CN107766896B - Spectrum dimensionality reduction method based on hue clustering - Google Patents
Spectrum dimensionality reduction method based on hue clustering Download PDFInfo
- Publication number
- CN107766896B CN107766896B CN201711221418.5A CN201711221418A CN107766896B CN 107766896 B CN107766896 B CN 107766896B CN 201711221418 A CN201711221418 A CN 201711221418A CN 107766896 B CN107766896 B CN 107766896B
- Authority
- CN
- China
- Prior art keywords
- hue
- sample
- color
- saturation
- tested
- 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.)
- Active
Links
Images
Classifications
-
- 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
-
- 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
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- 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
- G06V10/58—Extraction of image or video features relating to hyperspectral data
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Spectrometry And Color Measurement (AREA)
Abstract
The invention relates to a spectrum dimensionality reduction method based on hue clustering, which is characterized by comprising the following steps of: selecting and starting from the characteristics of the spectrum sample, determining the main hues according to the color characteristics of the training samples and the hue circle principle, determining the training sample corresponding to each main hue by utilizing hue clustering, and then performing the spectral reflectivity dimension reduction treatment on the sample to be tested. The invention fully considers the requirements of practical application, and has the characteristics of simple calculation, easy operation, less characteristic vectors, high dimension reduction precision and the like; and the multispectral color information is compressed and stored with high precision, so that the use is more convenient for users.
Description
Technical Field
The invention belongs to the technical field of cross-media color reproduction based on spectrum in color science technology, and particularly relates to a method for reducing dimension and compressing multispectral data, in particular to a spectrum dimension reducing method based on hue clustering.
Background
In cross-media color reproduction, one often requires that the color visual perception between the reproduced color and the original color be consistent under various circumstances, i.e., achieve a "what you see is what you get" color matching goal. Therefore, quantitative description and matching of colors has been an important research point in color science. The traditional color quantitative description and matching is mainly based on the Grassmann law, the color is quantitatively described in a three-dimensional color space form by simulating the characteristics of a human visual system, and the color matching is carried out based on the color space; the color matching is based on chromaticity and belongs to condition matching, the pursued standard is color matching under certain fixed observation conditions, when the observation conditions or the light source are changed, the color perception of the color is changed, and the color is not matched any more. The other way of quantitatively describing the color is the spectral characteristics of the object, can realize color matching under any condition, belongs to unconditional matching, can realize the true color matching target of 'what you see is what you get', and fundamentally ensures the consistency of color reproduction across media.
Spectral-based color reproduction techniques are based on spectral matching, always using spectral characteristics of objects instead of colorimetric values for the communication of color information. The color information is transmitted and reproduced without depending on the specific observation environment and the influence of the light source. The method is a fundamental way for truly realizing cross-media color reproduction, and brings essential promotion to the current color reproduction technology.
Since the obtained spectral color information is multi-dimensional data, there are problems of methods such as data redundancy, large calculation amount, large storage space, and the like in the process of being used for color reproduction and color transfer. In many application scenarios (e.g. artwork reproduction, computer vision, e-commerce, etc.), multi-dimensional spectroscopy is not suitable for direct spectral-image processing, cross-media color reproduction, spectral gamut mapping. The spectral dimension reduction process is a process of reducing the space of high-dimensional spectral data to a low-dimensional space by using the correlation between spectral reflectivity samples or spectral wave bands and a dimension reduction method, and time complexity and processing complexity required in the color processing process are reduced. Establishing a spectral reflectivity dimension reduction method facing color reproduction is an important link for realizing color information transmission and communication.
The multispectral dimensionality reduction methods currently applied to color reproduction are not too many, and can be basically divided into two main categories according to the dimensionality reduction principle: one is metamerism black compensation; the other is multivariate statistical analysis. The metamerism black compensation method is mainly characterized in that the metamerism black compensation method can be decomposed into a basic spectrum and a metamerism black spectrum according to color spectrum information; the basic spectrum determines the chromaticity information of the color, the metamerism black spectrum determines the precision of the spectrum, and the main algorithms of the method include a LabPQR dimension reduction method, a LabRGB dimension reduction method, a XYZLS dimension reduction method and the like; however, this method requires a given light source condition, and is not robust to color differences when the illumination is changed. The multivariate statistical analysis method is mainly used for reducing the dimension of the spectral data by utilizing a multivariate statistical theory based on the correlation among the spectral data or the correlation among wave bands, and comprises a principal component analysis method, a nonlinear principal component method, an independent principal component analysis method and the like; however, the method only starts from the sample global, and does not consider the characteristics of the spectrum sample, so that the spectral dimension reduction precision is low. Based on the reasons, the invention provides a method for reducing the dimension of the spectral data by using the sample characteristics of hue clustering to realize the dimension reduction and compression of the multispectral data.
Disclosure of Invention
In order to make up for the defects of the prior art, the invention provides a dimensionality reduction and compression method of multispectral data with the characteristics of simple calculation, easy operation, less characteristic vectors, high dimensionality reduction precision and the like, and a spectral dimensionality reduction system based on hue clustering is constructed by the dimensionality reduction and compression method.
The invention is realized by the following technical scheme:
a spectrum dimensionality reduction method based on hue clustering is characterized by comprising the following steps:
step S1, determining the main hues according to the color characteristics of the training samples and the hue circle principle, and determining the training sample subset corresponding to each main hue by utilizing hue clustering; the number of the obtained partitions is m, each partition corresponds to a subset of spectral values with the same color characteristics, and the hue angle and the saturation of each training sample are calculated.
Step S2, determining whether to divide the low saturation color partition:
(a) if the low saturation color subareas are not divided, determining the hue angle range corresponding to each subarea, marking the subareas according to the sequence of the hue angle ranges from small to large, wherein the subareas are respectively H (1), H (2) and H (3) … H (m), and the corresponding spectral value subsets of each subarea are respectively marked as A (1), A (2) and A (3) … A (m).
(b) If the low saturation color partition is divided, a saturation threshold corresponding to the low saturation color partition is determined. And when the saturation corresponding to the training sample is smaller than the saturation threshold value, the low-saturation color partition belongs to the low-saturation color partition, when the saturation corresponding to the training sample is not smaller than the saturation threshold value, the low-saturation color partition belongs to other partitions, the low-saturation color partition is marked as H (1), and the corresponding spectral value set is marked as A (1). And determining the hue angle ranges corresponding to other partitions, marking the hue angle ranges from small to large as H (2), H (3) and H (4) … H (m), and recording the sets of the spectral values corresponding to the partitions as A (2), A (3) and A (4) … A (m).
Step S3, obtaining a characteristic vector of each subarea by adopting a multivariate statistical analysis method for the spectral value set of each subarea; fully considering the actual production requirement, and selecting the corresponding quantity of the feature vectors as the dimension reduction dimension of the partition;
step S4, judging the partition where the sample to be tested is located by utilizing hue clustering, and performing spectrum dimensionality reduction on the sample to be tested by utilizing the characteristic vector of the partition; and finally, spectrum reconstruction is carried out by searching the partition to obtain the spectrum reflectivity of the sample to be tested after reconstruction.
The spectrum dimensionality reduction method based on hue clustering provided by the invention can also have the following characteristics: wherein, step S1 includes the following steps:
step S1-1, determining main hues according to the color characteristics of the training samples and the hue circle principle, and determining the training samples corresponding to each main hue by utilizing hue clustering;
step S1-2, the number of the obtained subareas is m, and each subarea corresponds to a spectrum value set with the same color characteristics;
step S1-3, calculating the hue angle and saturation of the training sample; the specific methods include two methods:
the method comprises the following steps: under a fixed observation environment, calculating CIE XYZ values corresponding to the spectral reflectances of the training samples, namely:
wherein the content of the first and second substances,kin order to adjust the factor(s),
in order to train the corresponding spectral reflectance of the sample,
is the relative spectral distribution curve of the spectrum,
is the CIE color matching function.
The CIE XYZ values of the training samples are converted to CIE Lab values, hue angle h and saturation C by a color space conversion function, i.e.:
wherein the content of the first and second substances,
to train the CIE XYZ values of the samples,
tristimulus values for CIE standard illuminant illuminated to a fully diffuse reflector surface;
the method 2 comprises the following steps: performing multivariate statistical analysis (such as principal component analysis, independent principal component analysis, non-negative principal component analysis and the like) on the spectral value set corresponding to the training sample set to obtain a feature vector; the variance contribution rates of the feature vectors (i.e., the variance contribution rates represent the proportion of the information amount contained in the feature vectors) are arranged from large to small, and are denoted as pc (1), pc (2), and pc (3) … pc (n).
Wherein the content of the first and second substances,
is the firstiThe number of feature vectors is determined by the number of feature vectors,
is a feature vectoriCorresponding coefficient, n is the spectral sampling interval.
Three preceding eigenvectors pc (1), pc (2), pc (3) are taken, the hue angle h and the saturation C of training sample are calculated, namely:
the cross-media color gamut mapping method for keeping the hue constant provided by the invention can also have the following characteristics: wherein, step S4 includes the following steps:
step S4-1, selecting a method consistent with the method for calculating the training sample to obtain the hue angle and the saturation, and calculating the hue angle and the saturation of the sample to be tested;
step S4-2, judging the partition where the sample to be tested is located by utilizing hue clustering, and marking the partition where the sample to be tested is located; if the sample to be tested is just positioned at the boundary of the two subareas, the dimension of the sample to be tested is reduced in the two subareas, the color difference between the numerical value after the dimension reduction under the condition of changing various light sources and the sample to be tested is calculated, the subarea where the sum of the color difference values of the multiple light sources is the minimum is selected as the subarea where the sample to be tested is positioned, and the subarea where the subarea is positioned is marked.
Step S4-3, using the feature vector of the partition to perform spectrum dimensionality reduction on the sample to be tested; and finally, spectrum reconstruction is carried out by searching the partition to obtain the spectrum reflectivity of the sample to be tested after reconstruction.
The invention has the beneficial effects that:
the method fully considers the requirements of practical application, selects the characteristic of the spectral sample according to the self, determines the main hues according to the color characteristic of the training sample and the hue circle principle, determines the training sample corresponding to each main hue by hue clustering, and has the characteristics of simple calculation, easy operation, less characteristic vector, high dimensionality reduction precision and the like; and the multispectral color information is compressed and stored with high precision, so that the use is more convenient for users.
Drawings
FIG. 1 is a flow chart of a spectral dimension reduction method based on hue clustering according to the present invention.
Detailed Description
In order to make the technical means, the creation features, the achievement purposes and the effects of the present invention easy to understand, the following embodiments specifically describe the spectral dimension reduction method based on hue clustering in conjunction with the drawings.
FIG. 1 is a flow chart of a spectral dimension reduction method based on hue clustering according to the present invention.
As shown in fig. 1, the training sample set may be selected from a munsell color system, an oswald color system, a chinese color system, and the like. In this embodiment, the training sample set used in the spectral dimension reduction method performs spectral dimension reduction on the sample to be measured by taking the munsell color system as an example, so as to implement dimension reduction and compression of multispectral data. The spectrum dimensionality reduction method based on hue clustering comprises the following specific steps:
step S1, determining main hues according to the color characteristics of the munsell color system and the color wheel principle, which are red (R), yellow-red (YR), yellow (Y), green-yellow (GY), green (G), blue-green (BG), blue (B), violet-blue (PB), violet (P), and red-violet (RP); according to the requirements of actual requirements, low-saturation color partitions can be divided; then determining a training sample corresponding to each main hue by utilizing hue clustering; the number of the obtained subareas is m, each subarea corresponds to a spectral value set with the same color characteristic at the moment, and the hue angle and the saturation of the training sample are calculated. The specific operation steps are as follows:
step S1-1, determining main hues according to the color characteristics of the Munsell color system and the hue circle principle, wherein the main hues are respectively red (R), yellow-red (YR), yellow (Y), green-yellow (GY), green (G), blue-green (BG), blue (B), violet-blue (PB), violet (P) and red-violet (RP); according to the requirements of actual requirements, low-saturation color partitions can be divided; then determining a training sample corresponding to each main hue by utilizing hue clustering;
step S1-2, the number of the obtained subareas is m, and each subarea corresponds to a spectrum value set with the same color characteristics;
step S1-3, calculating the hue angle and saturation of the training sample; the specific method comprises two methods, wherein one method is selected for calculation:
the method comprises the following steps: under a fixed observation environment, calculating CIE XYZ values corresponding to the spectral reflectances of the training samples, namely:
wherein the content of the first and second substances,kin order to adjust the factor(s),
in order to train the corresponding spectral reflectance of the sample,
is the relative spectral distribution curve of the spectrum,
is the CIE color matching function.
The CIE XYZ values of the training samples are converted to CIE Lab values, hue angle h and saturation C by a color space conversion function, i.e.:
wherein the content of the first and second substances,
to train the CIE XYZ values of the samples,
tristimulus values for CIE standard illuminant illuminated to a fully diffuse reflector surface;
the method 2 comprises the following steps: performing multivariate statistical analysis (such as principal component analysis, independent principal component analysis, non-negative principal component analysis and the like) on the spectral value set corresponding to the training sample set to obtain a feature vector; the variance contribution rates of the feature vectors (i.e., the variance contribution rates represent the proportion of the information amount contained in the feature vectors) are arranged from large to small, and are denoted as pc (1), pc (2), and pc (3) … pc (n).
Wherein the content of the first and second substances,
is the firstiThe number of feature vectors is determined by the number of feature vectors,
is a feature vectoriCorresponding coefficient, n is the spectral sampling interval.
Three preceding eigenvectors pc (1), pc (2), pc (3) are taken, the hue angle h and the saturation C of training sample are calculated, namely:
step S2, determining whether to divide the low saturation color partition:
(a) if the low saturation color subareas are not divided, determining the hue angle range corresponding to each subarea, marking the subareas according to the sequence of the hue angle ranges from small to large, wherein the hue angle ranges are respectively H (1), H (2) and H (3) … H (m), and the set of the spectral values corresponding to the subareas is marked as A (1), A (2) and A (3) … A (m).
(b) If the low saturation color partition is divided, a saturation threshold corresponding to the low saturation color partition is determined. And when the saturation corresponding to the training sample is smaller than the saturation threshold value, the low-saturation color partition belongs to the low-saturation color partition, when the saturation corresponding to the training sample is not smaller than the saturation threshold value, the low-saturation color partition belongs to other partitions, the low-saturation color partition is marked as H (1), and the corresponding spectral value set is marked as A (1). And determining the hue angle ranges corresponding to other partitions, marking the hue angle ranges from small to large as H (2), H (3) and H (4) … H (m), and recording the sets of the spectral values corresponding to the partitions as A (2), A (3) and A (4) … A (m).
Step S3, obtaining a characteristic vector of each subarea by adopting a multivariate statistical analysis method for the spectral value set of each subarea; fully considering the actual production requirement, and selecting the corresponding quantity of the feature vectors as the dimension reduction dimension of the partition;
step S4, judging the partition where the sample to be tested is located by utilizing hue clustering, and performing spectrum dimensionality reduction on the sample to be tested by utilizing the characteristic vector of the partition; and finally, spectrum reconstruction is carried out by searching the partition to obtain the spectrum reflectivity of the sample to be tested after reconstruction. The specific operation is as follows:
step S4-1, selecting a method consistent with the method for calculating the training sample to obtain the hue angle and the saturation, and calculating the hue angle and the saturation of the sample to be tested;
step S4-2, judging the partition where the sample to be tested is located, and judging the partition where the sample to be tested is located by utilizing hue clustering; and if the sample to be tested is just positioned at the boundary of the two subareas, reducing the dimension of the sample to be tested in the two subareas, calculating the color difference between the numerical value subjected to dimension reduction under the multispectral condition and the sample to be tested, and selecting the subarea where the sum of the color difference values of the multiple light sources is the minimum as the subarea where the sample to be tested is positioned.
Step S4-3, using the feature vector of the partition to perform spectrum dimensionality reduction on the sample to be tested; and finally, spectrum reconstruction is carried out by searching the partition to obtain the spectrum reflectivity of the sample to be tested after reconstruction.
The method fully considers the requirements of practical application, selects the characteristic of the spectral sample according to the self, determines the main hues according to the color characteristic of the training sample and the hue circle principle, determines the training sample corresponding to each main hue by hue clustering, and has the characteristics of simple calculation, easy operation, less characteristic vector, high dimensionality reduction precision and the like; and the multispectral color information is compressed and stored with high precision, so that the use is more convenient for users.
Effects and effects of the embodiments
According to the spectral dimension reduction method based on hue clustering provided by the embodiment, the main hues are determined according to the color characteristics of the training samples and the hue circle principle by selecting the spectral sample according to the characteristics of the spectral sample, the training samples corresponding to each main hue are determined by utilizing the hue clustering, and then the spectral reflectivity dimension reduction processing of the samples to be tested is carried out.
According to the spectral dimension reduction method based on hue clustering provided by the embodiment, the requirements of practical application are fully considered, and the method has the advantages of simple calculation, easy operation, less characteristic vectors, high dimension reduction precision and the like; and the multispectral color information is compressed and stored with high precision, so that the use is more convenient for users.
The above embodiments are preferred examples of the present invention, and are not intended to limit the scope of the present invention.
Claims (3)
1. A spectrum dimensionality reduction method based on hue clustering is characterized by comprising the following steps:
step S1, determining the main hues according to the color characteristics of the training samples and the hue circle principle, and determining the training sample subset corresponding to each main hue by utilizing hue clustering; the number of the obtained subareas is m, each subarea corresponds to a spectrum value subset with the same color characteristic, and the hue angle and the saturation of each training sample are calculated;
step S2, determining whether to divide the low saturation color partition:
(a) if the low saturation color subareas are not divided, determining a hue angle range corresponding to each subarea, marking the subareas according to the sequence of hue angle ranges from small to large, wherein the hue angle ranges are respectively H (1), H (2) and H (3) … H (m), and the corresponding spectral value subsets of each subarea are respectively marked as A (1), A (2) and A (3) … A (m);
(b) if the low saturation color subareas are divided, firstly determining a saturation threshold corresponding to the low saturation color subareas; when the saturation corresponding to the training sample is smaller than a saturation threshold value, the training sample belongs to a low-saturation color partition, when the saturation corresponding to the training sample is not smaller than the saturation threshold value, the training sample belongs to other partitions, the low-saturation color partition is marked as H (1), and a corresponding spectrum value set is marked as A (1); determining the hue angle ranges corresponding to other subareas, marking the hue angle ranges from small to large as H (2), H (3) and H (4) … H (m), and recording the spectral value sets corresponding to the subareas as A (2), A (3) and A (4) … A (m);
step S3, obtaining a characteristic vector of each subarea by adopting a multivariate statistical analysis method for the spectral value set of each subarea; selecting the number of corresponding feature vectors as the dimension reduction dimension of the partition;
step S4, judging the partition where the sample to be tested is located by utilizing hue clustering, and performing spectrum dimensionality reduction on the sample to be tested by utilizing the characteristic vector of the partition; and finally, spectrum reconstruction is carried out by searching the partition to obtain the spectrum reflectivity of the sample to be tested after reconstruction.
2. The spectral dimension reduction method based on hue clustering according to claim 1, wherein: wherein, step S1 includes the following steps:
step S1-1, determining main hues according to the color characteristics of the training samples and the hue circle principle, and determining the training samples corresponding to each main hue by utilizing hue clustering;
step S1-2, the number of the obtained subareas is m, and each subarea corresponds to a spectrum value set with the same color characteristics;
step S1-3, calculating the hue angle and saturation of the training sample; the specific methods include two methods:
the method comprises the following steps: under a fixed observation environment, calculating CIE XYZ values corresponding to the spectral reflectances of the training samples, namely:
wherein k is an adjustment factor, r (lambda) is the spectral reflectivity corresponding to the training sample, I (lambda) is the relative spectral distribution curve of the spectrum,
is the CIE color matching function;
the CIE XYZ values of the training samples are converted to CIE Lab values, hue angle h and saturation C by a color space conversion function, i.e.:
wherein (X, Y, Z) is CIE XYZ value of training sample, (X)n,Yn,Zn) Tristimulus values for CIE standard illuminant illuminated to a fully diffuse reflector surface;
the method 2 comprises the following steps: performing multivariate statistical analysis on the spectral value set corresponding to the training sample set to obtain a feature vector; arranging according to the variance contribution rate of the feature vectors from large to small, and marking as pc (1), pc (2) and pc (3) … pc (n);
where pc (i) is its i-th eigenvector, d (i) is the coefficient corresponding to the eigenvector i, and n is the spectral sampling interval;
three preceding eigenvectors pc (1), pc (2), pc (3) are taken, the hue angle h and the saturation C of training sample are calculated, namely:
3. the spectral dimension reduction method based on hue clustering according to claim 1, wherein: wherein, step S4 includes the following steps:
step S4-1, selecting a method consistent with the method for calculating the training sample to obtain the hue angle and the saturation, and calculating the hue angle and the saturation of the sample to be tested;
step S4-2, judging the partition where the sample to be tested is located by utilizing hue clustering, and marking the partition where the sample to be tested is located; if the sample to be tested is just positioned at the boundary of the two subareas, the dimension of the sample to be tested is reduced in the two subareas, the color difference between the numerical value after the dimension reduction under the condition of changing various light sources and the sample to be tested is calculated, the subarea where the sum of the color difference values of the multiple light sources is the minimum is selected as the subarea where the sample to be tested is positioned, and the subarea where the subarea is positioned is marked;
step S4-3, using the feature vector of the partition to perform spectrum dimensionality reduction on the sample to be tested; and finally, spectrum reconstruction is carried out by searching the partition to obtain the spectrum reflectivity of the sample to be tested after reconstruction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711221418.5A CN107766896B (en) | 2017-11-28 | 2017-11-28 | Spectrum dimensionality reduction method based on hue clustering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711221418.5A CN107766896B (en) | 2017-11-28 | 2017-11-28 | Spectrum dimensionality reduction method based on hue clustering |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107766896A CN107766896A (en) | 2018-03-06 |
| CN107766896B true CN107766896B (en) | 2021-01-29 |
Family
ID=61276814
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711221418.5A Active CN107766896B (en) | 2017-11-28 | 2017-11-28 | Spectrum dimensionality reduction method based on hue clustering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107766896B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109655155B (en) * | 2018-12-05 | 2020-09-29 | 北京印刷学院 | Method and device for evaluating color rendering of illumination light source |
| CN109508724B (en) * | 2019-01-16 | 2023-10-20 | 北京印刷学院 | Method for classifying observer color matching function by using cluster analysis |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105023239A (en) * | 2015-08-18 | 2015-11-04 | 西安电子科技大学 | Hyperspectral data dimensionality reduction method based on ultra-pixel and maximum boundary distribution |
| CN105069234A (en) * | 2015-08-13 | 2015-11-18 | 武汉大学 | Spectrum dimensionality reduction method and system based on visual perception feature |
| CN105869161A (en) * | 2016-03-28 | 2016-08-17 | 西安电子科技大学 | Method for selecting wave bands of hyperspectral image based on image quality assessment |
| CN106408619A (en) * | 2016-09-13 | 2017-02-15 | 齐鲁工业大学 | Method of realizing cross-media color reproduction based on spectral domain |
-
2017
- 2017-11-28 CN CN201711221418.5A patent/CN107766896B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105069234A (en) * | 2015-08-13 | 2015-11-18 | 武汉大学 | Spectrum dimensionality reduction method and system based on visual perception feature |
| CN105023239A (en) * | 2015-08-18 | 2015-11-04 | 西安电子科技大学 | Hyperspectral data dimensionality reduction method based on ultra-pixel and maximum boundary distribution |
| CN105869161A (en) * | 2016-03-28 | 2016-08-17 | 西安电子科技大学 | Method for selecting wave bands of hyperspectral image based on image quality assessment |
| CN106408619A (en) * | 2016-09-13 | 2017-02-15 | 齐鲁工业大学 | Method of realizing cross-media color reproduction based on spectral domain |
Non-Patent Citations (2)
| Title |
|---|
| "Dimension Reduction Using Spatial and Spectral Regularized Local Discriminant Embedding for Hyperspectral Image Classification";Yicong Zhou;《IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING》;20151231;第53卷(第2期);第1082-1095页 * |
| "面向颜色再现的光谱降维方法研究";刘攀;《包装工程》;20151231;第36卷(第3期);第119-123页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107766896A (en) | 2018-03-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Mojsilovic | A computational model for color naming and describing color composition of images | |
| US6480299B1 (en) | Color printer characterization using optimization theory and neural networks | |
| Sangwine et al. | The colour image processing handbook | |
| US8594420B2 (en) | Color naming, color categorization and describing color composition of images | |
| Zukauskas et al. | Statistical approach to color quality of solid-state lamps | |
| CN105069234B (en) | The spectrum dimension reduction method and system of a kind of view-based access control model Perception Features | |
| Pastilha et al. | Describing natural colors with Munsell and NCS color systems | |
| CN107424197B (en) | Method for realizing cross-media color reproduction based on spectral domain mapping | |
| CN107766896B (en) | Spectrum dimensionality reduction method based on hue clustering | |
| CN105496414A (en) | Make-up color diagnosis method customized by skin color and make-up color diagnosis device customized by skin color | |
| Minz et al. | Evaluation of RGB cube calibration framework and effect of calibration charts on color measurement of mozzarella cheese | |
| Derhak et al. | Introducing Wpt (Waypoint): A color equivalency representation for defining a material adjustment transform | |
| CN110324476B (en) | Method for representing color generation performance of mobile phone screen | |
| Darling et al. | Real-time multispectral rendering with complex illumination | |
| CN104742517B (en) | Color free collection and matching device | |
| CN105825020A (en) | Computing method of three-dimensional perceptual color gamut | |
| EP2866008B1 (en) | Tone type identification method | |
| Derhak et al. | Wpt (waypoint) shift manifold difference metrics for evaluation of varying observing‐condition (observer+ illuminant) metamerism and color inconstancy | |
| Wu | Charamer mismatch-based spectral gamut mapping | |
| Berns et al. | Spectral color reproduction of paintings | |
| Habib et al. | Spectral reproduction: drivers, use cases and workflow | |
| Zhang et al. | Backward spectral characterization of liquid crystal display based on forward spectral characterization | |
| Mirzaei | Object-Color Description Under Varying Illumination | |
| Mirzaei et al. | Gaussian-based hue descriptors | |
| Jin et al. | Testing of the uniformity of color appearance space |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20221212 Address after: 250306 Room 3115, No. 135, Ward Avenue, Ping'an Street, Changqing District, Jinan City, Shandong Province Patentee after: Shandong Jiqing Technology Service Co.,Ltd. Address before: 250353 College of printing and packaging engineering, 3501 Daxue Road, Changqing District, Jinan City, Shandong Province Patentee before: Qilu University of Technology |