CN106295679A - A kind of coloured image light source colour method of estimation based on category correction - Google Patents

A kind of coloured image light source colour method of estimation based on category correction Download PDF

Info

Publication number
CN106295679A
CN106295679A CN201610606092.7A CN201610606092A CN106295679A CN 106295679 A CN106295679 A CN 106295679A CN 201610606092 A CN201610606092 A CN 201610606092A CN 106295679 A CN106295679 A CN 106295679A
Authority
CN
China
Prior art keywords
image
light source
edge feature
correction
feature
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.)
Granted
Application number
CN201610606092.7A
Other languages
Chinese (zh)
Other versions
CN106295679B (en
Inventor
李永杰
张明
高绍兵
任燕泽
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN201610606092.7A priority Critical patent/CN106295679B/en
Publication of CN106295679A publication Critical patent/CN106295679A/en
Application granted granted Critical
Publication of CN106295679B publication Critical patent/CN106295679B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
    • G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The invention discloses a kind of coloured image light source colour method of estimation based on category correction, first on the image of one group of known luminaire color, extract the edge feature of image, then learnt by method of least square, obtain the correction matrix between edge feature and light source, to pending test image zooming-out edge feature and it is multiplied with correction matrix again, obtains rough light source and estimate；By the way of finding K width adjacent image at feature space, find a class training image close with pending test characteristics of image afterwards, thus relearn, obtain light source accurately and estimate.The present invention relates to parameter few, and owing to the feature of extraction is simple and negligible amounts, so also having the features such as calculating is simple, speed is fast；Additionally, the present invention is method based on study, so high treating effect, degree of accuracy is high, is very suitable for the occasion that the accuracy of estimation of light source colour requires comparison high.

Description

A kind of coloured image light source colour method of estimation based on category correction
Technical field
The invention belongs to computer vision and technical field of image processing, be specifically related to a kind of colour based on category correction The design of image light source color method of estimation.
Background technology
Under natural environment, same object can present different colors, such as green under the irradiation of the light of different colours Leaves is inclined yellow under morning twilight irradiates, and time-division blueness partially between the lights.The visual system of people can resist this light source colour Change, thus the color of constant perceptual object, namely visual system has color constancy.But, limited by technical conditions System, machine does not has this ability, physical equipment, such as photographing unit the picture photographed is due to the change meeting of light source colour Produce serious colour cast.Therefore, how to remove, according to existing image information, the light source colour accurately estimating in scene and moved Remove thus obtain object color under standard white light and be just particularly important.
Computational color constancy is just directed to solve this problem, and its main purpose is to calculate any piece image institute The color of the unknown light source comprised, then carries out light source colour with this calculated light source colour image to being originally inputted Show under the white light of standard after correction, obtain so-called standard picture.Owing to standard picture eliminates light source colour Impact, thus for follow-up calculating task, such as based on color scene classification, image retrieval does not the most exist leads because of colour cast The misclassification caused or false retrieval.
Computational color constancy method can be divided into two classes: based on the method learnt and traditional static method.Traditional Static method extracts simple feature from image and estimates for light source, and this kind of method is bigger due to the error estimated, it is impossible to very Good meets requirement of engineering.Based on this demand, side based on study of being born on the basis of tradition is not based on the method for study Method, has the G.D.Finlayson method proposition in 2013 than more typical method based on study, list of references: G.D.Finlayson, " Corrected-moment illuminant estimation " in Proc.Comput.Vis.IEEE Int.Conf., 2013, pp.1904 1911, the method is by feature extraction and utilizes the method for recurrence to find feature and light source Between relation.This method is owing to employing the means of recurrence, thus estimates that the accuracy of light source is of a relatively high, but the method All images are all used same correction matrix, cause the light source error that parts of images estimates very big, thus cannot meet right Estimate the occasion that light source colour accuracy requirement is the highest, such as before the equipment receiving image of intelligent robot or automatic Pilot etc. End.Therefore, it is achieved a kind of method of correction matrix different to different image study is just particularly important.
Summary of the invention
The invention aims to solve image scene light source colour method of estimation of the prior art cannot meet right The problem estimating the occasion that light source colour accuracy requirement is the highest, it is proposed that a kind of coloured image light source face based on category correction Color method of estimation.
The technical scheme is that a kind of coloured image light source colour method of estimation based on category correction, including with Lower step:
S1, the edge feature of extraction training image: using the coloured image of N width known luminaire as original training set T, respectively Do convolution algorithm with template G after Gauss distribution derivation, obtain the marginal value that each pixel of image is corresponding, extract edge special Levy, obtain the edge feature matrix M of N width training image；
S2, learning correction matrix: by method of least square, learn to be instructed with N width by step S1 calculated eigenmatrix M Practice the correction matrix C between standard light source L of image；
S3, rough light source are estimated: use the method in step S1 to extract the edge feature of test image, with step S2 Acquistion to correction matrix C be multiplied, obtain rough light source estimated result L1；
S4, find the training image corresponding with testing image: test image is removed light respectively with original training set T Source, then use the method in S1 to extract edge feature respectively, form feature space；In feature space, find out and test image special Levy close K width training image, as new training set TN；
S5, accurately light source estimate: repeat step S1-S4, training set T in step S1 replaced with in step S4 every time New training set TN obtained, training image number is also become K from N accordingly, until the TN obtained in step S4 and last behaviour Till TN that in work, step S4 obtains is identical, the light source estimated result L1 that step S3 in last operation is obtained is as finally Light source estimated result.
Further, in step S1, template G after Gauss distribution derivation is Gauss gradient operator.
Further, the computing formula extracting edge feature in step S1 is:
M x y z = ( Σ i = 1 N 1 R i x G i y B i z N 1 ) 1 / ( x + y + z ) - - - ( 1 )
R in formulai、Gi、BiRepresent each pixel marginal value at tri-passages of R, G, B, N respectively1Represent pixel in image The number of point, MxyzFor different x, the value of edge feature corresponding under y, z, x, y, z are to meet x >=0, y >=0, z >=0 and x+y+z All combinations of=3.
Further, in step S4, the span of K is
Further, step S4 specifically include following step by step:
S41, original N width training image is removed standard light source L, and use method in step S1 to extract edge feature；
S42, to the light source L1 of rough estimate in test image removal step S3, and use the method in step S1 to extract limit Edge feature, the edge feature extracted with N width training image in step S41 is collectively forming feature space；
S43, in feature space, find out and test the characteristics of image the most close K width image of distance, as test image New training image set TN.
Further, the characteristic distance in step S43 is Euclidean distance.
The invention has the beneficial effects as follows: first the present invention extracts the edge of image on the image of one group of known luminaire color Feature, is then learnt by method of least square, obtains the correction matrix between edge feature and light source, then to pending Test image zooming-out edge feature is also multiplied with correction matrix, obtains rough light source and estimates；Afterwards by seeking at feature space The mode looking for K width adjacent image finds a class training image close with pending test characteristics of image, thus relearns, Estimate to light source accurately.Owing to pending test image is different in the distance of feature space from training image, suitably regulate The value of corresponding training image number K, can obtain being better suited for the result of dissimilar training image, and K is unique ginseng here Number.The present invention relates to parameter few (only parameter K), and owing to the feature of extraction is simple and negligible amounts, so also gathering around There are the features such as calculating is simple, speed is fast；Additionally, the present invention is method based on study, so high treating effect, degree of accuracy is high, It is very suitable for the occasion that the accuracy of estimation of light source colour requires comparison high, such as, is built in intelligent robot or automatically drives The front end etc. receiving vision facilities sailed.
Accompanying drawing explanation
A kind of based on category correction the coloured image light source colour method of estimation flow chart that Fig. 1 provides for the present invention.
Fig. 2 is the pending test image tools_ph-ulm.GIF of the embodiment of the present invention one.
Fig. 3 is the error amount schematic diagram between light source and the real light sources of each step estimation of the embodiment of the present invention one.
Fig. 4 is that the final light source estimated result of the embodiment of the present invention one contrasts schematic diagram with real light sources.
Fig. 5 is that the light source colour value utilizing step S5 to calculate of the embodiment of the present invention two carries out tone to original test image Result schematic diagram after correction.
Detailed description of the invention
Below in conjunction with the accompanying drawings embodiments of the invention are further described.
The invention provides a kind of coloured image light source colour method of estimation based on category correction, as it is shown in figure 1, include Following steps:
S1, the edge feature of extraction training image: using the coloured image of N width known luminaire as original training set T, respectively Do convolution algorithm with template G after Gauss distribution derivation, obtain the marginal value that each pixel of image is corresponding, extract edge special Levy, obtain the edge feature matrix M of N width training image.
Wherein, template G after Gauss distribution derivation is Gauss gradient operator.
The computing formula extracting edge feature is:
M x y z = ( Σ i = 1 N 1 R i x G i y B i z N 1 ) 1 / ( x + y + z ) - - - ( 1 )
R in formulai、Gi、BiRepresent each pixel marginal value at tri-passages of R, G, B, N respectively1Represent pixel in image The number of point, MxyzFor different x, the value of edge feature corresponding under y, z, x, y, z are to meet x >=0, y >=0, z >=0 and x+y+z All combinations of=3, total number of combinations is 19, so can obtain 19 edge features here.
S2, learning correction matrix: by method of least square, learn to be instructed with N width by step S1 calculated eigenmatrix M Practice the correction matrix C between standard light source L of image.
S3, rough light source are estimated: use the method in step S1 to extract the edge feature of test image, with step S2 Acquistion to correction matrix C be multiplied, obtain rough light source estimated result L1.
S4, find the training image corresponding with testing image: test image is removed light respectively with original training set T Source, then use the method in S1 to extract edge feature respectively, form feature space；In feature space, find out and test image special (span of K is to levy close K width training image), as new training set TN.
This step specifically include following step by step:
S41, original N width training image is removed standard light source L, and use method in step S1 to extract edge feature.
S42, to the light source L1 of rough estimate in test image removal step S3, and use the method in step S1 to extract limit Edge feature, the edge feature extracted with N width training image in step S41 is collectively forming feature space.
S43, in feature space, find out and test the characteristics of image the most close K width image of distance, as test image New training image set TN.
Wherein, the characteristic distance in step S43 is Euclidean distance.
S5, accurately light source estimate: repeat step S1-S4, training set T in step S1 replaced with in step S4 every time New training set TN obtained, training image number is also become K from N accordingly, until the TN obtained in step S4 and last behaviour Till TN that in work, step S4 obtains is identical, the light source estimated result L1 that step S3 in last operation is obtained is as finally Light source estimated result.
The final light source estimated result L1 of the image calculated after step S5 is used directly for follow-up meter Calculation machine vision is applied, and such as with the component of each Color Channel of the original color image of input divided by L1, can reach to remove coloured silk The purpose of light source colour in color image.Calculate additionally, the white balance of image and color correction are also required to use S5 Whole light source estimated result L1.
A kind of based on category correction the coloured image light source colour with a specific embodiment, the present invention provided below Method of estimation is described further:
Embodiment one:
Download all pictures of the most internationally recognized image library SFU object for estimating scene light source colour (altogether 321 width) and real light sources color (standard light source) L of correspondence, image size is 468 × 637, selects front 214 width of image library Image, as training set image, selects piece image tools_ph-ulm.GIF (as shown in Figure 2) conduct in residual image to wait to locate The test image of reason is tested, all images the most not through the pretreatment of any camera itself (such as tint correction, gamma Value correction).Then the detailed step of the present invention is as follows:
S1, the edge feature of extraction training image: using the coloured image of 214 width known luminaire as original training set T, point Template G (Gauss gradient operator) not and after Gauss distribution derivation does convolution algorithm, obtains the limit that each pixel of image is corresponding Edge value, then extract the edge feature of 19 dimensions respectively, finally obtain the edge feature matrix of the training set image that size is 214*19 M。
S2, learning correction matrix: by method of least square, study is by step S1 calculated eigenmatrix M and 214 width Correction matrix between standard light source L of training image, obtains the correction matrix C that size is 19*3:
C=[-150.0689 ,-30.1462 ,-21.5186；-96.5582 ,-196.1642 ,-348.5298；52.6551, 76.4461,115.5982；-200.5289 ,-240.3650 ,-179.6495；-79.6311,72.4539,125.1126；- 56.1276 ,-130.2963 ,-226.1518；683.9180,552.9035,366.8781；214.1444 ,-15.5379 ,- 52.8198；-149.3407,138.0260,397.1888；154.6218,240.3336,128.2161；156.5752 ,- 50.6503,69.4182；22.7103,90.3730,274.5781；-65.9786 ,-384.7642 ,-66.2556；- 112.7044 ,-104.0913 ,-12.8868；-349.7427,81.5115 ,-215.8972；-79.0109 ,-48.0727 ,- 32.2072；-98.2723 ,-22.7039 ,-51.2091；108.6481 ,-52.0896 ,-265.9989；172.6056, 171.2726,95.1991].
S3, rough light source are estimated: use the method in step S1 to extract 19 dimension edge features of test image, obtain big The edge feature matrix M1 of the little test image for 1*19:
M1=[0.0002,0.0004,0.0002,0.0014,0.0017,0.0012,0.0015,0.0013,0.0014, 0.0036,0.0040,0.0031,0.0037,0.0034,0.0039,0.0037,0.0032,0.0034,0.0035].
Again the correction matrix C that M1 obtains with the study of step S2 is multiplied, obtains rough light source estimated result L1= [0.1985,0.2151,0.2360].
S4, find the training image corresponding with testing image: to test image and the original training with 214 width images Collection T removes light source respectively, then uses the method in S1 to extract 19 dimension edge features respectively, forms feature space.At feature space In find out the K width training image close with testing characteristics of image, thus obtain a class image close with its feature, by this K width Image is as new training set TN.In the embodiment of the present invention, choose K=100.
This step specifically include following step by step:
S41, original 214 width training images are removed standard light sources L, and use method in step S1 to extract 19 dimension edges Feature, obtains the eigenmatrix M0 that size is 214*19.
S42, to the light source L1 of rough estimate in test image removal step S3, and use the method in step S1 to extract limit Edge feature, obtains the eigenmatrix M2 that size is 1*19:
M2=[0.0060,0.0089,0.0038,0.0366,0.0371,0.0224,0.0365,0.0282,0.0285, 0.0937,0.0900,0.0581,0.0921,0.0790,0.0909,0.0768,0.0673,0.0664,0.0777].
Again M2 is collectively forming feature space with the edge feature M0 of 214 width training images extractions in step S41.
S43, in feature space, find out and test the characteristics of image the most close 100 width images of distance, as test image New training image set TN.
S5, accurately light source estimate: repeat step S1-S4, training set T in step S1 replaced with in step S4 every time New training set TN obtained, training image number is also become K from N accordingly, until the TN obtained in step S4 and last behaviour Till TN that in work, step S4 obtains is identical, the light source estimated result L1 that step S3 in last operation is obtained is as finally Light source estimated result.
In the embodiment of the present invention, for saving time, repetitive operation twice.The light source that repetitive operation obtains the most afterwards is estimated Meter result is L1=[0.3412,0.3591,0.3168], and the light source estimated result obtained after repetitive operation twice is L1= [0.3312,0.3365,0.3430].The light source estimated result L1=that to perform to obtain after twice [0.3312,0.3365, 0.3430] as final light source estimated result.
As it is shown on figure 3, first pillar represents angular error between light source and the real light sources of rough estimate in step S3 Value, second pillar is the angle error value between light source and the real light sources estimated after repetitive operation once in step S5, 3rd pillar is the angle error value between light source and the real light sources that in step S5, repetitive operation is estimated for twice afterwards.Three The broken line connected between pillar has reacted the downward trend of estimation difference, shows that the light source estimated is more and more accurate.
It is illustrated in figure 4 under the final calculated three primary colors space of step S5 the direction of the red response with green component With real light sources redness and the direction of the response of green component, Fig. 4 shows by the calculated response value of step S5 and true field The information of scape light source colour very close to.
The light source estimated result finally given the present invention with a specific embodiment below with the tint correction of image is Simply demonstrating when example makees an actual application:
Embodiment two:
Utilize the light source colour value under step S5 each color component calculated, correct original input picture respectively The pixel value of each color component.A pixel (0.335,0.538,0.601) with the test image of input in step S3 As a example by, its correction after result be (0.335/0.3312,0.538/0.3365,0.601/0.3430)=(1.0115, 1.5988,1.7522), become after normalized (0.2319,0.3665,0.4016), then the value after correction is multiplied by Standard white backscatter extinction logarithmic ratioObtain (0.1339,0.2116,0.2319) correction chart picture as final output Pixel value, other pixel of original input picture also does similar calculating, finally obtains the coloured image after correction, as figure Shown in 5.
Those of ordinary skill in the art it will be appreciated that embodiment described here be to aid in reader understanding this Bright principle, it should be understood that protection scope of the present invention is not limited to such special statement and embodiment.This area It is each that those of ordinary skill can make various other without departing from essence of the present invention according to these technology disclosed by the invention enlightenment Planting concrete deformation and combination, these deform and combine the most within the scope of the present invention.

Claims (6)

1. a coloured image light source colour method of estimation based on category correction, it is characterised in that comprise the following steps:
S1, the edge feature of extraction training image: using the coloured image of N width known luminaire as original training set T, respectively with height Template G after this distribution derivation does convolution algorithm, obtains the marginal value that each pixel of image is corresponding, extracts edge feature, Edge feature matrix M to N width training image；
S2, learning correction matrix: by method of least square, learn by step S1 calculated eigenmatrix M and N width training figure Correction matrix C between standard light source L of picture；
S3, rough light source are estimated: use the method in step S1 to extract the edge feature of test image, learn with step S2 To correction matrix C be multiplied, obtain rough light source estimated result L1；
S4, find the training image corresponding with testing image: test image is removed light source respectively with original training set T, then Use the method in S1 to extract edge feature respectively, form feature space；Feature space is found out and tests characteristics of image phase Near K width training image, as new training set TN；
S5, accurately light source estimate: repeat step S1-S4, training set T in step S1 replaced with in step S4 every time and obtain New training set TN, training image number is also become K from N accordingly, until in step S4 the TN that obtains with in last operation Till TN that step S4 obtains is identical, the light source estimated result L1 that step S3 in last operation is obtained is as final light source Estimated result.
Coloured image light source colour method of estimation based on category correction the most according to claim 1, it is characterised in that institute Stating in step S1 template G after Gauss distribution derivation is Gauss gradient operator.
Coloured image light source colour method of estimation based on category correction the most according to claim 1, it is characterised in that institute State and step S1 is extracted the computing formula of edge feature be:
M x y z = ( Σ i = 1 N 1 R i x G i y B i z N 1 ) 1 / ( x + y + z ) - - - ( 1 )
R in formulai、Gi、BiRepresent each pixel marginal value at tri-passages of R, G, B, N respectively1Represent pixel in image Number, MxyzFor different x, the value of edge feature corresponding under y, z, x, y, z are to meet x >=0, y >=0, z >=0 and x+y+z=3's All combinations.
Coloured image light source colour method of estimation based on category correction the most according to claim 1, it is characterised in that institute Stating the span of K in step S4 is
Coloured image light source colour method of estimation based on category correction the most according to claim 1, it is characterised in that institute State step S4 specifically include following step by step:
S41, original N width training image is removed standard light source L, and use method in step S1 to extract edge feature；
S42, to the light source L1 of rough estimate in test image removal step S3, and it is special to use the method in step S1 to extract edge Levying, the edge feature extracted with N width training image in step S41 is collectively forming feature space；
S43, in feature space, find out and test the characteristics of image the most close K width image of distance, as the new instruction of test image Practice image collection TN.
Coloured image light source colour method of estimation based on category correction the most according to claim 5, it is characterised in that institute Stating the characteristic distance in step S43 is Euclidean distance.
CN201610606092.7A 2016-07-28 2016-07-28 A kind of color image light source colour estimation method based on category correction Active CN106295679B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610606092.7A CN106295679B (en) 2016-07-28 2016-07-28 A kind of color image light source colour estimation method based on category correction

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610606092.7A CN106295679B (en) 2016-07-28 2016-07-28 A kind of color image light source colour estimation method based on category correction

Publications (2)

Publication Number Publication Date
CN106295679A true CN106295679A (en) 2017-01-04
CN106295679B CN106295679B (en) 2019-06-25

Family

ID=57663052

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610606092.7A Active CN106295679B (en) 2016-07-28 2016-07-28 A kind of color image light source colour estimation method based on category correction

Country Status (1)

Country Link
CN (1) CN106295679B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110060308A (en) * 2019-03-28 2019-07-26 杭州电子科技大学 A kind of color constancy method based on light source colour distribution limitation
CN112995634A (en) * 2021-04-21 2021-06-18 贝壳找房(北京)科技有限公司 Image white balance processing method and device, electronic equipment and storage medium
CN116188797A (en) * 2022-12-09 2023-05-30 齐鲁工业大学 Scene light source color estimation method capable of being effectively embedded into image signal processor

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103258334A (en) * 2013-05-08 2013-08-21 电子科技大学 Method of estimating scene light source colors of color image
US20130243085A1 (en) * 2012-03-15 2013-09-19 Samsung Electronics Co., Ltd. Method of multi-view video coding and decoding based on local illumination and contrast compensation of reference frames without extra bitrate overhead

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130243085A1 (en) * 2012-03-15 2013-09-19 Samsung Electronics Co., Ltd. Method of multi-view video coding and decoding based on local illumination and contrast compensation of reference frames without extra bitrate overhead
CN103258334A (en) * 2013-05-08 2013-08-21 电子科技大学 Method of estimating scene light source colors of color image

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110060308A (en) * 2019-03-28 2019-07-26 杭州电子科技大学 A kind of color constancy method based on light source colour distribution limitation
CN110060308B (en) * 2019-03-28 2021-02-02 杭州电子科技大学 Color constancy method based on light source color distribution limitation
CN112995634A (en) * 2021-04-21 2021-06-18 贝壳找房(北京)科技有限公司 Image white balance processing method and device, electronic equipment and storage medium
CN112995634B (en) * 2021-04-21 2021-07-20 贝壳找房(北京)科技有限公司 Image white balance processing method and device, electronic equipment and storage medium
CN116188797A (en) * 2022-12-09 2023-05-30 齐鲁工业大学 Scene light source color estimation method capable of being effectively embedded into image signal processor
CN116188797B (en) * 2022-12-09 2024-03-26 齐鲁工业大学 Scene light source color estimation method capable of being effectively embedded into image signal processor

Also Published As

Publication number Publication date
CN106295679B (en) 2019-06-25

Similar Documents

Publication Publication Date Title
CN107103613B (en) A kind of three-dimension gesture Attitude estimation method
Hemrit et al. Rehabilitating the colorchecker dataset for illuminant estimation
CN104537625A (en) Bayer color image interpolation method based on direction flag bits
CN106384117B (en) A kind of vehicle color identification method and device
US9286525B2 (en) Method for evaluating an object recognition device of a motor vehicle
CN106295679A (en) A kind of coloured image light source colour method of estimation based on category correction
CN109643446A (en) Circuit device, electronic equipment and error-detecting method
JP6882874B6 (en) Ground change interpretation support device, ground change interpretation support method, and program
CN109754444B (en) Image coloring method and device
CN106097833A (en) A kind of Educational toy external member and digit recognition method thereof
CN103839270B (en) A kind of image matching method and device
US20140267782A1 (en) Apparatus And Method For Automated Self-Training Of White Balance By Electronic Cameras
CN116490901A (en) Device and method for analyzing marks included in facility plan
CN106296658B (en) A kind of scene light source estimation accuracy method for improving based on camera response function
CN102567969B (en) Color image edge detection method
CN104346817A (en) Video image color cast detection method and device
US20160225159A1 (en) Image processing method and image processing apparatus
US8320662B2 (en) Distinguishing colors of illuminated objects using machine vision
Funt et al. Does colour really matter? Evaluation via object classification
CN104751487B (en) A kind of moving target detecting method based on the plane discoloration frames of colored RGB tri- difference
CN106960188A (en) Weather image sorting technique and device
Koli et al. Image processing based language converter for deaf and dumb people
KR102169041B1 (en) Apparatus and method for determining the state of area including water in an image using color labeling
CN104463917B (en) Based on the normalized image vision conspicuousness detection method of division
Retzlaff et al. Combining synthetic image acquisition and machine learning: accelerated design and deployment of sorting systems

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant