JP3216818B2 - Color measurement device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the management of paint colors and dyeing degrees at production sites in various industries, or the measurement of color of products,
Alternatively, the present invention relates to a color measurement device that can be used for color measurement of a subject in medical and academic fields.

[0002]

2. Description of the Related Art Conventionally, a conventional color measuring device measures reflected light intensity in a wavelength region representing three primary colors of red (R), green (G), and blue (B). CIE)
There is known a color coordinate system standardized by, for example, an XYZ color system, an L ^* a ^* b ^* color system, or the like, which displays colors as numerical values to identify colors.

[0003] In such a color measuring device, an object color can be represented as an absolute value by a color system which is considered to be close to human senses.

[0004] However, the color discrimination ability for detecting a slight difference in the object color is due to the characteristics of the primary color system.
After all, it is determined by the values of R, G, and B. Therefore,
Such a color measuring apparatus cannot clearly discriminate a difference in a color to be distinguished, that is, a difference in a reflection spectrum to be classified. For example, colors whose spectra differ in the wavelength region of the color matching function of G cannot be classified by the measured value of G.

As this type of apparatus, there is a multi-channel spectrophotometer that measures a reflection spectrum of an object and identifies a difference in color based on the difference in the spectrum. However, this apparatus requires expensive equipment such as a diffraction grating and a highly sensitive detector array, and furthermore, when measuring a spectrum, the number of dimensions per data increases, so that it is processed and analyzed. There is a disadvantage that the scale of the device becomes large, the configuration of the device becomes complicated, and the cost increases.

[0006]

By the way, if a data example of each class of spectrum to be classified can be prepared in advance, a data input method effective for classification by a statistical method,
That is, a method of determining an input wavelength region depending on the purpose instead of the R, G, and B inputs described above can be considered.

As a statistical method suitable for such a method, reference 1 (DHFoley and JWSammon. Jr, IEEE Trans. C)
ompp, C-24, 281 (1975) ". This document 1 describes a method of obtaining a direct support axis of a space suitable for classifying two classes so as to maximize an evaluation criterion called Fisher Ratio. In order to apply this method to color classification, first, a certain wavelength region is sampled into n pieces, and an n-dimensional vector having reflected light intensity at each wavelength as an element is considered as data.
A plurality of data examples (hereinafter, referred to as “training sets”) known to belong to one of the two classes are prepared in advance. From this training set, the Fish
An er Ratio is calculated, and an n-dimensional vector suitable for classifying the two classes is calculated. This n-dimensional vector represents the characteristics of the spectral filter. Thus, it is considered that the characteristics of the spectral filter for class classification can be calculated by the statistical method.

However, although the spectral characteristics obtained by this method are suitable for class classification, there is a drawback that the original color of the object cannot be known.

A method of recognizing colors by a statistical method is disclosed in Reference 2 "J. Parkkinen and T. Jaaskelainen, Appl Opt,
26,4240 (1987). In the method described in Document 2, first, principal component analysis is performed on a training set for each class, and a subspace is set for each class. Then, a rotation operation is performed so that each subspace does not overlap between classes.
In this method, classification between classes can be performed using a spectral filter calculated by a statistical method.

A method for estimating the original n-dimensional spectrum data from the vector projected on each subspace is also discussed. According to the method, the original color can be estimated.

However, this method requires a large number of computations to determine the characteristics of the spectral filter by repeatedly performing the principal component analysis, that is, the operation of finding the covariance matrix and solving the eigenvalue problem. . In addition, since the subspace is set for each class, in order to determine which class the measurement value belongs to, the projection operation, that is, the operation of measuring the light intensity through the spectral filter, must be repeated by the number of classes. Therefore, extremely complicated processing is required.

Therefore, in the conventional color measuring device,
Since it measures the light intensity in the R, G, and B wavelength regions, it is not an optimal input method for the object to be classified.
Inconvenience may occur. In addition, when a statistical method is applied, the method is suitable for classification but cannot identify the original color of the target object, and adding a color identification function requires a large amount of calculation and is not suitable for practical use. There was such a problem.

The present invention has been made in view of the above circumstances, and can classify a specific color in a wavelength range that is difficult to classify with an R, G, B filter, and can easily design a color classification filter used for the classification. It is an object to provide a simple color measuring device.

[0014]

According to the present invention, in order to achieve the above object, an invention according to claim 1 comprises a light source for generating illumination light for illuminating an object to be measured; imaging means for converting the reflected light into an electrical signal, said
Multiple classes to which the reflection spectrum of the object
Sequence data of multiple samples prepared for each sample
Using a training set that is
Designed by statistical methods to classify
And a filter means having optical properties, and a color filter having optical properties for inputting color information of three primary colors of the measurement object, and a filter control means for switching if necessary between the color filter and the filter means A color measuring device characterized by comprising:

[0015]

According to the present invention, when performing color classification, the filter control means switches to the filter means . With this, the difference in the reflection spectrum of the object to be measured is classified.
Are classified.

When the absolute color of the object to be measured is measured, the color control is switched to the color filter by the filter control means, and the normal color measurement is performed. In the the present invention as described above, the three primary colors of the color information of the filter means, also the measurement object in the case of relatively classifying a difference in reflection spectrum of the measuring object, i.e., absolute color information In the case of obtaining the color filter, the color filter is switched to a filter having a different function if necessary, so when designing the filter means , it is necessary to consider obtaining the color information of the three primary colors of the object to be measured. There is no need to design, and the design of the filter means becomes easy.

[0017]

Embodiments will be described below with reference to the drawings.

FIG. 1 shows functional blocks of a color measuring apparatus according to a first embodiment of the present invention.

The color measuring apparatus includes a light source box 1 for generating monochromatic light, and a light source color generator for converting the monochromatic light generated by the light source box 1 into illumination light having a predetermined light intensity (spectrum) and outputting the light. Device 2 and this light source color creating device 2
A light guide 3 for guiding the illumination light output from the object to a predetermined position, a detector head 4 for irradiating the illumination light to the object and converting the reflected light from the object into an electric signal and outputting the electric signal, and a detector head 4 And a man-machine interface 6 for inputting an operator's instruction to the processor 5.

FIG. 2 shows the configuration of the light source box 1 and the light source color generator 2. In the light source box 1, a lamp power supply 7 and a white lamp 8 to which power is supplied by the power supply are provided. The white light emitted from the white lamp 8 is collimated by the lens 9 and guided to the light source color creator 2.

A lens 11 into which light from the light source side is incident is provided in the light source color creator 2. This lens 11
Is provided with a slit 12 at the focusing position. Concave mirror 1
3, the diffraction grating 14 and the concave mirror 15 are arranged in the same manner as in a known spectroscope. The liquid crystal filter 16 is disposed on the light-collecting surface of the concave mirror 15, that is, in a position where the output-side slit is disposed in the ordinary spectroscope. This liquid crystal filter 16
Is set to a desired transmittance by a liquid crystal filter driver 17 controlled by the processor 5. In addition, the transmittance can be set at the imaging position of each wavelength.
The liquid crystal filter 16 realizes a color classification filter described later. Light having a predetermined wavelength, which is the first-order diffracted light by the diffraction grating 14, is attenuated by the liquid crystal filter 16 and converted into illumination light having a desired spectrum. This illumination light is collected on the end face of the incident end side connector 19 of the optical fiber bundle guided into the light guide 3.

The detector head 4 is configured as shown in FIG. The illumination light guided by the light guide 3 is applied to the illumination lens 2 installed at the tip of the detector head 4.
The object to be measured is radiated through 1. The reflected light from the object is focused on the light receiving surface of the photo detector 23 by the objective lens 22 provided on the detector head 4. The output signal from the photodetector 23 is sent to the processor 5 via the amplifier 24.

FIG. 4 is a diagram showing the internal configuration of the processor 5. The processor 5 includes an internal bus 26
, A ROM 27, a CPU memory 28, an A / D converter 29, a statistical analysis processor 31, a liquid crystal filter driver interface 32, a man-machine interface 33, and an external recording device interface 34 are respectively connected.

The CPU 25 controls each component and controls the entire apparatus. The ROM 27 stores an operating system and an execution program managed by the operating system. The execution program is read and executed as needed. The CPU memory 28 is configured by a RAM, and records a program and data when executing a process. A
The / D converter 29 digitally inputs an output signal from the detector head 4. The statistical analysis processor 31 includes a product-sum calculator, a memory, and the like, and is configured to perform a matrix operation at a high speed, and is a dedicated processor for performing statistical analysis in preprocessing described later.
The liquid crystal filter driver interface 32 is an interface for sending a command signal for determining an address and transmittance to the filter driver 17. The man-machine interface 33 is an interface for inputting a signal from a keyboard in outputting to a display device for displaying a menu of a program or a measurement result. An external recording device interface 34 is an interface with an external recording device 35, ie, a hard disk driver or a floppy disk driver, which is used when recording programs and measurement results on a recording medium such as a hard disk or a floppy disk and reading programs and data therefrom. It is.

A color classification filter suitable for color classification of a specific color can be designed as follows.

Each color to be classified (reflection spectrum light)
Are prepared for each class and statistical analysis is performed.

To perform statistical analysis, a training set is first created. This is because, as shown in FIG. 5, the transmittance distribution of the liquid crystal filter 16 is changed to a specific wavelength region R1.
Is set to be the maximum. The light source color creator 2 having the liquid crystal filter 16 set as described above.
The sample is illuminated by illumination light close to monochromatic light from the sample. The intensity of the reflected light from the sample illuminated by such illumination light is measured and recorded in the CPU memory 28 in the processor 5.

Next, the wavelength region R1 is moved, and the same operation as above is repeated. N regions (n
The intensity of reflected light in each wavelength region divided into several tens is recorded. FIG. 6 shows an example of the spectrum data measured in this way. The data example shown here is spectral data obtained by measuring the reflected light intensity in a total of 11 wavelength regions every 25 nm over a wavelength range of 450 to 700 nm.

For all the prepared samples, the spectrum data of each class is measured to prepare a training set. When the total data amount becomes large, the data is temporarily recorded on a hard disk or a floppy disk by the external recording device 35.

Next, a method for designing a color classification filter having spectral characteristics suitable for classifying a class from the training set created as described above will be described. Note that the number of classes is k, and the number of data of the training set of each class is mi (i = 1,... K).

A first design method will be described. This is a filter characteristic design method that can optimize the class classification (color classification) when the number of classes k = 2.

The two classes will be referred to as class 1 and class 2, and the color of class 1 is the reference color, and the color of class 2 is a color of class 1 with a slightly different color. I do. The training set for each class is represented as follows: The training sets of class 1 and class 2 are represented by equation (1).

[0033]

(Equation 1) However, each data is represented by an n-dimensional vector represented by the equation (2).

[0034]

(Equation 2) For this training set, Fi expressed by equation (3)
A vector d that maximizes the sher Ratio (hereinafter referred to as “evaluation criterion F”) is obtained.

[0035]

(Equation 3) Here, S ₁ is an inter-class covariance matrix, which is calculated by equation (4).

[0036]

(Equation 4) Where Pi is the probability of occurrence of the i-th class,

[0037]

(Equation 5) Is the average spectrum of the ith class,

[0038]

(Equation 6) Is the average spectrum of all training,

[0039]

(Equation 7) Is

[0040]

(Equation 8) Is the probability of occurrence of

S ₂ is the average of the intra-class covariance matrix and is expressed by equation (5).

[0042]

(Equation 9) Here, the intra-class covariance matrix S ₂ ⁽ⁱ⁾ is defined by equation (6).

[0043]

(Equation 10) The vector d is finally derived by equation (7).

[0044]

[Equation 11] However,

[0045]

(Equation 12)

[0046]

[Equation 13] is

[0047]

[Equation 14] Is a normalization constant for

Next, a vector d ₂ that maximizes the evaluation criterion F in a space orthogonal to the vector d ₁ is obtained. This vector d ₂ is derived from equation (8).

[0049]

(Equation 15) However,

[0050]

(Equation 16) Is

[0051]

[Equation 17] Is a normalization constant for

The evaluation criterion F is an evaluation criterion indicating the degree of separation between classes when the data of the training set is projected onto the vector d. That is, the vector d that maximizes the degree of separation between classes represents the characteristic (classification vector) of a filter that can classify classes 1 and 2 most efficiently. Note that the vectors d ₁ and d ₁
Reference numeral ₂ denotes the first two orthogonal vectors that maximize the evaluation criterion F in the n-dimensional space, and (n-1) orthogonal vectors can be theoretically calculated in descending order of the evaluation criterion F. However, for classification of two classes, one classification vector d ₁ or two classification vectors d _1, d
₂ is enough.

Next, as a second method, a method of deriving a classification vector for an arbitrary number of classes k (k ≧ 2) will be described.

First, k is set in the same manner as in the first method.
Prepare a training set with example data for classes.

[0055]

(Equation 18) In this second method, ZHGu and SHLee, Opt.
The method described in 3,727 (1984) is used. This is for Hottering Trace for any number of classes greater than one.
This is a method for finding the optimal classification vector based on an evaluation criterion called Criterion (HTC). According to this mathematical statistical method, a matrix H that maximizes (HTC) J is obtained.

J = tr [S ₂ ⁻¹ S ₁ ] (10) The matrix H is derived by the following procedure. First, the eigen equation of the mean S ₂ of the intra-class covariance matrix is solved.

[0057]

[Equation 19] Next, a transformation matrix for whitening S ₂

[0058]

(Equation 20) Transforms S ₁ to find a matrix D.

[0059]

(Equation 21) Note that in the above equation (12)

[0060]

(Equation 22) Is the equation (11)

[0061]

(Equation 23) , And T is a matrix obtained by removing the eigenvector corresponding to the eigenvalue “0” by T ′ in equation (11).

[0062]

(Equation 24) Here, T ″ is a submatrix of the eigenvector corresponding to the eigenvalue 0.

Then, the transformation matrix H is obtained by the equation (15).

[0064]

(Equation 25) Note that S _{2 in} equation (11) and S _{1 in} equation (12) are:
It is obtained in the same manner as in equations (4) and (5). Also, (15)
The matrix H in the equation is composed of (k-1) valid vectors di (i
= 1,2, ... k-1).

[0065]

(Equation 26) The classification vector d obtained as described above represents a transmission characteristic of a filter effective for color classification.
A characteristic curve as shown in FIG. The color classification filter and the R, G, B color image filters having such transmission characteristics are realized by the liquid crystal filter 16 .

[0066] However, in general, the classification vector d, as shown in FIG. 7, a positive, because it has a negative component, the classification vector d positive vector components d ⁺ and a negative vector component d ^- be divided into a. In order to obtain the characteristic of the vector d, two types of filters, a positive component and a negative component, are provided, and the mutual transmitted light intensity of each filter is subtracted.

Next, the operation of the present embodiment configured as described above will be described.

First, a case where a specific color is classified by a color classification filter in which a transmittance distribution that can be derived by the above-described method is set will be described.

[0069] For example, when performing color classification by color classification vector d _1, d _2, the transmittance distribution of the liquid crystal filter 16 are four spectral transmission characteristics of the positive and negative components of the respective color classification vector d _1, d ₂ are respectively realized Are set sequentially.

The object is illuminated by the illumination light of a specific wavelength extracted by the four color classification filters set as described above, and the reflected light at that time is converted into an electric signal, and each reflected spectrum light The positive and negative vector components Z ¹⁺ , Z ¹⁻ , Z ²⁺ , and Z ^{2− of} are measured.

The processor 5 subtracts the measured value of each color classification vector based on the positive and negative components as follows.

Z ¹ = Z ^{1 +} −Z ^{1 −} , Z ² = Z ^{2 +} −Z ^{2 −} The following analysis is performed on the above subtraction result.

That is, in a classification space having the classification vectors d _{1 and} d ₂ as orthogonal axes, it is assumed that the distribution of colors of two kinds of classes can be represented as shown in FIG. In this case, a straight line

[0074]

[Equation 27] If the two classes can be separated by using the threshold as
One of the with class are also designed to allow efficient separation), and there measurement vector ^{_{Z 1 (Z 1 1, Z}} 1 2), which geometrical distance between the straight line L represents the class of separability Becomes That is, in FIG.
₁ (Z ₁ ¹ , Z ₁ ² ) is determined to belong to class 1, but the degree of separation is the distance t ₁ between the intersection vector Z ₁ ′ of a perpendicular drawn from the vector to the straight line L and the measured value vector Z _1. Can be expressed by

[0075] ^{_{t1 = [(Z 1 1 -Z}} 1 1 ') 2 + (Z 1 2 -Z 1 2') 2] 1/2 ... (17) However,

[0076]

[Equation 28] Therefore, if the measured value vector Z is obtained, the above (1)
According to the equation 8), an intersection vector Z ′ between the straight line L and a perpendicular line is obtained.
Is calculated, and the class to be included is determined from the magnitude relation between the calculated values. Next, according to the equation (17), the degree of separation t
Calculate 1 and display the result on the display of the man-machine interface 6. In addition, on the display,
As shown in FIG. 8, the mapping in the classification space is displayed to facilitate visual understanding.

Next, when the absolute color of the object is measured, the transmittance distribution of the liquid crystal filter 16 is sequentially changed to R,
It is set so as to realize the color matching function of each of the primary colors G and B.
Then, R, G, and B color filters are used for R, G, and B, respectively.
Normal colorimetry is performed by inputting the G and B intensities. In this case, the measured R, G, B values or R, G, B values are displayed on the display of the man-machine interface 6.
(X, y, Y) values converted from G and B values and (L ^* , a
^* , B ^* ) values are controlled so that measured values based on the color system standardized by the CIE are displayed.

As described above, according to the present embodiment, the color classification filter and the color filter can be switched as necessary, so that the specific color can be classified by the color classification filter, and the normal color measurement can be performed by the color filter. it can. Therefore, it is not necessary to estimate the absolute color of the object from the measurement values obtained by the color classification filter, and it is not necessary to design the color classification filter in consideration of the estimation of the absolute color. The design is simplified, and the burden on the designer can be greatly reduced.

Further, from the training set creation to the design of the color classification filter based on the classification vectors d _{1 and} d ₂ , the actual colorimetry can be consistently performed by the same apparatus with a relatively simple configuration.

Next, a second embodiment of the present invention will be described.

The present embodiment comprises a pre-processing device for performing from the creation of a training set to the design of a color classification filter, and a main processing device for performing actual color measurement.

FIG. 9 shows the configuration of the pre-processing device of this embodiment. The pretreatment apparatus of the present embodiment includes a light source device 100, a light incident end 101 on which reflected light of an object is incident, and a light incident end 101.
The reflected light taken in from the light guide 102 is inputted through the light guide 102, the polychromator 103 acts as a spectroscope, the detector array 104 in which the spectral data of the reflected light separated by the polychromator 103 is photoelectrically converted, and the detector array 104 105, the analyzer 10 in which the spectrum data is input and a predetermined operation is executed.
5 includes a man-machine interface 106 for inputting a command. Note that the light source device 100 is configured independently. The polychromator 103, the analyzer 105, the man-machine interface 10
Reference numeral 6 is the same as that shown in the first embodiment.

In the preprocessing device configured as described above, a training set is created in the same manner as in the first embodiment, and a transmission characteristic suitable for classification of a specific color is calculated in the analyzer 105 from the training set. You.

FIG. 1 shows the configuration of the processing apparatus of this embodiment.
0 is shown.

This processing apparatus includes a light source device 100 for illuminating an object under the same conditions as in the pre-processing, a detector head 107 for converting reflected light from the object into an electric signal, and a calculation performed by the pre-processing. It comprises a processor 108 to which the color classification vector data is input, and a man-machine interface 109.

FIG. 11 shows a specific configuration of the detector head 107.

This detector head 107 is provided with a lens array 110 composed of seven lenses radially arranged on the incident surface. A photodetector 112 divided into seven parts is provided on the image plane of each lens of the lens array 110. Color filters 111-1 to 111-7 having seven types of filter characteristics are attached to the light receiving surface of each divided surface of the photodetector 112.

The color filters 111-1 to 111-7 are
Each is composed of an interference filter, a colored glass filter, a Wratten filter, or a combination thereof. And these color filters 111-
The positive / negative characteristics of the _two color classification vectors d _{1 and} d ₂ are set for four of the images 111 to 111-7, respectively.
R, G, and B characteristics are set for each sheet.

The output signal of each photodetector 112 is
The selector 114 is connected via the corresponding amplifier 113.
Is input to The selector 114 is connected to one of the amplifiers 1
Thirteen outputs are selected, and the signal captured by the color classification filter or the primary color signal captured by the R, G, B filters is output to the processor 108. The selector 114 is switched by a control signal given from the processor 108.

In the present processing apparatus configured as described above,
The reflected light from the object is input to the detector head 107, and the input light intensity in the wavelength region extracted by the color filters 111-1 to 111-7 having the above-described filter characteristics is detected, and amplified by the respective amplifiers 113. Is done. The measurement values captured by the color filters 111-1 to 111-7 are selected by the selector 114,
Input to the processor 108.

The processor 108 includes the detector head 1
Classification and measurement of the absolute color are performed based on the input signal from 07. Then, the measurement result and the selection of the operation by the observer are displayed on the man-machine interface 109.

As described above, according to this embodiment, a plurality of color filters 111-1 each having a predetermined spectral transmission characteristic are provided.
To 111-7, the selector 114
This makes it possible to perform measurement with a filter having a predetermined spectral transmission characteristic only by extremely simple electrical switching, which is effective when a set of spectral filters is used constantly to some extent. For example, there is a case where the standard of color classification of an object can be limited to some extent, such as a paint inspection at a mass production site or a color inspection of a fresh food, in which the same processing is repeatedly performed for a large amount or for a long time.

[0093]

As described in detail above, according to the present invention,
It is possible to provide a color measuring device that can accurately classify a specific color in a wavelength range that is difficult to classify with the R, G, and B filters, and that can easily design a color classification filter used for the classification.

[Brief description of the drawings]

FIG. 1 is a block diagram of a color measurement device according to a first embodiment of the present invention.

FIG. 2 is a configuration diagram of a light source box and a light source color creator of the device according to the first embodiment.

FIG. 3 is a configuration diagram of a detector head of the device according to the first embodiment.

FIG. 4 is a configuration diagram in a processor of the device according to the first embodiment.

FIG. 5 is a diagram for explaining a training set creation operation.

FIG. 6 is a diagram showing reflected light intensity actually collected to create a training set.

FIG. 7 is a diagram illustrating characteristics of a color classification vector.

FIG. 8 is a diagram showing a classification result in the first embodiment.

FIG. 9 is a configuration diagram of a pretreatment device according to a second embodiment.

FIG. 10 is a configuration diagram of the present processing apparatus in a second embodiment.

FIG. 11 is a configuration diagram of a detector head according to a second embodiment.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Light source box, 2 ... Light source color creator, 3 ... Light guide, 4 ... Detector head, 5 ... Processor, 6 ... Man-machine interface.

Continuation of front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G01J 3/00-3/52

Claims (1)

(57) [Claims] A light source configured to generate illumination light for illuminating the measurement target; an imaging unit configured to convert reflected light from the measurement target into an electrical signal; and a reflection spectrum of the measurement target. Multiple to belong
Spectral data of multiple samples prepared for each class
The reflection spectroscopy scan is performed using a training set
Designed by statistical methods to classify vector differences
Filter means having optical characteristics, color filters having optical properties for inputting color information of the three primary colors of the object to be measured , and filter control means for switching between the filter means and the color filter as necessary That you have
Characteristic color measuring device.