CN107231549A - AWB detecting system - Google Patents
AWB detecting system Download PDFInfo
- Publication number
- CN107231549A CN107231549A CN201610186408.1A CN201610186408A CN107231549A CN 107231549 A CN107231549 A CN 107231549A CN 201610186408 A CN201610186408 A CN 201610186408A CN 107231549 A CN107231549 A CN 107231549A
- Authority
- CN
- China
- Prior art keywords
- colour
- region
- white
- light source
- curve
- 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
- 238000003384 imaging method Methods 0.000 claims abstract description 18
- 238000001514 detection method Methods 0.000 claims abstract description 10
- 230000000694 effects Effects 0.000 claims description 3
- 238000006467 substitution reaction Methods 0.000 claims 1
- 230000003595 spectral effect Effects 0.000 abstract description 12
- 238000000034 method Methods 0.000 abstract description 9
- 238000000638 solvent extraction Methods 0.000 abstract description 2
- 230000008859 change Effects 0.000 description 13
- 238000012360 testing method Methods 0.000 description 12
- 238000001228 spectrum Methods 0.000 description 11
- 239000003086 colorant Substances 0.000 description 8
- 238000009826 distribution Methods 0.000 description 8
- 230000005457 Black-body radiation Effects 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000012937 correction Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 5
- 230000035945 sensitivity Effects 0.000 description 5
- 230000000052 comparative effect Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000005286 illumination Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000001429 visible spectrum Methods 0.000 description 3
- 230000004075 alteration Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 241000282693 Cercopithecidae Species 0.000 description 1
- 241000736199 Paeonia Species 0.000 description 1
- 235000006484 Paeonia officinalis Nutrition 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 230000009514 concussion Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000004043 dyeing Methods 0.000 description 1
- 230000005662 electromechanics Effects 0.000 description 1
- 235000021384 green leafy vegetables Nutrition 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 229910044991 metal oxide Inorganic materials 0.000 description 1
- 150000004706 metal oxides Chemical class 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000002834 transmittance Methods 0.000 description 1
- 238000013316 zoning Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/80—Camera processing pipelines; Components thereof
- H04N23/84—Camera processing pipelines; Components thereof for processing colour signals
- H04N23/88—Camera processing pipelines; Components thereof for processing colour signals for colour balance, e.g. white-balance circuits or colour temperature control
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Color Television Image Signal Generators (AREA)
Abstract
The present invention relates to AWB detecting system, the composition of uniqueness is employed.It is combined using the characteristic of camera itself with blackbody curve characteristic, replaces existing region and weight scheduling algorithm with white/Grey Point netrual colour region is found out.By to two regions of the light source characteristic using warm colour light source and cool colour light source, the information for changing calculating R/G and B/G using variable image partitioning technique is used as decision element, the region of reference point uses non-close adjustable way, it is determined that in the way of variable minimum and maximum curve of approximation.Colour temperature calculating is carried out to regional according to detection reference color temperature/detection colour temperature precision, exclude the colored areas in color temperature regions and exclude the colored areas in brightness, finally judge excluding white/gray area number behind colored areas, to judging that white/grey netrual colour region carries out calculating R, B gain and colour temperature.The influence of the characteristic of the characteristic of imaging sensor in the camera module of the invention for preferably solving light source spectral characteristic and imaging sensor, IRCF characteristics and camera lens, improves the accuracy rate of white balance.
Description
Technical field
The present invention relates to a kind of AWB detecting system.
Background technology
White balance (White balance) is in the current digital image world, particularly in photography and image domains,
It is very important factor, the repeatability influence on image color is very big.
The essence of white balance, is because human brain " Memory color "-C.J.Bartleson (1960) memory colour.I.e.
Under any lighting environment, no matter how illumination light changes, it is necessary to which eyes are seen that man memory is arrived in " white " correction in other words
White in color.Eyes are seen " white " --- it is actually reflection of the white object to light source, and digital camera/video camera
The light that white object reflects is become image by imaging sensor.It is real physical quantity in no progress white balance processing image
Reflection, and change because of the change of light source, but this change and white in man memory color is constant and contradiction.
The algorithm of white balance about solving this contradiction proposes that much main being substantially classified as is following in recent years
Several ways：
Blackbody curve approximate region method in B/G and R/G coordinates as shown in Figure 1.(R：Red, G：Green, B：It is blue
Color)
Such as PHILIPS Co. (CN200580044593), Nikon Corp (CN200810095706) and Core
Logic, Inc. (JPA2012518361) etc. are main representative.PHILIPS Co. limits predetermined color space and black body radiation is bent
1-1 in line, figure, is known, by a left side, and right mould intends Proximate blackbody curve, and 1-2 in figure calculates white balance；NIKON exists
On blackbody radiance curve or nearby, 1 in figure, the 1-3 in several regions, such as Fig. 1 is divided, white balance is calculated；Core Logic are public
Department, then be to have divided more source regions, the 1-4 in such as Fig. 1, and each region makes corrections by the dark angle of camera lens (Shading)
Reference point is chosen, and calculates white balance.
Fig. 2 rgb values carry out color space conversion aberration domain mode.
Such as Spreadtrum Communications Inc. (CN200610117166), Toshiba Corporation (CN200710185785,
CN201210320233), the Central China University of Science and Technology (CN201410197703), Samsung Electro-Mechanics Co., Ltd
(CN200810146275), Qualcomm Inc (CN200780039526) etc. is main representative.Spreadtrum company is by inciting somebody to action
RGB is converted into YCbCr, and whether average color difference is fallen under predefined light source in scope, according to distance weighted heavy, 2-4 in figure, carries out
Calculate；Toshiba calculates 2-3 in decision gate (region), figure by the way that RGB is converted into YUV, judges white, decision threshold pair
The each light source determined according to white balance gains delimit the color temperature scope of the object as blank level adjustment；The Central China University of Science and Technology
Uv is converted into, using fitting uv curves more than certain brightness, 2-1 in figure is calculated；Samsung is by the way that RGB is changed
Into Lab, it is contemplated that brightness L and a b intervals relation, luminance segmentation is calculated；Qualcomm is united by the way that RGB is converted into YCbCr
2-2 in the Cr/Cb of quadrilateral area, figure is counted, is calculated.
Explanation is needed, RGB, which is carried out color space, is converted into YUV/YCrCb/Lab, substantially similar.Fig. 3 often uses lamp box
UV the and ab coordinate position figures of 18% gray card of SpectraLight III (follow-up explanation) light source, Fig. 3 left sides 3-1 is YUV,
The right 3-2 is Lab.As illustrated, UV and ab are substantially similar.In various documents, during only Samsung is Lab in view of brightness
L, and to brightness carry out segmentation calculating.And other documents then exclude the influence of brightness substantially, such as Central China University of Science and Technology uses one
Determine the reference point selection after fitting uv curves more than brightness, and the dark angle correction of Core Logic companies suggestion camera lens, it is relevant
The influence of brightness will illustrated subsequently.
In addition, the advantage and disadvantage of the other manner, such as Gray world/ color constancies outside 2 kinds of modes of the above, upper
State and all on the books in document no longer enumerate herein.The algorithm of white balance can be attributed in a word：Black matrix Radiometric profiles are simulated, are chosen
Reference light source white point, setting curve and reference point region carries out approximate/weight and judged, in those steps all Shortcomings it
Place.
First, blackbody curve error influence factor --- the influence of spectral change.As shown in figure 4, blackbody radiation is in visible light
Colour temperature near spectral limit represents curve.4-1 is colour temperature 2500K, and 4-2 is colour temperature 3500K, and 4-3 is that colour temperature 5500K, 4-4 are color
Warm 7500K curve.It can be seen that blackbody radiation curve frequency spectrum consecutive variations, the low region in wavelength of light length of colour temperature will height
In the short region of wavelength, colour temperature it is high then conversely, the short region of wavelength of light is higher than the region of wavelength length.And Ordinary Light Sources Have
Frequency spectrum then has peak change.As shown in figure 5, conventional lamp box SpectraLight III light source frequency spectrum.5-1 is H light in figure
Source (2300K), 5-2 is A light sources (2856K), and 5-3 is CWF light sources (4150K), and 5-4 is TL84 light sources (4100K), and 5-5 is D65
Light source (6504K).Each light source specific name and color temperature value have regulation in CIE 1931 or 1964 (International Commission on Illumination), this
In no longer tell in detail.Compared from lighting source with blackbody radiation spectrum curve, A the and H light source blackbody radiation curves of low colour temperature
Unanimously, CWF the and TL84 light sources of middle colour temperature have very big peak change, and high color temperature D65 light sources also have peak change but totality
It is upper more approximate with black matrix.Although in addition, relative spectral power distributions of the International Commission on Illumination CIE to the various light sources of illumination
There is the prompting of standard frequency spectrum/colour temperature etc., when actually light source manufacturer establishing criteria is produced, simulated spectra institute additive
Difference can produce difference, cause end user to determine and in use, the differences such as colour temperature can be produced.And the influence of this difference, it can lead
Cause mixes up white balance product, is shifted in other similar scenes.
Secondly, the imaging sensor of digital camera/video camera (CCD---Charge-coupled Device or
CMOS---Complementary Metal-Oxide Semiconductor) influence to blackbody curve.For accurately, dress
There is the influence of the camera module of imaging sensor.The influence of camera module can be divided mainly into：The shadow of imaging sensor characteristic
Ring, the influence and the influence of lens properties of IRCF (Infrared Ray Cut Filter) IR-cut filter plate characteristic.As schemed
Shown in 6, RGB relative spectral sensitivity map in imaging sensor.6-1 is blue in figure, and 6-2 is green (note in figure：Gr's and Gb
Average value), in figure 6-3 be the waveform in red, figure for RGB frequency spectrums each wavelength sensitivity value and maximum sensitivity value in green
Than the ratio of maximum sensitivity value is defined as 1 in green spectral.The RGB photobehaviors of imaging sensor, because of producer and the difference of model
And it is different.6-2 greens in such as figure, the peak value 6-4 for being mainly manifested in green is different with the wide 6-5 of ripple.Blue 6-1 and red 6-2 with
Shown green 6-2 waveforms are the same, and the peak value and ripple on frequency spectrum are wide also can be different.Therefore camera module can be caused, even if other
Under building block and lighting environment all same cases, it can also produce different images --- i.e. the yield value of white balance is different.
The influence of IRCF characteristics is mainly manifested in IRCF filter curve.As shown in fig. 7, various conventional IRCF are in visible ray
Filtering characteristic distribution in spectral limit.IRCF filtering characteristics Fig. 7 is such as overlapped with light source Fig. 5 as can be seen that in the field of wavelength length
But the IRCF filter effects for depending near-infrared field alone are larger.Different IRCF, mutually sympathizes with building block with lighting environment
Under condition, different images can be also produced.
The influence of lens properties is mainly manifested in the dark angle influence of transmittance curve and camera lens (Shading) of camera lens.Such as Fig. 8
It is shown, transmission characteristic distribution of the camera lens in limit of visible spectrum.As IRCF, different camera lenses has different transmissions bent
Line and the dark angle of camera lens (Shading), different camera lens under building block and lighting environment all same cases, can also have been produced
The image of difference.
Fig. 9 be two cameras under same light source environment D65, the R/G and B/G of 24 pieces of colour atlas (Macbeth) of shooting
Distribution map example.Transverse axis is R/G, and vertical pivot is B/G.9-3 in light source reference point D75, D65, D55, TL84, figure, and light source
9-4 in reference Point C WF, A, H, figure, left and right is identical, using the data of the camera on the left side.Two cameras have identical image
Sensor and camera lens, but use different IRCF.The gray world points of the 9-1 cameras on the left side and the 9- on the right in figure in figure
The gray world of 2 cameras point compare as can be seen that the left side 9-1 cameras gray world points, with reference point
D65 light sources are identical, and the point for having the gray world of the right 9-2 cameras is higher than left side D75 reference points, and both have very big
Difference.The camera difference of left and right two only has IRCF different.It is understood that the Radiometric profiles formula of black matrix is constant, therefore, by
In the characteristic of light source and camera, be not suitable for being used as normative reference with constant blackbody curve, carry out around blackbody curve
Zoning judges.
For the selection of R/G and B/G white point reference points, existing frequently-used reference point chooses the ash using Kodak 18%
Card or 24 pieces of colour atlas (Macbeth) are more.Also both modes are used in aforementioned documents substantially, and for selection range,
Essentially polygons closed area.As shown in Figure 10, commonly use under lamphouse light source, shoot reference point R/G and B/G obtained by 18% gray card
The practical example of relation.Transverse axis is R/G, and vertical pivot is B/G.Captured image does not carry out dark angle (Shading) correction processing, from
In it can be seen from the figure that, figure between 10-1 D75 and D65 light sources, in figure between 10-2 CWF and TL84 light sources, 10-2 in figure
A and H light sources between, with intersection.Therefore, the multiple regions of division in any case, for 10-1 in figure, 10-2 and
Come as 10-3 for continuum, it is impossible to efficiently differentiate and come.10-4 in Figure 10 is as the 9-3 in preceding Fig. 9
10-5 in light source reference point D75, D65, D55, TL84 regions, Figure 10 is light source reference point as the 9-4 in preceding Fig. 9
CWF, A, H region.10-4 and 10-5 in Figure 10 than comparatively dense if taking reference point, and the R/G and B/G of these light sources will
Connect together and cannot be distinguished by.Document beyond foregoing Samsung, all excludes the influence of brightness, to confirm reference point substantially
Region.In reality, the change of brightness is consecutive variations, is the important performance of image.Ignore the influence of brightness, can have to white balance
Very big influence.Secondly, the reference zone of most document is closed interval, and light source characteristic that is continuous and overlapping certainly will
The judgement precision of white balance can be influenceed.
Figure 11 is conventional test chart (left side), R/G the and B/G coordinates relation of each segmented areas is (right under d 65 illuminant
Figure).Transverse axis is R/G, and vertical pivot is B/G.11-1 is image gray world point in figure.11-2 be light source reference point D75, D65,
D55, TL84 region, 11-3 are light source reference Point C WF, A, H regions.The D65 references of region and Figure 10 shown in Figure 11 11-1
The gray world in point region are identical.11-2 is interval, and D65 reference points lower regions are continuous, to tell these continuums
White and other color in the correspondence right figure of domain is relatively difficult.Although passing through the week of the white reference point in R/G and B/G coordinates
Enclose setting and limit parameter or adjustable range equal weight, the judgement of " correct " can be produced, but this " correct " judgement is temporary
When and it is unstable, parameter setting can be quite numerous and diverse and needs repeated multiple times debugging, has a strong impact on efficiency.Moreover light source is constantly
Carving can all change, the change of each scene, can all influence to debug effect again.It should be noted that conventional test chart should
To contain the composition of various colors in image, just the gray world of this image point and light source D65 be relatively.
The content of the invention
For the characteristic of the imaging sensor in the camera module of above-mentioned light source spectral characteristic and imaging sensor, IRCF
The characteristic of characteristic and camera lens, the present invention uses 9-4 in two parts, such as warm colour light source region, Fig. 9, and cool colour source region, such as Fig. 9
Middle 9-3, the method for replacing single black matrix approximate.Warm colour source region is particularly longer in spectral wavelength according to light source frequency spectrum
Scope, energy is larger, and main representative is：CWF light sources, A light sources, H light sources.Cool colour source region, the shorter model of spectral wavelength
The scope based on energy is enclosed, main representative is：TL84 light sources and serial (D50/D65/D75) light sources of D.Certainly, the present invention is not limited
Two parts, according to the camera module characteristic of imaging sensor, can use more source region subregions.
For above-mentioned because brightness change reference point region is continuously overlapped, it is impossible to the 10-1 in the problem of distinguishing such as Figure 10,
Shown in 10-2,10-3, the present invention uses non-close, and abscissa R/G takes minimum and maximum region in direction.In ordinate B/G side
Upwards, open mode is taken, the 10- in the 10-4 cool colours source region and Figure 10 in B/G upper lower threshold value, such as Figure 10 is not limited
It is shown in solid in 5 warm colour source regions.Equally, the present invention does not limit aforesaid way, camera that can also be according to imaging sensor
Module, can also B/G directions take minimum and maximum region, the mode of R/G upper lower threshold value is not limited.10-4 in Figure 10 is cold
The abscissa R/G directions of coloured light source region, 10-6 is minimum value curve of approximation, and 10-7 is maximum curve of approximation.In Figure 10
The abscissa R/G directions of 10-5 warm colour source regions, 10-8 is minimum value curve of approximation, and 10-9 is maximum curve of approximation.Need
To illustrate that the 10-5 warm colour source regions in 10-4 cool colours source region and Figure 10 in Fig. 10 have intersection, this hair
Bright feature carries out counting/conventional direct the computing such as weight using to entering region part, therefore does not influence.In the past scheme
Conventional test image can be seen that R/G in image in Figure 11 of D65 light sources, and the B/G overwhelming majority falls within 11-2 cool colours
Between the maximum curve 11-5 and minimum curve 11-4 of source region, and span is continuous to TL84 always from D75, is partly into
Between the maximum curve 11-7 and minimum curve 11-6 of 11-3 warm colour source regions.These continuous parts, include colored portion
Divide and non-colored portion is white/grey parts, distinguish using colored and non-colored portion, white/grey is done at white balance
Manage relatively difficult, colored influence explanation in example below.
For it is above-mentioned existing the problem of, the technical problem to be solved in the present invention is to provide one kind can AWB detection
System, it is bent using the characteristic and black matrix of camera itself by the reference point selection mode of each colour temperature to improving existing platform
Line characteristic is combined, and is replaced existing region and Weight algorithm with white (netrual colour) region is found out, is improved the judgement of white balance
Accuracy rate, while the setting of existing parameter can be reduced, improves debugging efficiency.
Present invention employs following technical scheme：
Present invention employs unique composition.It is combined using the characteristic of camera itself with blackbody curve characteristic, with looking for
Go out Grey Point and replace existing region and Weight algorithm.First, calculated using two regions of warm colour light source and cool colour light source.The
Two, the information for changing calculating R/G and B/G using variable image partitioning technique is used as decision element.Typically, picture portion is using solid
Setting is more, and fixed picture portion is just widely used initial stage in digital image, and picture portion can be defeated in imaging sensor
The rear end gone out is carried out, and is easier to realize, while for compared with the mode calculated using each pixel, it is possible to reduce internal patrols
Collect circuit；Reduce because of the influence of picture element signal noise.3rd, R/G and the B/G information of reference point are taken as decision element.
The region of reference point, non-close adjustable way, it is determined that in the way of variable minimum and maximum curve of approximation.
The present invention's is specific using technical method and sharp point, is illustrated below in conjunction with accompanying drawing.
Brief description of the drawings
The blackbody curve approximate region method of the existing white balances of Fig. 1 is summed up.
The aberration region method of the existing white balances of Fig. 2 is summed up.
Fig. 3 is often with UV the and ab coordinate position figures of 18% gray card under lamp box
Colour temperature of Fig. 4 blackbody radiations near visible spectral regions represents curve.
Fig. 5 often uses the light source frequency spectrum of lamp box
RGB relative spectral sensitivity map in Fig. 6 imaging sensors
Fig. 7 is often distributed with filtering characteristics of the IRCF in limit of visible spectrum
Transmission characteristic distribution of Fig. 8 camera lenses in limit of visible spectrum
Fig. 9 cameras R/G and B/G of 24 pieces of colour atlas under d 65 illuminant distribution map example
The practical example of gray card reference point R/G and the B/G relations of Figure 10 18%
Figure 11 often uses R/G and B/G coordinate examples under test chart D65 light sources
Figure 12 imaging sensor photographing modules simply constitute schematic diagram
The AWB detecting system flow chart of Figure 13 present invention
The AWB detecting system decision flow chart of Figure 14 present invention
The AWB detecting system testing result comparative example 1 of Figure 15 present invention
The AWB detecting system testing result comparative example 2 of Figure 16 present invention
Embodiment
Figure 12 imaging sensor photographing modules simply constitute schematic diagram.12-1 is that (AutoExpose exposes AE automatically in figure
Light), AWB (AutoWhiteBalance AWBs), AF (AutoFocus auto-focusings) are commonly called as 3A controls.In usual 3A
AWB AWBs in image procossing, such as cell phone platform image debug time in occupy significant portion.AWB
Accuracy directly affect the color reprodubility of image, while the result of AWB can influence the dark angle of camera lens
(Shading) in correction and image procossing as ColorMatrix being called, therefore the image quality of image can be influenceed.
The AWB detecting system flow chart of Figure 13 present invention.13-0 in Figure 13 receives to be passed by the image in Figure 12
The original stable image for removing black benchmark OB (OpticalBlack) for the process AE/AF controls that sensor is sent.13-1 in figure, according to
Picture portion domain, which is carried out, according to the parameter setting at initial stage of the A in Figure 13 calculates R/G and B/G values and scope.It is central, the A in Figure 13
Parameter setting at initial stage include：White portion minimum number, detects reference color temperature, and image segmentation number detects colour temperature precision, curve is most
Big value and minimum value, monochromatic scope setting, the green/colour of skin/blue spectrum etc., the setting of these first period parameters is extremely weighed
Will, according to the size of image, reference point choose result and change, so as to improve detection efficiency and accuracy of detection.13-2 in figure,
To the R/G of image, B/G scopes are compared with the monochromatic scope setting of parameter setting at initial stage, if less than monochromatic scope setting,
Illustrate that image graph picture compares dull coloring, number can be split with expanded view picture, be so conducive to improving detection efficiency.In Figure 13
13-3 is for the R/G of image regional, and whether B/G values are in cool colour light source and warm colour light source curve maximum and minimum value scope
Interior, the region outside curve maximum and minimum value scope, obvious is that colored areas does not do white balance processing, and curve
Region in the range of maximum and minimum value, including white/grey and and colored region.As with reference to 13-4 in figure with grey
World's mode calculates R, and B gain and colour temperature are used as backup.13-5 in figure, according to detection reference color temperature/detection colour temperature precision
Colour temperature calculating is carried out to regional, the colored areas in color temperature regions is excluded and excludes the colored areas in brightness,
13-6 finally judges that excluding white/gray area number behind colored areas 13-7 is to judging white/grey area in figure in figure
Domain carries out calculating R, B gain and colour temperature, and 13-8 is judged in figure result of determination.
White balance is in practice, and white balance is inaccurate and white balance concussion is subject matter.Figure 14 is the 13- in Figure 13
One embodiment that 8 result judgements are used.Figure 13 receives every frame images and carries out white balance judgement.To reduce white balance shake
Swing, result of determination is recorded successively in fig. 14.Shown in Figure 14 for judgement cycle that 3 two field pictures are white balance --- implement
The sequential of R, B gain.3 frames are an example, can be used as two field picture using more.14-1 in figure, judges that present image is
No is the 3rd frame, and no talks about, and 14-2 determines whether that the no of 14-2 is the first two field picture, 3 two field pictures in the second frame, figure in figure
There is 14-3 in the judgement for whether thering is white portion to detect, figure.14-4 and 14-5 in first and second two field pictures, figure, can be recorded
There is the result of calculation of the 13-7 in Figure 13 of white portion.When 3 two field picture, 14-6 in figure, if detecting white/grey
Region, and judge result of calculation in reference color temperature, then using 14-8 in current result of calculation figure, and more capable meter
Calculate the record of result.If the 3rd two field picture does not detect white/gray area, then into 14-7 in monochromatic process decision chart,
14-12 monochromes in 14-11 in 14-10 in green figure, blue figure, broca scale are carried out to judge.Here monochrome judges also only one
Application Example.There can be more other monochromatic judgements.And monochromatic institute's comparison range is set at Figure 13 initial stage in parameter setting
It is fixed.14-13 in figure, the monochromatic result of determination of record, the mark handled as back-end image.For monochrome image, it can determine whether
Record calculates the result of calculation of 14-14, i.e. the 1/th the 2nd two field picture in result figure, if historical record then returns record using preceding
14-15 in result of calculation figure, without use definition value at initial stage figure in 14-16.
The AWB detecting system testing result comparative example of Figure 15 present invention.The shooting environmental of figure in the middle of Figure 15
For indoor light source, background is wooden peony bookshelf, pink colour wall, pale yellow timber floor, yellow doll monkey, white flowerpot and green
Color vegetation.The left side is the image result of the cell phone platform of commercial type.The right is AWB detecting system of the invention
Testing result.Top in Figure 15 is the R/G in each region, B/G distribution map.15-1 judges white balance for the weighting of mobile phone in figure
15-2 is result of determination of the invention in point, figure.It can be seen that the weighting of 15-1 mobile phones judges white balance point in figure
Above and below be all that the number of point in consecutive variations, and figure above 15-1 is significantly less than in figure the number below 15-1,
For more than number, weight ratio is larger, and 15-2 is visible in the judgement point diagram of the contrast present invention, and point of the invention does not have deviation
More than number.The value of section RGB in comparison diagram shown in 15-3, below Figure 15, it can be seen that 15-4's is corresponding white in figure
The colors of RGB tri- of dyeing defect basin are not overlapped --- and white balance is forbidden, and the AWB detecting system testing result of the present invention, figure
The colors of RGB tri- of white flowerpot corresponding to middle 15-5 are completely superposed.
The AWB detecting system testing result comparative example of Figure 16 present invention.The shooting environmental of figure in the middle of Figure 16 is
Outdoor light source, background blue sky and sparse white clouds --- brightness is brighter, green leaf --- brightness is dark, electric pole --- brightness
It is dark.This brightness change has other colored scenes influenceed easily to produce than larger " purple sky " the problem of --- white clouds are not
In vain, color is blue, and blue sky is not blue.The left side is the image result of the cell phone platform of commercial type.The right is automatic for the present invention's
White balance detecting system testing result.Top in Figure 16 is the R/G in each region, B/G distribution map.16-1 is mobile phone in figure
Weighting judges that 16-2 is result of determination of the invention in white balance point, figure.The weighting of 16-1 mobile phones judges the lower-left of white balance point
Side --- all it is consecutive variations, although green most of parts of lower left are any limitation as by 16-6, still for green, in limit
Still there are many green points between region processed and decision-point, these green points are weighted participation white balance and calculated so that
White balance is inaccurate.As being mentioned in being described in prior figures 11, although pass through the week of the white reference point in R/G and B/G coordinates
Enclose setting and limit parameter or adjustable range equal weight, the judgement of " correct " can be produced, but this " correct " judgement is temporary
When and it is unstable, parameter setting can be quite numerous and diverse and needs repeated multiple times debugging, has a strong impact on efficiency.16-3 in comparison diagram
Below shown section RGB value, Figure 16, it can be seen that the colors of RGB tri- of 16-4 corresponding white clouds are not weighed in figure
Close --- white balance is forbidden, and in 16-2 in the AWB detecting system testing result figure of the present invention, figure corresponding to 16-5
The colors of RGB tri- of white clouds are completely superposed.
As described above, the present invention is by the selection to reference point, by the maximum for dividing cool colour light source and warm colour light source
The curve of approximation feature of value and minimum value, is combined using the characteristic of camera itself with blackbody curve characteristic, with finding out white
(netrual colour) region replaces existing region and Weight algorithm, the accuracy rate of the judgement of white balance is improved, while can reduce existing
The setting of parameter, improves debugging efficiency.
Claims (5)
1. the present invention relates to a kind of AWB detecting system, to the accuracy and judgement effect of the white balance for improving image
Really.
2. according to the AWB detecting system described in claim 1, in R/G, B/G coordinates, using warm colour source region and
Cool colour source region two, the approximate single black body emission curve of substitution., can be with according to the camera module characteristic of imaging sensor
Using more source region subregion two or more parts.
3. according to the AWB detecting system described in claim 1, in R/G, B/G coordinates, light source reference point uses non-envelope
Close mode.Coordinate R/G takes minimum and maximum region in direction.On ordinate B/G directions, open mode is taken, B/G is not limited
Upper lower threshold value.The present invention does not limit aforesaid way, and camera module that can also be according to imaging sensor can also B/G directions
Minimum and maximum region is taken, the mode of R/G upper lower threshold value is not limited.
4. according to the AWB detecting system described in claim 1, according to according to claim 2 and according to described in claim 3, adopt
With the minimum value curve of approximation of cool colour source region, maximum curve of approximation.And warm colour source region for minimum value it is approximate
Curve, maximum curve of approximation.
5. according to the AWB detecting system described in claim 1, according to detection reference color temperature/detection colour temperature precision to each
Region progress colour temperature calculating, excludes the colored areas in color temperature regions and excludes the colored areas in brightness, white to judgement/
Grey netrual colour region carries out calculating R, B gain.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610186408.1A CN107231549A (en) | 2016-03-25 | 2016-03-25 | AWB detecting system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610186408.1A CN107231549A (en) | 2016-03-25 | 2016-03-25 | AWB detecting system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107231549A true CN107231549A (en) | 2017-10-03 |
Family
ID=59932349
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610186408.1A Pending CN107231549A (en) | 2016-03-25 | 2016-03-25 | AWB detecting system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107231549A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107610073A (en) * | 2017-10-31 | 2018-01-19 | 努比亚技术有限公司 | A kind of image processing method and terminal, storage medium |
| CN108254159A (en) * | 2018-02-07 | 2018-07-06 | 北京镭创高科光电科技有限公司 | A kind of white balance detection device of laser display system |
| CN108280857A (en) * | 2017-12-28 | 2018-07-13 | 努比亚技术有限公司 | Camera method for parameter configuration, mobile terminal and computer readable storage medium |
| CN113691794A (en) * | 2020-05-18 | 2021-11-23 | 佳能株式会社 | Image processing apparatus, image processing method, and storage medium |
| CN114189666A (en) * | 2021-11-30 | 2022-03-15 | 信利光电股份有限公司 | Method for improving image quality effect of camera confusing color and shooting device |
-
2016
- 2016-03-25 CN CN201610186408.1A patent/CN107231549A/en active Pending
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107610073A (en) * | 2017-10-31 | 2018-01-19 | 努比亚技术有限公司 | A kind of image processing method and terminal, storage medium |
| CN107610073B (en) * | 2017-10-31 | 2020-10-16 | 努比亚技术有限公司 | Image processing method, terminal and storage medium |
| CN108280857A (en) * | 2017-12-28 | 2018-07-13 | 努比亚技术有限公司 | Camera method for parameter configuration, mobile terminal and computer readable storage medium |
| CN108254159A (en) * | 2018-02-07 | 2018-07-06 | 北京镭创高科光电科技有限公司 | A kind of white balance detection device of laser display system |
| CN113691794A (en) * | 2020-05-18 | 2021-11-23 | 佳能株式会社 | Image processing apparatus, image processing method, and storage medium |
| CN113691794B (en) * | 2020-05-18 | 2023-10-31 | 佳能株式会社 | Image processing apparatus, image processing method, and storage medium |
| CN114189666A (en) * | 2021-11-30 | 2022-03-15 | 信利光电股份有限公司 | Method for improving image quality effect of camera confusing color and shooting device |
| CN114189666B (en) * | 2021-11-30 | 2024-04-12 | 信利光电股份有限公司 | Method for improving mixed color image quality effect of camera and shooting device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108024055B (en) | White balance processing method and device, mobile terminal and storage medium | |
| Barnard et al. | A data set for color research | |
| CN109151426B (en) | White balance adjusting method and device, camera and medium | |
| JP3550440B2 (en) | Auto white balance adjustment device | |
| CN105827995B (en) | Automatic explosion method and system based on histogram | |
| CN107231549A (en) | AWB detecting system | |
| EP0658058B1 (en) | White balance adjustment apparatus | |
| JP6494181B2 (en) | Imaging device, control method thereof, and control program | |
| CN102892010B (en) | White balance processing method and device under multiple light sources | |
| CN103227928B (en) | White balance adjusting method and device | |
| WO2016122910A1 (en) | Dual camera systems and methods for rapid 3a convergence and high dynamic range exposure metering | |
| KR100478064B1 (en) | Illumination light color estimating method of obtaining white balance and image sensing apparatus | |
| JP5162905B2 (en) | Imaging device | |
| US9307213B2 (en) | Robust selection and weighting for gray patch automatic white balancing | |
| CN107483908A (en) | White balance calibration method and electronic equipment | |
| TWI740316B (en) | Day-mode-night-mode switching method and related monitoring camera apparatus | |
| US9479708B2 (en) | Image processing device, image processing method, and image processing program | |
| JP5482850B2 (en) | Imaging device | |
| CN114866754A (en) | Automatic white balance method and device, computer readable storage medium and electronic equipment | |
| CN107580205A (en) | Method and device for adjusting white balance | |
| CN115100085A (en) | Image color correction method, device, storage medium and electronic device | |
| CN114071107B (en) | Automatic white balance method and device based on fusion cluster analysis and color temperature curve | |
| CN118337969A (en) | Image white balance method, device, apparatus, medium, and program | |
| CN104780322B (en) | Photographic device, image capture method, image processing apparatus | |
| US20200228769A1 (en) | Lens rolloff assisted auto white balance |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171003 |