CN107358255A - GAMMA based on convolutional neural networks ties up a classification learning method and device - Google Patents
GAMMA based on convolutional neural networks ties up a classification learning method and device Download PDFInfo
- Publication number
- CN107358255A CN107358255A CN201710504532.2A CN201710504532A CN107358255A CN 107358255 A CN107358255 A CN 107358255A CN 201710504532 A CN201710504532 A CN 201710504532A CN 107358255 A CN107358255 A CN 107358255A
- Authority
- CN
- China
- Prior art keywords
- gamma
- value
- little
- initial
- ties
- 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
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
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
Abstract
The invention discloses a kind of GAMMA based on convolutional neural networks to tie up a classification learning method, and the register configuration values for being tied up to the GAMMA of OLED modules a little carry out classification learning, and this method comprises the following steps：1) provide one and comprising target GAMMA tie up multiple initial luma values a little or/and multiple initial chromas and with each initial luma values or/and each data set of the corresponding multiple RGB register configuration values of the initial chroma；2) using the plurality of initial luma values or/and the plurality of initial chroma as input value, the plurality of RGB register configuration values are trained using convolutional neural networks to the data set as output valve, are obtained target GAMMA and are tied up GAMMA data models a little.The present invention is by tying up initial luma values or/and initial chroma and the classification learning of the relation of RGB register configuration values a little to each GAMMA, high-precision data model can be provided for the fast tuning of OLED module GAMMA curves, OLED module producing lines high-quality, efficient requirement can be met.
Description
Technical field
The present invention relates to OLED module GAMMA curve adjustment field, and in particular to a kind of based on convolutional neural networks
The relation of initial luma values and RGB register configuration values a little, or initial chroma and RGB are tied up to the GAMMA of OLED modules
The relation of register configuration values, or the relation of initial chroma, initial chroma and RGB register configuration values carry out taxology
The method and device of habit.
Background technology
OLED have wide self-luminous, clear beautiful, frivolous, fast response time, visual angle, low-power consumption, Applicable temperature scope it is big,
The features such as cost is low, manufacturing process is simple, it is considered as the novel flat-plate Display Technique that development potentiality is maximum after LCD, PDP.
In recent years, with OLED correlation techniques development and building the increase of production capacity, the application market of OLED products also expands rapidly, bag
Include TV, display, smart mobile phone, intelligence wearing, VR, automobile are shown, automotive lighting etc..
Instinctively, we will be considered that the expression of light and shade should equally spacedly correspond to corresponding brightness, but actually human eye
To the sensitivity ratio compared with brightness under dark situation in bright environment it is high, research find the sensation approximation direct ratio of human eye with
(1/ γ) power of brightness, this human eye feels that the relation curve between brightness is referred to as γ (GAMMA) curve, in order to preferably
So that OLED module displays effect more meets human eye vision curve, it is necessary to GAMMA corrections are done to module.Colour temperature is human eye to hair
The sensation of body of light or white reflection body, the color characteristics shown by OLED products are embodied in, standard blackbody is heated, temperature rise
The black matrix color starts dark red-pale red-orange yellow-white-indigo plant and gradually changed when to a certain extent, be typically chosen colour temperature for 6500K or
White field of the 9300K reference white as module, the chromaticity coordinate corresponding to 6500K is x=0.312, y=0.329, in order that obtaining
Unified colour temperature is presented in the white field of OLED modules, and a progress colourity regulation need to be tied up to the module IC each GAMMA bound.
At present, the GAMMA adjusting process of detection device manufacturer exploitation is shown both at home and abroad mainly for LCD modules, but due to
OLED modules are different from the luminescence mechanism of LCD modules so that are not fully appropriate for for the GAMMA adjusting process of LCD modules
OLED modules (regulating effect is not good enough, influences the product quality of OLED modules).In recent years, part also shows detection device manufacturer
GAMMA adjustment solution of the exploitation specifically for OLED modules is attempted, i.e., GAMMA each to OLED modules respectively ties up RGB a little
Register value is adjusted so that each GAMMA ties up brightness value a little and chromatic value falls in the error range of desired value, Ran Houzai
Repeatedly obtain each GAMMA and tie up brightness value and chromatic value a little, and adjust GAMMA and tie up RGB register values a little so that each GAMMA
Tie up brightness value a little and chromatic value is all fallen within the error range of desired value.But this GAMMA adjustment solutions for OLED modules
Certainly scheme has the following disadvantages：
1) this GAMMA adjustment solution for OLED modules is time-consuming longer, and needs one by one to enter OLED modules
Row GAMMA adjustment, it is less efficient, it is impossible to meet the requirement of OLED modules producing line batch production；
If 2) to lift GAMMA adjustment efficiency, this GAMMA adjustment solutions for OLED modules typically can only
Using the used time is less, simple algorithm, appraisal procedure, so as to cause GAMMA adjusting accuracies relatively low, and then OLED modules are influenceed
Product quality.
The content of the invention
For above-mentioned the deficiencies in the prior art, the present invention discloses a kind of GAMMA based on convolutional neural networks and ties up a classification
Learning method and device, initial luma values and RGB a little are tied up by each GAMMA to OLED modules based on convolutional neural networks
The relation of register configuration values, the either relation of initial chroma and RGB register configuration values or initial chroma, initial
The relation of chromatic value and RGB register configuration values carries out classification learning, and RGB registers a little can be tied up for each GAMMA of OLED modules
The rapid configuration of value provides high-precision data model, with meet OLED module producing line GAMMA adjustment high-quality, it is efficient will
Ask.
In order to solve the above technical problems, the present invention, which provides a kind of GAMMA based on convolutional neural networks, ties up a classification learning
Method, the register configuration values for being tied up to the GAMMA of OLED modules a little carry out classification learning, and this method comprises the following steps：
1) provide one comprising target GAMMA tie up multiple initial luma values a little or/and multiple initial chromas and with it is each
The data set of the initial luma values or/and the corresponding multiple RGB register configuration values of each initial chroma；
2) using the plurality of initial luma values or/and the plurality of initial chroma as input value, the plurality of RGB registers
Configuration Values are trained to the data set using convolutional neural networks as output valve, obtain target GAMMA and tie up a little
GAMMA data models.
Preferably, the obtaining step of the data set includes in above-mentioned technical proposal：
Obtain target GAMMA tie up all R, G a little, brightness value or/and chromatic value corresponding to the combination of B-register value, and
Take the minimum brightness value or/and chromatic value of error that target brightness value a little or/and aim colour angle value are tied up with target GAMMA
As initial luma values or/and initial chroma, R, G, B corresponding with brightness value or/and chromatic value that the error is minimum are taken
Register value combination is used as RGB register configuration values.
Preferably, the obtaining step of the data set includes in above-mentioned technical proposal：
Target GAMMA is obtained using CIE standard colorimetric systems and ties up all brightness values or/and chromatic value a little, and is taken
The minimum brightness value of error of target brightness value a little or/and aim colour angle value is tied up with target GAMMA or/and chromatic value is made
For initial luma values or/and initial chroma, R, G, B corresponding with brightness value or/and chromatic value that the error is minimum is taken to post
The combination of storage value is used as RGB register configuration values.
Preferably, this method is further comprising the steps of in above-mentioned technical proposal：
Initial value of the one initial network weight as the convolutional neural networks is provided；Wherein,
The target GAMMA for the OLED modules that the initial network weight includes other batches or other specifications ties up a little initial bright
The relation or initial color of angle value and the relation of RGB register configuration values, either initial chroma and RGB register configuration values
The relation of angle value, initial chroma and RGB register configuration values.
Preferably, the GAMMA data models include each initial luma values or/and each should in above-mentioned technical proposal
The network weight of initial chroma and each RGB register configuration values corresponding relations.
Preferably, above-mentioned technical proposal is trained the data set to obtain target GAMMA and tied up a little using neural network framework
Network weight.
In order to solve the above technical problems, the present invention additionally provides a kind of GAMMA based on convolutional neural networks to tie up a classification
Learning method, this method comprise the following steps：
1) obtained according to target GAMMA highest gray-scale intensity values, minimum gray scale brightness value and brightness-the GTG formula tied up a little
Target GAMMA ties up target brightness value and aim colour angle value a little；
2) obtain target GAMMA and tie up initial luma values, initial chroma a little, and target GAMMA is tied up R a little,
G, B-register value is adjusted, and the brightness value that target GAMMA is tied up a little is fallen in the error range of the target brightness value, color
Angle value falls in the error range of aim colour angle value, obtains RGB register configuration values corresponding with the brightness value, chromatic value；
3) using the initial luma values or/and the initial chroma as input value, the RGB register configuration values are as defeated
Go out value, the target GAMMA GAMMA data models tied up a little are trained using convolutional neural networks.
Preferably, the error range of the target brightness value is ± 2% in above-mentioned technical proposal；The error of the aim colour angle value
Scope is ± 0.001.
In order to solve the above technical problems, the present invention additionally provides a kind of GAMMA based on convolutional neural networks to tie up a classification
Learning device, including memory, processor and it is stored in the computer journey that can be run in the memory and on the processor
Sequence, the step of processor is configured as realizing above-mentioned technical proposal methods described when performing the computer program.
In order to solve the above technical problems, the present invention additionally provides a kind of computer-readable recording medium, this is computer-readable
Storage medium is stored with computer program, and the computer program realizes above-mentioned technical proposal methods described when being executed by processor
Step.
The beneficial effects of the present invention are：
1) present invention is deposited by the initial luma values tied up to the target GAMMA of substantial amounts of OLED modules sample a little with RGB
The relation of device Configuration Values carries out deep learning, or the relation of initial chroma and RGB register configuration values carries out deep learning,
Or the relation of initial chroma, initial chroma and RGB register configuration values carries out deep learning, can directly be OLED modules
Offer each GAMMA of rapid configuration ties up GAMMA data models a little, and then the GAMMA that can greatly lift OLED module producing lines is adjusted
School efficiency.
2) present invention accurately obtains each GAMMA using the algorithm of Various Complex, appraisal procedure and ties up original intensity a little
The relation or initial chroma of value and the relation of RGB register configuration values, either initial chroma and RGB register configuration values
The relation of value, initial chroma and RGB register configuration values, can not influence the feelings of OLED module producing line GAMMA adjustment efficiency
Under condition, the GAMMA adjusting accuracies and product quality of OLED modules are greatly lifted.
Brief description of the drawings
The GAMMA that Fig. 1 embodiment of the present invention one provides ties up a classification learning flow chart；
The GAMMA that Fig. 2 embodiment of the present invention two provides ties up a classification learning flow chart.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.As long as in addition, technical characteristic involved in each embodiment of invention described below
Conflict can is not formed each other to be mutually combined.
A kind of GAMMA based on convolutional neural networks disclosed by the invention ties up a classification learning method, for a collection of
The RGB register configuration values that each GAMMA of secondary OLED modules ties up a little carry out classification learning.
As shown in figure 1, embodiment one ties up RGB register configuration values a little with some target GAMMA of OLED modules
Classification learning process illustrates, and it comprises the following steps：
1) the step of target GAMMA ties up input/output relation data set a little is obtained：Substantial amounts of OLED modules sample is entered
Row GAMMA curve adjustments, and the target GAMMA for obtaining each OLED modules sample ties up initial luma values L a little0(or just
Beginning chromatic value X0Y0, or initial luma values L0With initial chroma X0Y0) and final RGB register configuration values；
2) the input/output relation data set tied up to target GAMMA a little is trained the step of obtaining GAMMA data models：
By these initial luma values L0As input value, using these RGB register configuration values as output valve, convolutional neural networks are utilized
It is trained to obtain target GAMMA and ties up GAMMA data models a little.
In above-described embodiment, ergodic theory method can be used to obtain the target GAMMA of substantial amounts of OLED modules sample and tied up
The input/output relation data set of point.Wherein, the target of any of which OLED module samples is obtained using ergodic theory method
GAMMA, which ties up the step of input/output relation data set a little, to be included：Obtain target GAMMA and tie up all R, G a little, B-register value
Combination and its corresponding brightness value L (or chromatic value XY, or brightness value L and chromatic value XY), and by all brightness
Value L obtains tying up the error of target brightness value a little with target GAMMA compared with the target brightness value that target GAMMA is tied up a little
Minimum brightness value L is as initial luma values L0, take R, G, B-register value corresponding to the brightness value L minimum with the error to combine
As RGB register configuration values.For example, OLED modules have tri- color adaptation options of R, G, B, each color adaptation option has again
33 are tied up a little, and the regulated value each tied up a little has 4096 kinds of possibility, then for target GAMMA tie up for a little have 4096*3 kinds R,
G, B-register value combine, then obtain respectively this 4096*3 kind R, G, B-register value combination corresponding to brightness value, and respectively with
The target brightness value that target GAMMA ties up a little is contrasted, and the error for obtaining tying up target brightness value a little with target GAMMA is minimum
One brightness value takes R, G, B-register value corresponding to the brightness value minimum with the error to combine and be used as RGB as initial luma values
Register configuration values.
In above-described embodiment, CIE standard colorimetrics system can also be used to obtain the input and output that target GAMMA is tied up a little and closed
It is data set：I.e. using the CIE marks such as " CIE 1931XYZ standard colorimetrics system ", " CIE 1931xyY standard colorimetrics system "
Quasi- colorimetry system acquisition target GAMMA ties up all chromatic value XY a little, and all chromatic value XY and target GAMMA are tied up a little
Aim colour angle value be compared, obtain tying up the minimum chromatic value XY conducts of error of aim colour angle value a little with target GAMMA
Initial chroma X0Y0, take R, G, B-register value corresponding to the chromatic value XY minimum with the error to combine and match somebody with somebody as RGB registers
Put value.
In above-described embodiment, it is necessary to according to the IC service manuals of OLED modules, search OLED modules IC support it is each
GAMMA ties up point, RGB register address.
In above-described embodiment, initial luma values, the initial chroma of OLED modules are gathered using optical tester, and is passed through
Pictcure generator is controlled to adjust R, G in the IC of OLED modules, B-register value, according to existing Gamma Calibration Technologies scheme to this
OLED modules carry out Gamma adjustment.
, can be with before input/output relation data set a little is tied up according to target GAMMA and is trained in above-described embodiment
The network weight for being a little trained to obtain using the target GAMMA of other batches or the OLED modules of other specifications is tied up is used as this
One initial value of GAMMA data models.
In above-described embodiment, in the GAMMA data models comprising a large amount of OLED modules samples target GAMMA tie up a little at the beginning of
Beginning brightness value L0(or initial chroma X0Y0, or initial luma values L0With initial chroma X0Y0) and RGB registers
The network weight of Configuration Values corresponding relation.
In above-described embodiment, using Tensorflow, or Caffe, or CNTK, or Theano, or Torch, or MXNet,
Or Chainer, or the neural network framework such as Keras trains target GAMMA to tie up input/output relation data set a little, is somebody's turn to do
Target GAMMA ties up network weight a little.
As shown in Fig. 2 embodiment two ties up RGB register configuration values a little with some target GAMMA of OLED modules
Classification learning process illustrates, and it comprises the following steps：
1) target is obtained according to target GAMMA highest gray-scale intensity, minimum gray scale and brightness-the GTG formula tied up a little
GAMMA ties up target brightness value and aim colour angle value a little；
2) obtain target GAMMA and tie up initial luma values L a little0, initial chroma X0Y0, and target GAMMA is tied up a little
R, G, B-register value be adjusted, the brightness value for making target GAMMA tie up a little falls the error range in the target brightness value
Interior, chromatic value falls in the error range of aim colour angle value, obtains RGB register configurations corresponding with the brightness value, chromatic value
Value；
3) by initial luma values L0Or/and initial chroma X0Y0As input value, the RGB register configuration values are made
For output valve, the target GAMMA GAMMA data models tied up a little are trained using convolutional neural networks.
In above-described embodiment, it is necessary to according to the IC service manuals of OLED modules, search OLED modules IC support it is each
GAMMA ties up point, RGB register address；Wherein, highest GTG W255 and minimum gray scale W0 is that fixed GAMMA is tied up a little, other ashes
Rank is tied up a little for middle GAMMA.
In above-described embodiment, initial luma values, the initial chroma of OLED modules are gathered using optical tester, and is passed through
Pictcure generator is controlled to adjust R, G in the IC of OLED modules, B-register value, according to existing Gamma Calibration Technologies scheme to this
OLED modules carry out Gamma adjustment.
In above-described embodiment, the white balance by adjusting OLED modules determines highest GTG W255 brightness value L v1 and most
Low GTG W0 brightness value L v2, and " Lv=(target GAMMA ties up a grey decision-making/W255) GAMMA refers to according to brightness-GTG formula
Number * (Lv1-Lv2)+Lv2 " obtain target GAMMA and tie up target brightness value Lv a little.
In above-described embodiment, the selection principle that target GAMMA ties up aim colour angle value a little is：When target is tied up a little as highest ash
During rank W255, because W255 GTGs are as maximum brightness benchmark, the required precision to chromatic value is higher, and chromaticity coordinates x scopes exist
Between 0.315 and 0.309 (± 0.003), chromaticity coordinates y scopes (± 0.003) between 0.332 and 0.326；Tied up when target and be a little
During 11-254 GTGs, its chromatic value is setting value, and 0.312 nearby (± 0.001), chromaticity coordinates y is near 0.329 by chromaticity coordinates x
(±0.001)；When it is a little 0-10 GTGs that target, which is tied up, its chromatic value is also setting value, chromaticity coordinates x 0.312 nearby (±
0.050), chromaticity coordinates y is near 0.329 (± 0.050).
In OLED module producing lines, the GAMMA data models that directly can be obtained using above-described embodiment are to OLED modules
GAMMA curve adjustment is carried out, its process includes：The target GAMMA of collection OLED modules ties up initial luma values L a little1(can also
It is initial chroma X1Y1Or initial luma values L1With initial chroma X1Y1), and by L1With in GAMMA data models
Each L0It is compared, obtains and L1The minimum L of error0, then the minimum L of this error is taken0Corresponding RGB registers
Configuration Values tie up RGB register configuration values a little for target GAMMA.
Or
The mistake of GAMMA curve adjustment is directly carried out to OLED modules using the GAMMA data models that above-described embodiment obtains
Journey includes：The target GAMMA of collection OLED modules ties up initial luma values L a little1(can also be initial chroma X1Y1, can also
It is initial luma values L1With initial chroma X1Y1), and by L1With each L in GAMMA data models0It is compared, obtains
With L1The minimum L of error0.If the error range be less than ± 2% (if what is selected is chromatic value, X1Y1With X0Y0Error
No more than ± 0.001), then take the minimum L of this error0Corresponding RGB register configuration values are that target GAMMA is tied up a little
RGB register configuration values；If the error range is more than ± 2%, according to highest gray-scale intensity values, minimum gray scale brightness value and bright
Degree-GTG formula obtains target GAMMA and ties up target brightness value or/and aim colour angle value a little, and is posted by adjusting R, G, B
Storage is worth to target GAMMA and ties up RGB register configuration values corresponding to target brightness value or/and aim colour angle value a little.
Embodiment three ties up a classification learning device for a kind of GAMMA based on convolutional neural networks, including：Processor, deposit
Reservoir and it is stored in the computer program that can be run in the memory and on the processor.The computing device institute
Realize all or part of flow in above-described embodiment method when stating computer program, such as obtain target GAMMA and tie up a little defeated
The step of entering output relation data set, the input/output relation data set tied up to target GAMMA a little, which are trained, obtains GAMMA numbers
The step of according to model.
In above-mentioned technical proposal, alleged processor can be CPU (Central Processing Unit,
CPU), it can also be other general processors, digital signal processor (Digital Signal Processor, DSP), special
Integrated circuit (Application Specific Integrated Circuit, ASIC), ready-made programmable gate array
(Field-Programmable Gate Array, FPGA) either other PLDs, discrete gate or transistor are patrolled
Collect device, discrete hardware components etc..General processor can be microprocessor or the processor can also be any conventional
Processor etc., the processor are that the GAMMA ties up a control centre for classification learning device, using various interfaces, circuit or
Signal logic connects whole GAMMA and ties up a various pieces for classification learning device.
In above-described embodiment, the memory can be used for storing the computer program, the processor by operation or
The computer program being stored in the memory is performed, and calls the data being stored in memory, realizes the GAMMA
Tie up a various functions for classification learning device.The memory can mainly include storing program area and storage data field, wherein, deposit
Storing up program area can storage program area, application program needed at least one function etc.；Storage data field can be stored comprising target
GAMMA ties up the data such as input/output relation data set and GAMMA data models a little.In addition, memory can be included at a high speed
Random access memory, nonvolatile memory, such as hard disk, internal memory, plug-in type hard disk, intelligent memory card can also be included
(Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash card (Flash Card), at least
One disk memory, flush memory device or other volatile solid-state parts.
In above-described embodiment, if the GAMMA is tied up a classification learning device and realized simultaneously in the form of SFU software functional unit
As independent production marketing or in use, can be stored in a computer read/write memory medium.Based on such reason
Solution, the present invention realize all or part of flow in above-described embodiment method, can also instruct correlation by computer program
Hardware complete, described computer program can be stored in a computer-readable recording medium, the computer program is in quilt
During computing device, can be achieved above-mentioned each embodiment of the method the step of.Wherein, the computer program includes computer program
Code, the computer program code can be source code form, object identification code form, executable file or some intermediate forms
Deng.The computer-readable medium can include：Any entity or device, the record of the computer program code can be carried
Medium, USB flash disk, mobile hard disk, magnetic disc, CD, computer storage, read-only storage (ROM, Read-Only Memory), with
Machine access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium etc..
As it will be easily appreciated by one skilled in the art that the content that this specification is not described in detail belongs to this area professional technique
Prior art known to personnel, these are only presently preferred embodiments of the present invention, be not intended to limit the invention, all in this hair
All any modification, equivalent and improvement made within bright spirit and principle etc., should be included in protection scope of the present invention
Within.
Claims (10)
1. a kind of GAMMA based on convolutional neural networks ties up a classification learning method, for being tied up a little to the GAMMA of OLED modules
Register configuration values carry out classification learning, it is characterised in that this method comprises the following steps：
1) provide one comprising target GAMMA tie up multiple initial luma values a little or/and multiple initial chromas and with each this is first
The data set of beginning brightness value or/and the corresponding multiple RGB register configuration values of each initial chroma；
2) using the plurality of initial luma values or/and the plurality of initial chroma as input value, the plurality of RGB register configurations
Value is used as output valve, and the data set is trained using convolutional neural networks, obtains target GAMMA and ties up GAMMA numbers a little
According to model.
2. GAMMA according to claim 1 ties up a classification learning method, it is characterised in that the obtaining step of the data set
Including：
Obtain target GAMMA tie up all R, G a little, brightness value or/and chromatic value corresponding to the combination of B-register value, and take with
Target GAMMA ties up brightness value or/and the chromatic value conduct of the error minimum of target brightness value or/and aim colour angle value a little
Initial luma values or/and initial chroma, R, G, B corresponding with brightness value or/and chromatic value that the error is minimum is taken to deposit
The combination of device value is used as RGB register configuration values.
3. GAMMA according to claim 1 ties up a classification learning method, it is characterised in that the obtaining step of the data set
Including：
Target GAMMA is obtained using CIE standard colorimetric systems and ties up all brightness values or/and chromatic value a little, and take with should
Target GAMMA ties up the brightness value of the error minimum of target brightness value or/and aim colour angle value a little or/and chromatic value is used as just
Beginning brightness value or/and initial chroma, take R, G corresponding with brightness value or/and chromatic value that the error is minimum, B-register
Value combination is used as RGB register configuration values.
4. GAMMA according to claim 1 ties up a classification learning method, it is characterised in that this method also includes following step
Suddenly：
Initial value of the one initial network weight as the convolutional neural networks is provided；Wherein,
The target GAMMA of OLED module of the initial network weight comprising other batches or other specifications ties up initial luma values a little
With the relation of RGB register configuration values, either the relation of initial chroma and RGB register configuration values or initial chroma,
The relation of initial chroma and RGB register configuration values.
5. GAMMA according to claim 1 ties up a classification learning method, it is characterised in that the GAMMA data models include
The network weight of each initial luma values or/and each initial chroma and each RGB register configuration values corresponding relations
Weight.
6. GAMMA according to claim 5 ties up a classification learning method, it is characterised in that is instructed using neural network framework
Practice the data set and obtain target GAMMA and tie up network weight a little.
7. a kind of GAMMA based on convolutional neural networks ties up a classification learning method, for being tied up a little to the GAMMA of OLED modules
Register configuration values carry out classification learning, it is characterised in that this method comprises the following steps：
1) mesh is obtained according to target GAMMA highest gray-scale intensity values, minimum gray scale brightness value and brightness-the GTG formula tied up a little
Mark GAMMA ties up target brightness value and aim colour angle value a little；
2) obtain target GAMMA and tie up initial luma values, initial chroma a little, and target GAMMA R, G, B tied up a little is posted
Storage value is adjusted, and the brightness value that target GAMMA is tied up a little is fallen in the error range of the target brightness value, chromatic value falls
In the error range of aim colour angle value, RGB register configuration values corresponding with the brightness value, chromatic value are obtained；
3) using the initial luma values or/and the initial chroma as input value, the RGB register configuration values as output valve,
The target GAMMA GAMMA data models tied up a little are trained using convolutional neural networks.
8. GAMMA according to claim 7 ties up a classification learning method, it is characterised in that the error of the target brightness value
Scope is ± 2%；The error range of the aim colour angle value is ± 0.001.
9. a kind of GAMMA based on convolutional neural networks ties up a classification learning device, including memory, processor and it is stored in
In the memory and the computer program that can run on the processor, it is characterised in that the processor is configured as performing should
Realized during computer program such as the step of any one of claim 1-8 methods described.
10. a kind of computer-readable recording medium, the computer-readable recording medium storage has computer program, and its feature exists
In when the computer program is executed by processor the step of realization such as any one of claim 1-8 methods describeds.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710504532.2A CN107358255A (en) | 2017-06-28 | 2017-06-28 | GAMMA based on convolutional neural networks ties up a classification learning method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710504532.2A CN107358255A (en) | 2017-06-28 | 2017-06-28 | GAMMA based on convolutional neural networks ties up a classification learning method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107358255A true CN107358255A (en) | 2017-11-17 |
Family
ID=60273082
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710504532.2A Pending CN107358255A (en) | 2017-06-28 | 2017-06-28 | GAMMA based on convolutional neural networks ties up a classification learning method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107358255A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109029918A (en) * | 2018-06-06 | 2018-12-18 | 武汉精测电子集团股份有限公司 | A kind of gamma adjusts the method and device tied up a little |
| CN109191386A (en) * | 2018-07-18 | 2019-01-11 | 武汉精测电子集团股份有限公司 | A kind of quick Gamma bearing calibration and device based on BPNN |
| CN110459170A (en) * | 2019-10-11 | 2019-11-15 | 武汉精立电子技术有限公司 | A kind of mould group Gamma bearing calibration, terminal device and computer-readable medium |
| CN113223453A (en) * | 2020-01-21 | 2021-08-06 | 武汉精立电子技术有限公司 | Module Gamma initial value prediction method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106328070A (en) * | 2016-09-28 | 2017-01-11 | 武汉精测电子技术股份有限公司 | OLED module Gamma adjustment and calibration method and device |
| CN106384573A (en) * | 2016-11-04 | 2017-02-08 | 武汉精测电子技术股份有限公司 | OLED module Gamma adjustment and calibration method based on linear interpolation calculation |
| CN106847173A (en) * | 2017-01-19 | 2017-06-13 | 武汉精测电子技术股份有限公司 | A kind of OLED modules gamma curve adjusting method and device |
-
2017
- 2017-06-28 CN CN201710504532.2A patent/CN107358255A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106328070A (en) * | 2016-09-28 | 2017-01-11 | 武汉精测电子技术股份有限公司 | OLED module Gamma adjustment and calibration method and device |
| CN106384573A (en) * | 2016-11-04 | 2017-02-08 | 武汉精测电子技术股份有限公司 | OLED module Gamma adjustment and calibration method based on linear interpolation calculation |
| CN106847173A (en) * | 2017-01-19 | 2017-06-13 | 武汉精测电子技术股份有限公司 | A kind of OLED modules gamma curve adjusting method and device |
Non-Patent Citations (1)
| Title |
|---|
| 陈敏, 华中科技大学出版社 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109029918A (en) * | 2018-06-06 | 2018-12-18 | 武汉精测电子集团股份有限公司 | A kind of gamma adjusts the method and device tied up a little |
| CN109029918B (en) * | 2018-06-06 | 2020-07-10 | 武汉精测电子集团股份有限公司 | Method and device for adjusting binding point by gamma |
| CN109191386A (en) * | 2018-07-18 | 2019-01-11 | 武汉精测电子集团股份有限公司 | A kind of quick Gamma bearing calibration and device based on BPNN |
| CN109191386B (en) * | 2018-07-18 | 2020-11-06 | 武汉精测电子集团股份有限公司 | BPNN-based rapid Gamma correction method and device |
| CN110459170A (en) * | 2019-10-11 | 2019-11-15 | 武汉精立电子技术有限公司 | A kind of mould group Gamma bearing calibration, terminal device and computer-readable medium |
| CN113223453A (en) * | 2020-01-21 | 2021-08-06 | 武汉精立电子技术有限公司 | Module Gamma initial value prediction method |
| CN113223453B (en) * | 2020-01-21 | 2022-07-19 | 武汉精立电子技术有限公司 | Module Gamma initial value prediction method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107346653A (en) | A kind of GAMMA curves adjusting process and device based on deep learning | |
| CN106384573B (en) | The OLED module Gamma adjusting process calculated based on linear interpolation | |
| CN106328070B (en) | OLED mould group Gamma adjusting process and device | |
| CN107358255A (en) | GAMMA based on convolutional neural networks ties up a classification learning method and device | |
| CN106297664B (en) | For the OLED module Gamma adjusting process of color sensitivity | |
| CN106847173B (en) | A kind of OLED mould group gamma curve adjusting method and device | |
| CN104821157B (en) | Show correction system | |
| CN104732949B (en) | Gamma electric voltage generative circuit, driving unit, display device and chromaticity coordinates adjusting method | |
| CN103929631B (en) | Use white balance adjustment method and the device thereof of scene detection | |
| CN110024020A (en) | Display device, calibrating installation and its calibration method | |
| CN103248793B (en) | The colour of skin optimization method of gamut conversion system and device | |
| CN108846351A (en) | Image processing method, device, electronic equipment and computer readable storage medium | |
| CN109559650A (en) | A kind of pixel rendering method and device, image rendering method and device, display device | |
| CN104200783B (en) | Method and system for enabling mobile phone to acquire uniform light effect through production line calibration | |
| CN105869580A (en) | Color temperature adjusting method and device, backlight source and display equipment | |
| CN102142224B (en) | Display device, uneven brightness correction method, correction data making device and correction data making method | |
| CN107680542A (en) | A kind of subregion method for controlling backlight thereof and its device | |
| CN113240112A (en) | Screen display adjusting method and device, electronic equipment and storage medium | |
| CN110197463A (en) | High dynamic range image tone mapping method and its system based on deep learning | |
| CN105632407B (en) | A kind of display adjusting method and mobile terminal of AMOLED display screens | |
| CN104115215B (en) | Video display devices and radiovisor | |
| CN105848354B (en) | A kind of shading process being used in intelligent illuminating system and system | |
| CN105872511B (en) | The method that flat panel TV colour temperature uniformity is improved using a plurality of gamma | |
| CN109584792A (en) | A kind of grayscale gamma fast adjustment method of OLED screen | |
| CN100492457C (en) | Color display device and color compensation method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| CB02 | Change of applicant information |
Address after: 430070 Hubei City, Hongshan Province, South Lake Road, No. 53, Hongshan Venture Center, building on the 4 floor, No. Applicant after: Wuhan fine test electronics group Limited by Share Ltd Address before: 430070 Hubei City, Hongshan Province, South Lake Road, No. 53, Hongshan Venture Center, building on the 4 floor, No. Applicant before: Wuhan Jingce Electronic Technology Co., Ltd. |
|
| CB02 | Change of applicant information | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |