CN109839189A - Utilize the method for multispectral camera self-adapting reconstruction spectral reflectance - Google Patents
Utilize the method for multispectral camera self-adapting reconstruction spectral reflectance Download PDFInfo
- Publication number
- CN109839189A CN109839189A CN201811611765.3A CN201811611765A CN109839189A CN 109839189 A CN109839189 A CN 109839189A CN 201811611765 A CN201811611765 A CN 201811611765A CN 109839189 A CN109839189 A CN 109839189A
- Authority
- CN
- China
- Prior art keywords
- test sample
- sample
- spectral reflectance
- response
- training
- 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
Abstract
The invention discloses a kind of methods using multispectral camera self-adapting reconstruction spectral reflectance, include the following steps: 1. multispectral images that all training samples are shot using multispectral camera, obtain the multichannel camera response of all training samples；2. keeping multispectral camera position constant, the multispectral image of all test samples is shot, the multichannel camera response of test sample is obtained；3. calculating the correlation of each test sample with each training sample in multichannel response vector space, and obtain corresponding spectral reflectance.The present invention has the characteristics that can be improved test sample spectral reflectance reconstruction accuracy.
Description
Technical field
The present invention relates to imaging types to obtain object spectra and chrominance information technical field, can be improved more particularly, to one kind
The method using multispectral camera self-adapting reconstruction spectral reflectance of test sample spectral reflectance reconstruction accuracy.
Background technique
Due to the presence of metamerism phenomenon, the chrominance information of object is difficult to reliably characterize this source information of object, and object
The spectral reflectance of body is unrelated and unrelated with the capture light source physical quantity of equipment, can characterize this source information of object.Therefore,
High-precision color measuring may be implemented in the accurate spectral reflectance for obtaining object, and loyal can reappear object in any light source
Under color appearance.Multispectral camera is imaged due to increasing more channels, therefore spectral reflectance acquisition can be improved
Precision.
Existing multispectral camera Spectral Reconstruction algorithm is mostly to establish camera using training sample based on training sample
Transformational relation between response and spectral reflectance.However, the training sample of existing multispectral camera and the light of test sample
It composes reflection characteristic and differs larger, Spectral Reconstruction effect is poor, seriously affects test sample spectral reflectance reconstruction accuracy.
Therefore, it is adaptive to design a kind of utilization multispectral camera that can be improved test sample spectral reflectance reconstruction accuracy
The method for reconstructing spectral reflectance, just seems very necessary.
Summary of the invention
The present invention is the existing multispectral camera in order to overcome in the prior art, is asked there are Spectral Reconstruction precision is poor
Topic provides a kind of utilization multispectral camera self-adapting reconstruction spectrum that can be improved test sample spectral reflectance reconstruction accuracy
The method of reflectivity.
To achieve the above object, the invention adopts the following technical scheme:
A method of using multispectral camera self-adapting reconstruction spectral reflectance, include the following steps:
(1-1) shoots the multispectral image of all training samples using multispectral camera, obtains the more of all training samples
Channel camera response；
(1-2) keeps multispectral camera position constant, shoots the multispectral image of all test samples, obtains test sample
Multichannel camera response；
(1-3) it is related to each training sample to calculate each test sample in multichannel response vector space
Property, and obtain corresponding spectral reflectance.
The present invention is the closer sample of the adaptively selected spectral characteristic of test sample as training sample, and utilizes selection
Training sample establish multichannel response to the transition matrix between spectral reflectance, the transition matrix is more suitable for testing
Sample.The present invention has the characteristics that can be improved test sample spectral reflectance reconstruction accuracy.
Preferably, using multispectral camera self-adapting reconstruction spectral reflectance method the step of (1-3) further include as
Lower step:
(1-3-1) utilizes following formula, calculates the correlation of each test sample with each training sample:
Wherein, piIndicate the multichannel response of i-th of sample in test sample, pjIndicate j-th of sample in training sample
Multichannel response,WithRespectively indicate i-th of test sample and j-th of training sample multichannel response vector itself
Average value, cijIndicate the correlation of i-th of test sample and j-th of training sample.
Preferably, the step of for the method for measurement insect bodies table spectral reflectance indirectly (1-3) further includes walking as follows
It is rapid:
(1-3-2) for i-th of test sample, by cijValue all training sample descendings are arranged, select before m training
Sample selects m training sample for having maximum correlation with i-th of test sample, then utilize this m training sample
The spectral reflectance of multichannel response and i-th of test sample establishes transition matrix between the two, and utilizes the conversion
The multichannel response p of matrix and test sampleiCalculate corresponding spectral reflectance.
Preferably, step (1-3) further includes following steps:
(1-3-3) picks out phase by step (1-3-2) for each test sample except i-th of test sample respectively
Then the maximum m training sample of closing property utilizes the spectrum of the multichannel response of this m training sample and the test sample
Reflectivity establishes transition matrix between the two, and utilizes the calculating pair of the multichannel response of the transition matrix and test sample
The spectral reflectance answered.
Preferably, the multispectral camera is that colour filter is wheeled, n channel is formed by n interference filter.
It therefore, is that the adaptively selected spectral characteristic of test sample more connects the invention has the following beneficial effects: (1) present invention
Close sample is established multichannel response using the training sample of selection and is turned between spectral reflectance as training sample
Matrix is changed, which is more suitable for test sample, to improve the spectral reflectance reconstruction accuracy of test sample；(2)
It is simple to operate, it is easy to implement.
Detailed description of the invention
Fig. 1 is a kind of flow chart of the invention.
Specific embodiment
The present invention will be further described with specific embodiment with reference to the accompanying drawing:
Embodiment: multispectral camera of the present invention is that colour filter is wheeled, forms 8 by 8 interference filters and leads to
Road, the full widths at half maximum of 8 colour filters are 20nm, peak transmittance wavelength be respectively 420nm, 460nm, 500nm, 540nm,
580nm,620nm,660nm,700nm.Using 168 color lumps in DC colour atla, surrounding and center in primary colors card is given up and have repeated
Black, white, grey block and 8 glossiness color lumps.Using odd number color lump as training sample, even number color lump is as test specimens
This, then trained and test sample respectively has 84.Based on training sample, multichannel response is calculated to spectrum using the wiener estimation technique
The transition matrix of reflectivity.
A kind of method using multispectral camera self-adapting reconstruction spectral reflectance as shown in Figure 1, includes the following steps:
(1-1) shoots the multispectral image of 84 training samples using multispectral camera, and obtain 84 training samples 8 are logical
Road camera response；
(1-2) keeps multispectral camera position constant, shoots the multispectral image of 84 test samples, obtains test sample
8 channel camera responses；
(1-3) it is related to each training sample to calculate each test sample in 8 channel response value vector spaces
Property, and obtain corresponding spectral reflectance:
(1-3-1) utilizes following formula, calculates the correlation of each test sample with each training sample:
Wherein, piIndicate 8 channel response value vectors of i-th of sample in test sample, pjIt indicates in training sample j-th
8 channel response value vectors of sample,WithRespectively indicate i-th of test sample and j-th of 8 channel response value of training sample
The average value of vector itself, cijIndicate the correlation of i-th of test sample and j-th of training sample；
(1-3-2) for i-th of test sample, by cijValue all training sample descendings are arranged, select it is preceding 30 instruct
Practice sample, that is, selects 30 training samples that there is maximum correlation with i-th of test sample, then utilize this 30 trained samples
The spectral reflectance of this 8 channel response values and i-th of test sample establishes transition matrix between the two, and is turned using described
Change 8 channel response value p of matrix and test sampleiCalculate corresponding spectral reflectance.
(1-3-3) picks out phase by step (1-3-2) for each test sample except i-th of test sample respectively
Then maximum 30 training samples of closing property utilize the light of the multichannel response of this 30 training samples and the test sample
The transition matrix of reflectivity foundation between the two is composed, and is calculated using the multichannel response of the transition matrix and test sample
Corresponding spectral reflectance.
It is computed, for 84 test samples, does not use the average color difference of the spectral reflectance of self-adapting reconstruction for 1.37
Δ E00 (CIEDE2000 color difference unit), and use the average color difference of the spectral reflectance of self-adapting reconstruction for 1.07 Δ E00, it demonstrate,proves
The feasibility and validity of this method is illustrated.
It should be understood that this embodiment is only used to illustrate the invention but not to limit the scope of the invention.In addition, it should also be understood that,
After having read the content of the invention lectured, those skilled in the art can make various modifications or changes to the present invention, these etc.
Valence form is also fallen within the scope of the appended claims of the present application.
Claims (5)
1. a kind of method using multispectral camera self-adapting reconstruction spectral reflectance, characterized in that include the following steps:
(1-1) shoots the multispectral image of all training samples using multispectral camera, obtains the multichannel of all training samples
Camera response；
(1-2) keeps multispectral camera position constant, shoots the multispectral image of all test samples, obtains the more of test sample
Channel camera response；
(1-3) calculates the correlation of each test sample with each training sample in multichannel response vector space, and
Obtain corresponding spectral reflectance.
2. the method according to claim 1 using multispectral camera self-adapting reconstruction spectral reflectance, characterized in that step
Suddenly (1-3) further includes following steps:
(1-3-1) utilizes following formula, calculates the correlation of each test sample with each training sample:
Wherein, piIndicate the multichannel response of i-th of sample in test sample, pjIndicate the more of j-th sample in training sample
Channel response value,WithRespectively indicate the flat of i-th of test sample and j-th training sample multichannel response vector itself
Mean value, cijIndicate the correlation of i-th of test sample and j-th of training sample.
3. the method according to claim 2 using multispectral camera self-adapting reconstruction spectral reflectance, characterized in that step
Suddenly (1-3) further includes following steps:
(1-3-2) for i-th of test sample, by cijValue all training sample descendings are arranged, select before m trained sample
This, that is, select m training sample for having maximum correlation with i-th of test sample, then utilizes the more of this m training sample
The spectral reflectance of channel response value and i-th of test sample establishes transition matrix between the two, and utilizes the conversion square
The multichannel response p of battle array and test sampleiCalculate corresponding spectral reflectance.
4. the method according to claim 3 using multispectral camera self-adapting reconstruction spectral reflectance, characterized in that step
Suddenly (1-3) further includes following steps:
(1-3-3) picks out correlation by step (1-3-2) for each test sample except i-th of test sample respectively
Then maximum m training sample utilizes the spectral reflectance of the multichannel response of this m training sample and the test sample
Than establishing transition matrix between the two, and it is corresponding using the calculating of the multichannel response of the transition matrix and test sample
Spectral reflectance.
5. the method according to claim 1 or 2 or 3 or 4 using multispectral camera self-adapting reconstruction spectral reflectance,
It is characterized in, the multispectral camera is that colour filter is wheeled, forms n channel by n interference filter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811611765.3A CN109839189A (en) | 2018-12-27 | 2018-12-27 | Utilize the method for multispectral camera self-adapting reconstruction spectral reflectance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811611765.3A CN109839189A (en) | 2018-12-27 | 2018-12-27 | Utilize the method for multispectral camera self-adapting reconstruction spectral reflectance |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109839189A true CN109839189A (en) | 2019-06-04 |
Family
ID=66883446
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811611765.3A Pending CN109839189A (en) | 2018-12-27 | 2018-12-27 | Utilize the method for multispectral camera self-adapting reconstruction spectral reflectance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109839189A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110660112A (en) * | 2019-09-29 | 2020-01-07 | 浙江大学 | Drawing spectrum reconstruction method based on special color card and multispectral imaging |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106153192A (en) * | 2016-07-22 | 2016-11-23 | 浙江大学 | A kind of method utilizing multispectral camera virtual responsive value to obtain spectral reflectance |
| CN106841055A (en) * | 2017-03-22 | 2017-06-13 | 浙江大学 | A kind of training sample selection method for reconstructing art drawing spectrum picture |
-
2018
- 2018-12-27 CN CN201811611765.3A patent/CN109839189A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106153192A (en) * | 2016-07-22 | 2016-11-23 | 浙江大学 | A kind of method utilizing multispectral camera virtual responsive value to obtain spectral reflectance |
| CN106841055A (en) * | 2017-03-22 | 2017-06-13 | 浙江大学 | A kind of training sample selection method for reconstructing art drawing spectrum picture |
Non-Patent Citations (1)
| Title |
|---|
| 徐鹏: "基于多光谱成像技术的光谱图像重构与显示研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110660112A (en) * | 2019-09-29 | 2020-01-07 | 浙江大学 | Drawing spectrum reconstruction method based on special color card and multispectral imaging |
| CN110660112B (en) * | 2019-09-29 | 2021-09-24 | 浙江大学 | Drawing spectrum reconstruction method based on special color card and multispectral imaging |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106548467B (en) | The method and device of infrared image and visual image fusion | |
| Van De Weijer et al. | Edge-based color constancy | |
| CN108020519B (en) | Virtual multi-light-source spectrum reconstruction method based on color constancy | |
| CN101454806B (en) | Method and apparatus for volume rendering using depth weighted colorization | |
| CN111047539A (en) | Fabric image color calibration algorithm based on spectral reflectivity reconstruction | |
| KR101291219B1 (en) | Method for data fusion of panchromatic and multi-spectral images and apparatus thereof | |
| CN108109201B (en) | Three-dimensional reconstruction method and system for complex color surface object | |
| CN107730493A (en) | Product colour difference detecting method, device, medium and computer equipment | |
| CN109738067B (en) | Method for estimating synthesized spectral sensitivity of narrow-band multispectral camera | |
| CN101933321A (en) | Image sensor apparatus and method for scene illuminant estimation | |
| DiCarlo et al. | Illuminating illumination | |
| CN108195312B (en) | Method and system for three-dimensional reconstruction of colored object based on dynamic weight | |
| CN103091615B (en) | Method to measure response curve of image sensor and device | |
| CN105976351A (en) | Central offset based three-dimensional image quality evaluation method | |
| CN106153192B (en) | A kind of method that spectral reflectance is obtained using multispectral camera virtual responsive value | |
| CN106841055B (en) | A kind of training sample selection method of reconstruct art drawing spectrum picture | |
| CN109839189A (en) | Utilize the method for multispectral camera self-adapting reconstruction spectral reflectance | |
| CN109253862A (en) | A kind of colour measurement method neural network based | |
| CN108510474B (en) | Method and system for evaluating quality of tobacco leaf image, memory and electronic equipment | |
| CN101893481B (en) | Method and system for testing and determining color attribute | |
| CN109712093A (en) | A kind of color of image restoring method and device based on sky and ocean background | |
| CN108680535A (en) | Based on the spectral reflectance recovery method for improving R matrixes | |
| CN106023238A (en) | Color data calibration method for camera module | |
| CN112651945A (en) | Multi-feature-based multi-exposure image perception quality evaluation method | |
| Urban et al. | Recovering camera sensitivities using target-based reflectances captured under multiple LED-illuminations |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190604 |
|
| RJ01 | Rejection of invention patent application after publication |