CN101320469A - Image illumination emendation method and device thereof - Google Patents
Image illumination emendation method and device thereof Download PDFInfo
- Publication number
- CN101320469A CN101320469A CNA2008101163566A CN200810116356A CN101320469A CN 101320469 A CN101320469 A CN 101320469A CN A2008101163566 A CNA2008101163566 A CN A2008101163566A CN 200810116356 A CN200810116356 A CN 200810116356A CN 101320469 A CN101320469 A CN 101320469A
- Authority
- CN
- China
- Prior art keywords
- image
- input picture
- equalization
- brightness
- gamma matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The present invention relates to an image illumination correction method which comprises the steps as follows: an input image is processed for histogram equalization to get an equalized image; the equalized image projects towards a characteristic subspace used for reconstructing the image to get a projection vector; the projection vector is used for reconstructing the equalized image to get a reconstruction image; the equalized image is subtracted from the reconstruction image to get a difference value image; the difference value image is transformed into a Gamma matrix; the Gamma matrix is used for correcting the illumination of the equalized image. In the embodiment, the reconstruction image is the image with normal light, so the difference value image of the equalized image and the reconstruction image can reflect the polarization status of the equalized image more exactly; the Gamma matrix which is gotten from difference value image mapping is used for correcting the equalized image, so a realer and more natural corrected image is obtained. The present invention also provides an image illumination correction device.
Description
Technical field
The present invention relates to technical field of image processing, relate in particular to a kind of image irradiation correcting method and a kind of image irradiation means for correcting.
Background technology
In recent years, computer vision and pattern-recognition correlation technique have obtained development rapidly, and as a wherein important branch, face recognition technology also is subjected to more concern gradually, and is widely used in fields such as access control and attendance, information security, public security investigation.But the relatively poor problem of face identification method ubiquity adaptability that has proposed at present, the effect of identification is easy to be subjected to the influence of environmental change factors such as facial expression, attitude, shelter and illumination.In the application system of reality, illumination is the key factor of system for restricting performance especially.According to the test result of authority's recognition of face contest FRVT2002, be 1% o'clock at false acceptance rate, indoor control light conditions human face correct recognition rata can reach 93%, and the discrimination under the outdoor non-control light conditions can only reach 50%.Therefore, improve the robustness of recognition of face, the practicality that improves face identification system is had very realistic meanings illumination.
By to the image pre-service, can carry out illumination and proofread and correct.For example, gray-scale transformation method such as histogram equalization, Gamma correction all is the image enhancement technique of using always.Wherein, histogram equalization is the histogram modification method based on cumulative distribution functional transformation method, the tonal range of image can be drawn back, and makes histogram reach unanimity in bigger dynamic range, and image is more clear.
The Gamma bearing calibration is the mapping mode that determines input picture is arrived output image by the value of the Gamma factor, has promptly determined to strengthen low gray scale and has still strengthened high gray scale.As can be seen, the Gamma bearing calibration is a kind of method of controlling the illumination correction intensity by the Gamma factor, but traditional Gamma bearing calibration can't realize the purpose of correction intensity with the pixel adaptive change, promptly can't well proofread and correct high light and shade zone and the image of depositing.The new a kind of self-adaptation Gamma bearing calibration that proposes, adopt the correction model of nonlinear function mutual superposition to make the variation of Gamma value be tending towards reasonable more, weakened correction intensity to transitional region, enhancing is to the correction intensity of Gao Guang, shadow region, improved the adaptive faculty of Gamma correction, the image fault after avoiding proofreading and correct to illumination variation.
Though the new self-adaptation Gamma bearing calibration that proposes has improved the illumination calibration result to high light and shade zone and the image deposited to a certain extent, still there is defective in this method.
Summary of the invention
In view of this, technical matters to be solved by this invention provides a kind of image irradiation correcting method, with the illumination calibration result of further improvement to high light and shade zone and the image deposited.
In some embodiment of image irradiation correcting method, comprising: input picture is done histogram equalization handle, obtain the equalization image; The equalization image is obtained projection vector to the proper subspace projection that is used for image reconstruction; Utilize projection vector that the equalization image is rebuild, obtain reconstructed image; It is poor that equalization image and reconstructed image are done, and obtains error image; Convert error image to the Gamma matrix; Utilize the illumination of Gamma matrix correction equalization image.In this embodiment, because reconstructed image is the normal image of light, so the error image of equalization image and reconstructed image can reflect the polarisation situation of equalization image more accurately; Utilization is proofreaied and correct the equalization image with the Gamma matrix that error image mapping obtains, and obtains image after the correction of true nature more.In this embodiment, owing to do not have directly big or small as the standard of passing judgment on Gao Guang, shade and transitional region with the pixel value of input picture, but extract the polarisation zone with the error image of equalization image and reconstructed image, can put the intrinsic zone of image under transitional region to greatest extent like this, reduction is to this regional brightness adjustment, strengthened adjustment simultaneously, made and proofread and correct back image natural reality more Gao Guang, shadow region.
Another technical matters to be solved by this invention provides a kind of image irradiation means for correcting.
In some embodiment of image irradiation means for correcting, comprising: input picture is done the equalization unit that histogram equalization is handled; With the projecting cell of equalization image to the proper subspace projection that is used for image reconstruction; The reconstruction unit that utilizes projection vector that the equalization image is rebuild; Equalization image and reconstructed image are done poor subtrator; Error image is converted to the converting unit of Gamma matrix; Utilize the correcting unit of the illumination of Gamma matrix correction equalization image.In this embodiment, owing to do not have directly big or small as the standard of passing judgment on Gao Guang, shade and transitional region with the pixel value of original polarized light image, but extract the polarisation zone with the error image of original polarized light image and reconstructed image, can put the intrinsic zone of image under transitional region to greatest extent like this, reduction is to this regional brightness adjustment, strengthened adjustment simultaneously, made and proofread and correct back image natural reality more Gao Guang, shadow region.
Figure of description
Fig. 1 is an embodiment process flow diagram of image irradiation correcting method provided by the invention;
Fig. 2 is another embodiment process flow diagram of image irradiation correcting method provided by the invention;
Fig. 3 is another embodiment process flow diagram of image irradiation correcting method provided by the invention;
Fig. 4 is another embodiment process flow diagram of image irradiation correcting method provided by the invention;
Fig. 5 is an embodiment synoptic diagram of image irradiation means for correcting provided by the invention;
Fig. 6 is another embodiment synoptic diagram of image irradiation means for correcting provided by the invention;
Fig. 7 is another embodiment synoptic diagram of image irradiation means for correcting provided by the invention;
Fig. 8 is another embodiment synoptic diagram of image irradiation means for correcting provided by the invention;
Fig. 9 is another embodiment synoptic diagram of image irradiation means for correcting provided by the invention;
Figure 10 is another embodiment synoptic diagram of image irradiation means for correcting provided by the invention;
Figure 11 is another embodiment synoptic diagram of image irradiation means for correcting provided by the invention.
Embodiment
In aforesaid self-adaptation Gamma bearing calibration, rely on pixel value to judge Gao Guang, shade or transitional region merely, caused this method to have irrational place like this.For example, the pupil region brightness value of people's face is very low, handles if the shadow region is used as in this zone, improves brightness value, obviously is irrational.Therefore, need judge Gao Guang, shade or transitional region more exactly by certain standard, with some the intrinsic zone in the differentiate between images (for example pupil region of people's face), need not to proofread and correct for the brightness value in intrinsic zone, so just can make the image after the correction truer.
Fig. 1 shows a kind of flow process of image irradiation correcting method.
In step 11, to input picture I
InDo histogram equalization and handle, obtain the equalization image I
h
In step 12, with the equalization image I
hObtain projection vector Ω to the proper subspace projection that is used for image reconstruction.
Ω
T=[ω
1, ω
2..., ω
L], wherein,
L is the dimension of proper subspace, μ
j TBe j proper vector of proper subspace, I
AveThe average image for no polarisation sample set.
For guaranteeing not contain polarized component in follow-up reconstructed image, polarized component just should not contained in the training characteristics subspace that is used for image reconstruction, therefore will choose no polarisation composition of sample training set.Adopt the no polarisation sample set after principal component analysis (PCA) (PCA) method is trained histogram equalization, can obtain to be used for the proper subspace of image reconstruction.
In step 13, utilize projection vector Ω to the equalization image I
hRebuild, obtain reconstructed image I
r
By calculating
Can obtain reconstructed image.
Here, because first three feature is relatively more responsive to illumination variation, therefore since the 4th feature, to the equalization image I
hRebuild.If directly since i feature, to the equalization image I
hRebuild, then
Reconstructed image I
rKept the equalization image I
hPrincipal character, as profile information.
In step 14, with the equalization image I
hWith reconstructed image I
rIt is poor to do, and obtains error image I
d, i.e. I
d=I
h-I
r
So-called two images are done difference be meant corresponding pixel value in two images is subtracted each other that the image that is made of pixel value difference of acquisition is error image.By error image I
dIn pixel value, promptly reflect equalization image I such as Gao Guang, shade, transitional region
hThe illumination patterns feature.
In step 15, with error image I
dConvert the Gamma matrix to.
For in follow-up illumination is proofreaied and correct, taking different correcting schemes to highlight area, shadow region, transitional region, with error image I
dShould satisfy following condition when converting the Gamma matrix to:
Reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image.
There are a lot of existent method can be by above-mentioned condition with error image I
dConvert the Gamma matrix to, no longer these methods are enumerated one by one here.
Here, the Gao Guang of image, shade and transitional region can be understood as the illumination patterns feature of image.Be different from the image light and shade and distribute, the illumination patterns feature of image is used for the true illumination strength distribution situation of presentation video, can judge Gao Guang, shade or transitional region by the error image that makes up, and also just can obtain the illumination patterns feature of image.
The method of a kind of optional judgement highlight area, shadow region and transitional region is, set in advance a higher limit and a lower limit, pixel value is a highlight area greater than the zone of described higher limit in the error image, pixel value is the shadow region less than the zone of described lower limit, and the zone of pixel value between higher limit and lower limit is transitional region.
In step 16, utilize Gamma matrix correction equalization image I
hIllumination.
Illumination can realize the highlight area image is reduced brightness significantly after proofreading and correct, and the shadow region image is increased substantially brightness, keeps original image brightness for transitional region.
In the present embodiment, because reconstructed image is the normal image of light, so the error image of equalization image and reconstructed image can reflect the polarisation situation of equalization image more accurately; Utilization is proofreaied and correct the equalization image with the Gamma matrix that error image mapping obtains, and obtains image after the correction of true nature more.In the present embodiment, owing to do not have directly big or small as the standard of passing judgment on Gao Guang, shade and transitional region with the pixel value of input picture, but extract the polarisation zone with the error image of equalization image and reconstructed image, can put the intrinsic zone of image under transitional region to greatest extent like this, reduction is to this regional brightness adjustment, strengthened adjustment simultaneously, made and proofread and correct back image natural reality more Gao Guang, shadow region.
Need to prove, input picture is carried out histogram equalization handle, mainly is for unified with the pretreatment operation to training sample in the training process.If do not consider and the unification of training process, can omit step 11.Fig. 2 shows a kind of flow process of image irradiation correcting method.
In step 21, with input picture I
InObtain projection vector Ω to the proper subspace projection that is used for image reconstruction.
Ω
T=[ω
1, ω
2..., ω
L], wherein,
L is the dimension of proper subspace, μ
j TBe j proper vector of proper subspace, I
AveThe average image for no polarisation sample set.
Here, adopt the no polarisation sample set after the PCA method is trained histogram equalization, can obtain to be used for the proper subspace of image reconstruction.
In step 22, utilize projection vector Ω to input picture I
InRebuild, obtain reconstructed image I
r
By calculating
Can obtain reconstructed image.
Reconstructed image I
rKept input picture I
InPrincipal character, as profile information.
In step 23, with input picture I
InWith reconstructed image I
rIt is poor to do, and obtains error image I
d, i.e. I
d=I
In-I
r
By error image I
dIn pixel value, promptly reflect input picture I such as Gao Guang, shade, transitional region
InThe illumination patterns feature.
In step 24, with error image I
dConvert the Gamma matrix to.
For in follow-up illumination is proofreaied and correct, taking different correcting schemes to highlight area, shadow region, transitional region, with error image I
dShould satisfy following condition when converting the Gamma matrix to:
Reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image.
In step 25, utilize Gamma matrix correction input picture I
InIllumination.
In the present embodiment, because reconstructed image is the normal image of light, so the error image of input picture and reconstructed image can reflect the polarisation situation of input picture more accurately; Utilization is proofreaied and correct input picture with the Gamma matrix that error image mapping obtains, and obtains image after the correction of true nature more.In the present embodiment, owing to do not have directly big or small as the standard of passing judgment on Gao Guang, shade and transitional region with the pixel value of input picture, but extract the polarisation zone with the error image of input picture and reconstructed image, can put the intrinsic zone of image under transitional region to greatest extent like this, reduction is to this regional brightness adjustment, strengthened adjustment simultaneously, made and proofread and correct back image natural reality more Gao Guang, shadow region.
In fact, also can the mean variance regulationization substitute histogram equalization, the pixel mean variance of promptly adjusting training sample and input picture is a setting.Concrete grammar is as follows:
I′
i＝(I
i-m
i)×ustd/std
i+um
Wherein, I '
iBack image, I are handled in expression
iThe expression input picture, m
iAnd std
iRepresent input picture I respectively
iAverage gray and variance yields, um and ustd represent the benchmark of predefined average, variance respectively.
Fig. 3 shows the flow process of another kind of image irradiation correcting method.
In step 31, with input picture I
InObtain projection vector Ω to the proper subspace projection that is used for image reconstruction.
Ω
T=[ω
1, ω
2..., ω
L], wherein,
L is the dimension of proper subspace, μ
j TBe j proper vector of proper subspace, I
AveThe average image for no polarisation sample set.
Here, adopt the no polarisation sample set of PCA method training, can obtain to be used for the proper subspace of image reconstruction.
In step 33, utilize projection vector Ω to input picture I
InRebuild, obtain reconstructed image I
r
By calculating
Can obtain reconstructed image.
Reconstructed image I
rKept input picture I
InPrincipal character, as profile information.
In step 34, with input picture I
InWith reconstructed image I
rIt is poor to do, and obtains error image I
d, i.e. I
d=I
In-I
r
By error image I
dIn pixel value, promptly reflect input picture I such as Gao Guang, shade, transitional region
InThe illumination patterns feature.
In step 35, with error image I
dConvert the Gamma matrix to.
For in follow-up illumination is proofreaied and correct, taking different correcting schemes to highlight area, shadow region, transitional region, with error image I
dShould satisfy following condition when converting the Gamma matrix to:
Reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image.
In step 36, utilize Gamma matrix correction input picture I
InIllumination.
Fig. 4 shows the flow process of another kind of image irradiation correcting method.
In step 41, with input picture I
InObtain projection vector Ω to being used for the projection of image reconstruction proper subspace.
Ω
T=[ω
1, ω
2..., ω
L], wherein,
L is the dimension of proper subspace, μ
j TBe j proper vector of proper subspace, I
AveThe average image for no polarisation sample set.
Here, adopt the no polarisation sample set of PCA method training, can obtain to be used for the proper subspace of image reconstruction.
In step 42, utilize projection vector Ω to input picture I
InRebuild, obtain reconstructed image I
r
By calculating
Can obtain reconstructed image.
Reconstructed image I
rKept input picture I
InPrincipal character, as profile information.
In step 43, with input picture I
InWith reconstructed image I
rIt is poor to do, and obtains error image I
d, i.e. I
d=I
In-i
r
By error image I
dIn pixel value, promptly reflect input picture I such as Gao Guang, shade, transitional region
InThe illumination patterns feature.
In step 44, to error image I
dCarry out histogram equalization and handle, the error image I ' after the acquisition equalization
h
In step 45, with the error image I ' after the equalization
hConvert the Gamma matrix to.
For in follow-up illumination is proofreaied and correct, taking different correcting schemes to highlight area, shadow region, transitional region, with the error image I ' after the equalization
hShould satisfy following condition when converting the Gamma matrix to:
Reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image.
In step 46, utilize Gamma matrix correction input picture I
InIllumination.
By the various embodiments described above as can be seen, basic thought of the present invention is: the illumination patterns feature of extracting original polarized light image; The illumination patterns Feature Conversion of described original polarized light image is become to be used to proofread and correct the parameter matrix of illumination; Utilize described parameter matrix to proofread and correct the illumination of original polarized light image.Described original polarized light image is the image that is corrected, original polarized light image can but to be not limited to be input picture or equalization image.Described illumination patterns feature has mainly reflected the distribution situation of Gao Guang, shade and transitional region in the image.The method that obtains the illumination patterns feature has a lot, and a kind of optional mode is that the error image by original polarized light image and reconstructed image obtains the illumination patterns feature in the original polarized light image.In addition, because basic thought of the present invention is to extract exactly Gao Guang, shade, transitional region, and zones of different adopted different correction intensities, therefore last bearing calibration is not to adopt Gamma to proofread and correct, here be an application example just, feasibility of the present invention is described with the Gamma correction.As long as calibration curve can satisfy a little less than the transitional region correction, Gao Guang, shadow region to be proofreaied and correct and got final product more by force, this curve can be made of sine or tangent cutve.Owing to do not have directly big or small as the standard of passing judgment on Gao Guang, shade and transitional region with the pixel value of original polarized light image, but extract the polarisation zone with the error image of original polarized light image and reconstructed image, can put the intrinsic zone of image under transitional region to greatest extent like this, reduction is to this regional brightness adjustment, strengthened adjustment simultaneously, made and proofread and correct back image natural reality more Gao Guang, shadow region.In addition, image is carried out pretreated method also is not limited to histogram equalization, can adopt additive method that image is carried out pre-service, guarantee to the preprocess method of input picture with in the training process to identical the getting final product of preprocess method of training sample.
In addition, the method for acquisition reconstructed image also has a lot.In certain embodiments, at first original polarized light image is obtained projection vector Ω to the proper subspace projection that is used for image reconstruction, utilize described projection vector Ω that original polarized light image is rebuild and obtain reconstructed image I
rIn actual applications, also can adopt the average image of no polarisation sample set as reconstructed image I
r, and then obtain error image.The image that constitutes by average pixel value that so-called the average image is meant that the respective pixel to each image in the sample set is averaged that the back obtains.
Fig. 5 shows a kind of image irradiation means for correcting 500, and image irradiation means for correcting 500 comprises first module S51, second cell S 52 and the 3rd cell S 53.
First module S51 is used to extract the illumination patterns feature of original polarized light image; Second cell S 52 is used for the illumination patterns Feature Conversion of original polarized light image is become to be used to proofread and correct the parameter matrix of illumination; The 3rd cell S 53 is used to utilize parameter matrix to proofread and correct the illumination of original polarized light image.
The illumination patterns feature of original polarized light image has mainly reflected the distribution situation of Gao Guang, shade and transitional region in the original polarized light image.The mode of extracting the illumination patterns feature of original polarized light image has a lot, and a kind of optional mode is that the error image by original polarized light image and reconstructed image obtains the illumination patterns feature in the original polarized light image.In this case, first module S51 should comprise image reconstruction unit S511 and subtrator S512.
Image reconstruction unit S511 is used for original polarized light image is obtained projection vector to the proper subspace projection that is used for image reconstruction, utilizes described projection vector that original polarized light image is rebuild and obtains described reconstructed image; It is poor that subtrator S512 is used for original polarized light image and reconstructed image are done, and obtains error image, as shown in Figure 6.In addition, described original polarized light image is the image that is corrected, original polarized light image can but to be not limited to be input picture or equalization image.When original polarized light image when input picture is carried out equalization image that equalization obtained, the image irradiation means for correcting also should comprise input picture is carried out the equalization cell S 71 that equalization is handled, as shown in Figure 7.In device 700, the mode that 71 pairs of input pictures of equalization cell S carry out the equalization processing can have a lot, and a kind of optional mode is input picture to be carried out histogram equalization handle.
When original polarized light image was input picture, image reconstruction unit S511 was with input picture I
InObtain projection vector Ω to being used for the projection of image reconstruction proper subspace.
Ω
T=[ω
1, ω
2..., ω
L], wherein,
L is the dimension of proper subspace, μ
j TBe j proper vector of proper subspace, I
AveThe average image for no polarisation sample set.
For guaranteeing not contain polarized component in follow-up reconstructed image, polarized component just should not contained in the training characteristics subspace that is used for image reconstruction, therefore will choose no polarisation composition of sample training set.The mode that acquisition is used for the proper subspace of image reconstruction includes but not limited to following manner:
Adopt the no polarisation sample set after the PCA method is trained histogram equalization, obtain to be used for the proper subspace of image reconstruction; Perhaps, adopt the no polarisation sample set of PCA method training to obtain to be used for the proper subspace of image reconstruction.
By calculating
Can obtain reconstructed image I
r
Subtrator S512 is with input picture I
InWith reconstructed image I
rIt is poor to do, and obtains error image I
d, i.e. I
d=I
In-I
rBy error image I
dIn pixel value, promptly reflect input picture I such as Gao Guang, shade, transitional region
InThe illumination patterns feature.
When original polarized light image was the equalization image, image reconstruction unit S511 was with the equalization image I
hObtain projection vector Ω to the proper subspace projection that is used for image reconstruction.
Ω
T=[ω
1, ω
2..., ω
L], wherein,
L is the dimension of proper subspace, μ
j TBe j proper vector of proper subspace, I
AveThe average image for no polarisation sample set.
By calculating
Can obtain reconstructed image I
r
Subtrator S512 is with the equalization image I
hWith reconstructed image I
rIt is poor to do, and obtains error image I
d, i.e. I
d=I
h-I
rBy error image I
dIn pixel value, promptly reflect equalization image I such as Gao Guang, shade, transitional region
hThe illumination patterns feature.
Because basic thought of the present invention is to extract exactly Gao Guang, shade, transitional region, and zones of different is adopted different correction intensities.Therefore, satisfy a little less than the transitional region correction, strong this condition correcting mode is proofreaied and correct in Gao Guang, shadow region can be used, and a kind of optional mode is to adopt the Gamma correction.Under the situation of original polarized light image being carried out the Gamma correction, correspondingly, second cell S 52 need convert error image to the Gamma matrix.It is to be noted, no matter adopt which kind of correcting mode, for can be in follow-up illumination be proofreaied and correct, highlight area, shadow region, transitional region are taked different correcting schemes, when second cell S 52 becomes to be used to proofread and correct the parameter matrix of illumination in the illumination patterns Feature Conversion with original polarized light image, all should satisfy following condition: reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image as far as possible.
Fig. 8 shows a kind of image irradiation means for correcting 800, and image irradiation means for correcting 800 comprises first module S51, equalization cell S 71, second cell S 52 and the 3rd cell S 53.
First module S51 is used to extract the illumination patterns feature of input picture; Equalization cell S 71 is used for that the illumination patterns feature of input picture is done equalization and handles; Second cell S 52 is used for the illumination patterns Feature Conversion after the equalization is become to be used to proofread and correct the parameter matrix of illumination; The 3rd cell S 53 utilizes described parameter matrix to proofread and correct the illumination of input picture.
The mode of extracting the illumination patterns feature of original polarized light image has a lot, and a kind of optional mode is that the error image by original polarized light image and reconstructed image obtains the illumination patterns feature in the original polarized light image.In this case, first module S51 should comprise image reconstruction unit S511 and subtrator S512.
Image reconstruction unit S511 is used for original polarized light image is obtained projection vector to the proper subspace projection that is used for image reconstruction, utilizes described projection vector that original polarized light image is rebuild and obtains described reconstructed image; It is poor that subtrator S512 is used for original polarized light image and reconstructed image are done, and obtains error image.The mode that 71 pairs of error images of equalization cell S carry out the equalization processing can have a lot, and a kind of optional mode is error image to be carried out histogram equalization handle.No matter adopt which kind of correcting mode, for can be in follow-up illumination be proofreaied and correct, highlight area, shadow region, transitional region are taked different correcting schemes, when second cell S 52 becomes to be used to proofread and correct the parameter matrix of illumination in the illumination patterns Feature Conversion with original polarized light image, all should satisfy following condition: reduce the brightness of highlight area image as far as possible, improve the brightness of shadow region image, keep the brightness of transitional region image.
As can be seen, respectively install among the embodiment above-mentioned, owing to do not have directly big or small as the standard of passing judgment on Gao Guang, shade and transitional region with the pixel value of original polarized light image, but extract the polarisation zone with the error image of original polarized light image and reconstructed image, can put the intrinsic zone of image under transitional region to greatest extent like this, reduction has strengthened the adjustment to Gao Guang, shadow region simultaneously to this regional brightness adjustment, makes to proofread and correct back image natural reality more.In addition, image is carried out pretreated method also is not limited to histogram equalization, can adopt additive method that image is carried out pre-service, guarantee to the preprocess method of input picture with in the training process to identical the getting final product of preprocess method of training sample.
Fig. 9 shows a kind of image irradiation means for correcting 900, and this device comprises: input picture is done the equalization cell S 91 that histogram equalization is handled; With the projecting cell S92 of equalization image to the proper subspace projection that is used for image reconstruction; The reconstruction unit S93 that utilizes projection vector that the equalization image is rebuild; Equalization image and reconstructed image are poor subtrator S94; Error image is converted to the converting unit S95 of Gamma matrix; Utilize the correcting unit S96 of the illumination of Gamma matrix correction equalization image.
Figure 10 shows a kind of image irradiation means for correcting 1000, and this device comprises: with the projecting cell S92 of input picture to the proper subspace projection that is used for image reconstruction; The reconstruction unit S93 that utilizes projection vector that input picture is rebuild; Input picture and reconstructed image are poor subtrator S94; Error image is converted to the converting unit S95 of Gamma matrix; Utilize the correcting unit S96 of the illumination of Gamma matrix correction input picture.
Figure 11 shows a kind of image irradiation means for correcting 1100, and this device comprises: with input picture to the projecting cell S92 that is used for the projection of image reconstruction proper subspace; The reconstruction unit S93 that utilizes projection vector that input picture is rebuild; Input picture and reconstructed image are poor subtrator S94; Error image is carried out the equalization cell S 91 that histogram equalization is handled; Error image after the equalization is converted to the converting unit S95 of Gamma matrix; Utilize the correcting unit S96 of the illumination of Gamma matrix correction input picture.
The present invention also provides a kind of integrated circuit, is used to realize above-mentioned described device of arbitrary embodiment item or method.
The present invention also provides a kind of computer-readable medium, stores the program that is used to realize the described method of above-mentioned arbitrary embodiment.
Those skilled in the art can understand, various exemplary method step of describing in conjunction with the disclosed embodiments and device unit all can electronic hardware here, software or the combination of the two realize.In order to be clearly shown that the interchangeability between the hardware and software, more than various exemplary steps and unit are all carried out generally description with its functional form.This functional be to realize or realize depending on the design constraint that specific application and total system are realized with software with hardware.Those skilled in the art can be at each specific application, realize in many ways described functional, but the result of this realization should not be construed as and on the contrary deviates from scope of the present invention.
Utilize general processor, digital signal processor (DSP), special IC (ASIC), field programmable gate array (FPGA) or other programmable logical device, discrete gate or transistor logic, discrete hardware components or the combination in any among them, can realize or carry out the various exemplary unit of describing in conjunction with embodiment disclosed herein.General processor may be a microprocessor, but in another kind of situation, this processor may be processor, controller, microcontroller or the state machine of any routine.Processor also may be implemented as the combination of computing equipment, for example, and the combination of DSP and microprocessor, a plurality of microprocessor, one or more microprocessor or any other this kind structure in conjunction with the DSP core.
In conjunction with the step of the described method of above-mentioned disclosed embodiment can directly be presented as hardware, the software module carried out by processor or the combination of these two.Software module may be present in the storage media of RAM storer, flash memory, ROM storer, eprom memory, eeprom memory, register, hard disk, mobile disk, CD-ROM or any other form well known in the art.The coupling of a kind of exemplary storage medium and processor, thus make processor can be from this storage media read message, and can be to this storage media write information.In replacing example, storage media is the ingredient of processor.Processor and storage media may be present among the ASIC.This ASIC may be present in the subscriber station.Replace in the example at one, the discrete assembly that processor and storage media can be used as in the subscriber station exists.
According to described disclosed embodiment, can be so that those skilled in the art can realize or use the present invention.To those skilled in the art, the various modifications of these embodiment are conspicuous, and the general principles of definition here also can be applied to other embodiment on the basis that does not depart from the scope of the present invention with purport.Above-described embodiment only is preferred embodiment of the present invention, and is in order to restriction the present invention, within the spirit and principles in the present invention not all, any modification of being done, is equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (14)
1, a kind of image irradiation correcting method is characterized in that, comprising:
Input picture is done histogram equalization handle, obtain the equalization image;
The equalization image is obtained projection vector to the proper subspace projection that is used for image reconstruction;
Utilize projection vector that the equalization image is rebuild, obtain reconstructed image;
It is poor that equalization image and reconstructed image are done, and obtains error image;
Convert error image to the Gamma matrix;
Utilize the illumination of Gamma matrix correction equalization image.
2, the method for claim 1 is characterized in that, should satisfy following condition when converting error image to the Gamma matrix:
Reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image.
3, a kind of image irradiation correcting method is characterized in that, comprising:
Input picture is obtained projection vector to the proper subspace projection that is used for image reconstruction;
Utilize projection vector that input picture is rebuild, obtain reconstructed image;
It is poor that input picture and reconstructed image are done, and obtains error image;
Convert error image to the Gamma matrix;
Utilize the illumination of Gamma matrix correction input picture.
4, method as claimed in claim 3 is characterized in that, should satisfy following condition when converting error image to the Gamma matrix:
Reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image.
5, a kind of image irradiation correcting method is characterized in that, comprising:
Input picture is obtained projection vector to the proper subspace projection that is used for image reconstruction;
Utilize projection vector that input picture is rebuild, obtain reconstructed image;
It is poor that input picture and reconstructed image are done, and obtains error image;
Convert error image to the Gamma matrix;
Utilize the illumination of Gamma matrix correction input picture.
6, method as claimed in claim 5 is characterized in that, should satisfy following condition when converting error image to the Gamma matrix:
Reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image.
7, a kind of image irradiation correcting method is characterized in that, comprising:
Input picture is obtained projection vector to being used for the projection of image reconstruction proper subspace;
Utilize projection vector that input picture is rebuild, obtain reconstructed image;
It is poor that input picture and reconstructed image are done, and obtains error image;
Error image is carried out histogram equalization handle, the error image after the acquisition equalization;
Convert the error image after the equalization to the Gamma matrix;
Utilize the illumination of Gamma matrix correction input picture.
8, method as claimed in claim 7 is characterized in that, should satisfy following condition when converting error image to the Gamma matrix:
Reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image.
9, a kind of image irradiation means for correcting is characterized in that, comprising:
Input picture is done the equalization unit that histogram equalization is handled;
With the projecting cell of equalization image to the proper subspace projection that is used for image reconstruction;
The reconstruction unit that utilizes projection vector that the equalization image is rebuild;
Equalization image and reconstructed image are done poor subtrator;
Error image is converted to the converting unit of Gamma matrix;
Utilize the correcting unit of the illumination of Gamma matrix correction equalization image.
10, device as claimed in claim 9 is characterized in that, should satisfy following condition when converting error image to the Gamma matrix:
Reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image.
11, a kind of image irradiation means for correcting is characterized in that, comprising:
With the projecting cell of input picture to the proper subspace projection that is used for image reconstruction;
The reconstruction unit that utilizes projection vector that input picture is rebuild;
Input picture and reconstructed image are done poor subtrator;
Error image is converted to the converting unit of Gamma matrix;
Utilize the correcting unit of the illumination of Gamma matrix correction input picture.
12, device as claimed in claim 11 is characterized in that, should satisfy following condition when converting error image to the Gamma matrix:
Reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image.
13, a kind of image irradiation means for correcting is characterized in that, comprising:
With input picture to the projecting cell that is used for the projection of image reconstruction proper subspace;
The reconstruction unit that utilizes projection vector that input picture is rebuild;
Input picture and reconstructed image are done poor subtrator;
Error image is carried out the equalization unit that histogram equalization is handled;
Error image after the equalization is converted to the converting unit of Gamma matrix;
Utilize the correcting unit of the illumination of Gamma matrix correction input picture.
14, device as claimed in claim 13 is characterized in that, should satisfy following condition when converting error image to the Gamma matrix:
Reduce the brightness of highlight area image, improve the brightness of shadow region image, keep the brightness of transitional region image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008101163566A CN101320469A (en) | 2008-07-09 | 2008-07-09 | Image illumination emendation method and device thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008101163566A CN101320469A (en) | 2008-07-09 | 2008-07-09 | Image illumination emendation method and device thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101320469A true CN101320469A (en) | 2008-12-10 |
Family
ID=40180501
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2008101163566A Pending CN101320469A (en) | 2008-07-09 | 2008-07-09 | Image illumination emendation method and device thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101320469A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101630401B (en) * | 2009-07-31 | 2011-12-21 | 北京师范大学 | ATGP-VCA projection vector-obtaining method |
| CN102693532A (en) * | 2011-02-25 | 2012-09-26 | 微软公司 | Automatic localized adjustment of image shadows and highlights |
| CN108490652A (en) * | 2018-02-09 | 2018-09-04 | 深圳市华星光电半导体显示技术有限公司 | A kind of method and system of Gamma adjust automaticallies |
| CN108921033A (en) * | 2018-06-04 | 2018-11-30 | 北京京东金融科技控股有限公司 | Face picture comparison method, device, medium and electronic equipment |
| CN109145755A (en) * | 2018-07-25 | 2019-01-04 | 昆明聚信丰科技有限公司 | A kind of desk area recognizing method of combination perspective transform and K- mean algorithm |
| CN110930340A (en) * | 2019-10-11 | 2020-03-27 | 成都华为技术有限公司 | Image processing method and device |
| CN115731139A (en) * | 2022-12-08 | 2023-03-03 | 深圳明锐理想科技有限公司 | Image correction method and electronic equipment |
-
2008
- 2008-07-09 CN CNA2008101163566A patent/CN101320469A/en active Pending
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101630401B (en) * | 2009-07-31 | 2011-12-21 | 北京师范大学 | ATGP-VCA projection vector-obtaining method |
| CN102693532A (en) * | 2011-02-25 | 2012-09-26 | 微软公司 | Automatic localized adjustment of image shadows and highlights |
| CN102693532B (en) * | 2011-02-25 | 2016-08-10 | 微软技术许可有限责任公司 | The automatic local directed complete set of image shadow and highlight |
| CN108490652A (en) * | 2018-02-09 | 2018-09-04 | 深圳市华星光电半导体显示技术有限公司 | A kind of method and system of Gamma adjust automaticallies |
| CN108490652B (en) * | 2018-02-09 | 2020-05-22 | 深圳市华星光电半导体显示技术有限公司 | Gamma automatic adjustment method and system |
| CN108921033A (en) * | 2018-06-04 | 2018-11-30 | 北京京东金融科技控股有限公司 | Face picture comparison method, device, medium and electronic equipment |
| CN109145755A (en) * | 2018-07-25 | 2019-01-04 | 昆明聚信丰科技有限公司 | A kind of desk area recognizing method of combination perspective transform and K- mean algorithm |
| CN110930340A (en) * | 2019-10-11 | 2020-03-27 | 成都华为技术有限公司 | Image processing method and device |
| CN110930340B (en) * | 2019-10-11 | 2023-09-29 | 成都华为技术有限公司 | Image processing method and device |
| CN115731139A (en) * | 2022-12-08 | 2023-03-03 | 深圳明锐理想科技有限公司 | Image correction method and electronic equipment |
| CN115731139B (en) * | 2022-12-08 | 2023-09-01 | 深圳明锐理想科技有限公司 | Image correction method and electronic device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101320469A (en) | Image illumination emendation method and device thereof | |
| Ramzan et al. | External debt growth nexus: Role of macroeconomic polices | |
| US9369684B2 (en) | Image tone adjustment using local tone curve computation | |
| Gijsenij et al. | Computational color constancy: Survey and experiments | |
| CN106067287A (en) | Display device and the method being used for controlling this display device | |
| CN108229234B (en) | Scannable image generation method fusing digital coding | |
| CN103093203B (en) | A kind of human body recognition methods again and human body identify system again | |
| Ylioinas et al. | Age estimation using local binary pattern kernel density estimate | |
| CN108876756B (en) | Image similarity measurement method and device | |
| CN112884758B (en) | Defect insulator sample generation method and system based on style migration method | |
| CN112651342B (en) | Face recognition method and device, electronic equipment and storage medium | |
| Li et al. | A novel image enhancement method using fuzzy sure entropy | |
| CN106101679A (en) | The processing method of high dynamic range signal and processing system | |
| CN111209858A (en) | Real-time license plate detection method based on deep convolutional neural network | |
| CN104867121A (en) | Fast image defogging method based on dark channel prior and Retinex theory | |
| Lanitis et al. | Towards automatic face identification robust to ageing variation | |
| CN114627435A (en) | Intelligent light adjusting method, device, equipment and medium based on image recognition | |
| Ding et al. | Sparse codes fusion for context enhancement of night video surveillance | |
| Ma et al. | A new algorithm for backlight image enhancement | |
| CN111861943A (en) | Image restoration method for multi-scale pixel-level depth optimization | |
| CN111145316A (en) | Teaching animation production system | |
| CN111353976B (en) | Sand grain target detection method based on convolutional neural network | |
| CN114372941A (en) | Low-illumination image enhancement method, device, equipment and medium | |
| TWI244330B (en) | Image sensing device with pixel correction function and method for correcting pixel sensing-information in an image sensing device | |
| Ma et al. | Image restoration and enhancement in monitoring systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20081210 |