with reference to correction coefficient, adopt the described 2DC curved surface of binary high order power fitting of a polynomial of pixel coordinate, go out the camera lens shadow correction coefficient corresponding with each pixel coordinate according to described polynomial computation;

Camera lens shadow correction coefficient output unit is used for the camera lens shadow correction coefficient corresponding with each pixel coordinate that calculate according to described camera lens shadow correction coefficient determining unit, the camera lens shadow correction coefficient of output current pixel point coordinate correspondence;

Correcting unit is used for the described pixel value with described pixel value acquiring unit output, multiply by the camera lens shadow correction coefficient of the respective pixel point coordinate of described camera lens shadow correction coefficient output unit output, generates to proofread and correct the back image;

The image output unit is used for image behind the output calibration.

11, device as claimed in claim 10 is characterized in that, described camera lens shadow correction coefficient output unit comprises:

Extract storing sub-units, be used for the camera lens shadow correction coefficient of each pixel coordinate correspondence of calculating from described camera lens shadow correction coefficient determining unit, extract the camera lens shadow correction coefficient and the storage of part pixel coordinate correspondence by the decimation rule of setting;

Interpolation calculation output subelement is used for according to the camera lens shadow correction coefficient of storage and the interpolation algorithm of setting, calculates the camera lens shadow correction coefficient and the output of current pixel point coordinate correspondence.

Images