Spectral sensitivity estimation device and program therefor

Abstract

To provide a spectral sensitivity estimation device for estimating spectral sensitivity.SOLUTION: A spectral sensitivity estimation device 1 comprises: a singular value calculation unit 31 for calculating a singular value for a matrix in which as many rows as the sampling number of center wavelength are arranged, with the measured value of light whose center wavelength is a sampled wavelength having been measured at a predetermine wavelength interval by a spectroradiometer 5 as a component in a row direction; a pseudo-inverse matrix generation unit 32 for generating a plurality of pseudo-inverse matrixes for each singular value, with a predetermined number excluded in decreasing order of the singular value, in which the respective singular values are assumed to be permissible errors; a spectral sensitivity candidate calculation unit 33 for calculating a plurality of spectral sensitivity candidates by multiplying a response value outputted by a device T under test, as a column vector, to the plurality of pseudo-inverse matrixes; and a spectral sensitivity selection unit 34 for selecting a spectral sensitivity candidate whose sum total of differences between elements is minimum among the spectral sensitivity candidates as the spectral sensitivity of the device T under test.SELECTED DRAWING: Figure 5

Description

[Technical field]

[0001] The present invention relates to a spectral sensitivity estimation device and a program for estimating the spectral sensitivity of a device under test. [Background technology]

[0002] In order to evaluate the accuracy of color reproduction in color cameras, measurements by photometers and colorimeters, etc., it is important to understand the spectral sensitivity characteristics of each device. For this reason, devices for measuring spectral sensitivity have been developed (for example, see Non-Patent Document 1). In general, spectral sensitivity is measured by measuring the response of an imaging system to light (measurement light) sampled at desired wavelength intervals with a sufficiently small half-width using a spectrometer (monochromator) in order to measure with high resolution. For example, conventional spectral sensitivity measurements are performed using light with a sufficiently small half-width W of 5 nm or 10 nm and a central wavelength shifted by 1 nm or 5 nm as a desired wavelength interval D as shown in FIG. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] “Spectral response (spectral sensitivity) measurement”, [online], Bunkoukeiki Co., Ltd., [Retrieved June 26, 2023], Internet<URL:http: / / www.bunkoukeiki.co.jp / responsivity.html> Summary of the Invention [Problem to be solved by the invention]

[0004] When the half-width and wavelength interval of the measurement light are reduced as in the conventional method, the exposure time required in the device under test (eg, a camera) becomes longer, which causes a problem of longer measurement time. In addition, when the exposure time becomes long, the sensor noise in the device under test accumulates, and the accuracy of the spectral response measurement decreases. Therefore, the conventional method has a problem that it is necessary to increase the intensity of the measurement light at each wavelength as shown in Figure 11, and to use a light source with a very high output that is calibrated to equal energy.

[0005] Therefore, an object of the present invention is to provide a spectral sensitivity estimation device and a program therefor that are capable of estimating spectral sensitivity with high accuracy even when the half-width and wavelength interval of the measurement light are made larger than in the past and the light intensity is reduced. [Means for solving the problem]

[0006] In order to achieve the above object, a spectral sensitivity estimation device according to the present invention is a spectral sensitivity estimation device that estimates the spectral sensitivity of a device under test that outputs a response value to input light, and is configured to include a singular value calculation unit, a pseudo-inverse matrix generation unit, a spectral sensitivity candidate calculation unit, and a spectral sensitivity selection unit.

[0007] In such a configuration, the spectral sensitivity estimation device calculates, by the singular value calculation unit, a plurality of singular values ​​for a matrix in which the number of rows is the same as the number of sampled central wavelengths, with row-direction components being measurement values ​​obtained by measuring light having a central wavelength that is a sampled wavelength in the wavelength range to be measured at predetermined wavelength intervals using a spectroradiometer.

[0008] Then, the spectral sensitivity estimation device generates, by a pseudo-inverse matrix generation unit, a plurality of pseudo-inverse matrices for the matrix in which each singular value, excluding a predetermined number from the largest of the plurality of singular values, is set as a permissible error. In this manner, by generating pseudo-inverse matrices by setting a plurality of permissible errors, it is possible to generate a plurality of pseudo-inverse matrices that represent the characteristics of the spectral distribution measured by the spectroradiometer and that assume measurement noise.

[0009] Then, the spectral sensitivity estimation device calculates multiple spectral sensitivity candidates by using the spectral sensitivity candidate calculation unit to treat the response values ​​output by the device under test as column vectors for each light having the same central wavelength measured by the spectroradiometer, and multiplying these by multiple pseudo-inverse matrices. The spectral sensitivity candidates calculated in this manner reflect the characteristics of the actually measured spectral distribution, so that it is possible to make the wavelength interval and half-width of the spectral distribution of the measurement light larger than in the past, and to reduce the intensity of the light source.

[0010] Then, the spectral sensitivity estimation device uses the spectral sensitivity selection unit to select, as the spectral sensitivity of the device under test, the spectral sensitivity candidate with the smallest sum of differences between adjacent elements from among the multiple spectral sensitivity candidates. Alternatively, the spectral sensitivity estimation device may use the spectral sensitivity selection unit to calculate, as a likelihood, the inverse of the sum of differences between adjacent elements from among the multiple spectral sensitivity candidates, and select, as the spectral sensitivity of the device under test, the spectral sensitivity candidate with the largest likelihood. This allows the spectral sensitivity estimation device to estimate spectral sensitivities that are less affected by noise. The spectral sensitivity estimating apparatus can be operated by a program for causing a computer to function as the spectral sensitivity estimating apparatus. Effect of the Invention

[0011] According to the present invention, the wavelength interval and half width of the spectral distribution of the measurement light can be made larger than in the past, so that the spectral sensitivity can be estimated more accurately in a shorter time than in the past. [Brief description of the drawings]

[0012] [Figure 1] 1 is an overall configuration diagram showing a configuration of a spectral sensitivity measurement system using a spectral sensitivity estimation device according to an embodiment of the present invention. [Diagram 2] 4 is a graph showing an example of the spectral distribution of LED light emitted by the light emitting device. [Diagram 3] 4 is a graph showing an example of the spectral distribution of light output by a spectroscope. FIG. [Figure 4]FIG. 2 is a graph showing an example of the spectral distribution of radiance measured by a spectroradiometer. [Diagram 5] 1 is a block diagram showing a configuration of a spectral sensitivity estimating device according to an embodiment of the present invention. [Figure 6] 5 is a graph illustrating an example of calculation in an estimation unit of the spectral sensitivity estimation device according to the embodiment of the present invention. FIG. [Figure 7] 5 is a graph illustrating an estimation result of the spectral sensitivity estimation apparatus according to the embodiment of the present invention. FIG. [Figure 8] 4 is a graph showing the spectral distribution of measurement light used in the spectral sensitivity estimation device according to the embodiment of the present invention, multiplexed for each center wavelength. FIG. [Figure 9] 4 is a flowchart for explaining a data collection operation of the spectral sensitivity estimation apparatus according to the embodiment of the present invention. [Figure 10] 4 is a flowchart for explaining an estimation operation of a spectral sensitivity of the spectral sensitivity estimation device according to the embodiment of the present invention. [Figure 11] 1 is a graph showing the spectral distribution of measurement light used in conventional spectral sensitivity measurement, multiplexed for each center wavelength; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. [Spectral sensitivity measurement system configuration] The configuration of a spectral sensitivity measuring system 100 using a spectral sensitivity estimating device 1 according to an embodiment of the present invention will be described with reference to FIG. The spectral response measurement system 100 measures the spectral response of a device under test T. The spectral response is a characteristic of sensitivity to the wavelength of an input light beam.

[0014] The device under test T is a device for which the spectral sensitivity is to be measured. The device under test T may be anything that has a light receiving element that receives light and outputs a response value to the input light. For example, the device under test T may be a camera (imaging device), a photometer, a colorimeter, etc. It is assumed that the response value of the device under test T is linear with respect to the input light. The spectral sensitivity measurement system 100 includes a spectral sensitivity estimation device 1, a light emitting device 2, a spectroscope 3, an integrating sphere 4, and a spectroradiometer 5.

[0015] The spectral sensitivity estimation device 1 estimates the spectral sensitivity of a device under test T that outputs a response value to input light. This spectral sensitivity estimation device 1 estimates the spectral sensitivity of the device under test T from the response value of the device under test T to measurement light, which is the input light, and the measurement value of a spectroradiometer 5 to the same measurement light. The configuration and operation of the spectral sensitivity estimating device 1 will be described later.

[0016] The light emitting device 2 emits white light. This white light contains light of various wavelengths, as shown in the graph of the spectral distribution D1 in Fig. 2. In addition, in the spectral distribution D1, the horizontal axis indicates the wavelength (nm) and the vertical axis indicates the light intensity (relative intensity). For example, the light emitting device 2 uses an LED as a light source and emits white light. The light emitting device 2 outputs the emitted light to the spectrometer 3 via the optical fiber F.

[0017] The spectrometer 3 separates light having a specified wavelength from light containing various wavelengths. Here, the spectroscope 3 splits the white light input from the light emitting device 2 via the optical fiber F into light having a wavelength designated by the spectral sensitivity estimation device 1 as a center wavelength. The light split by the spectroscope 3 has a wavelength designated by the spectral sensitivity estimation device 1 at a center wavelength λ , as shown in the spectral distribution D2 in FIG. C In addition, in the spectral distribution D2, the horizontal axis indicates the wavelength (nm) and the vertical axis indicates the light intensity (relative intensity). The light dispersed by the spectroscope 3 becomes measurement light in the spectroradiometer 5 and the device under test T. The half-width W of the measurement light may be larger than 5 nm or 10 nm used in conventional methods. For example, the spectroscope 3 may output measurement light with a half-width W of about 30 nm. The spectroscope 3 outputs the dispersed measurement light to an integrating sphere 4 .

[0018] The integrating sphere 4 has a spherical inner surface, and uniformly scatters the incident light (measurement light) inside. The integrating sphere 4 takes in the measurement light from an aperture (entrance aperture) H0 and emits the scattered measurement light from two apertures (exit apertures) H1 and H2. The apertures H1 and H2 are provided at positions where they are not directly irradiated with the measurement light from the aperture H0. With these two openings H1 and H2, the integrating sphere 4 can irradiate the spectroradiometer 5 and the device T to be measured with the same measurement light.

[0019] The spectroradiometer 5 measures the radiance (spectral radiance) for each wavelength (for example, at 1 nm intervals) by dispersing and analyzing light. Spectroradiometer 5 is provided opposite the position of aperture H1 of integrating sphere 4, and measures the spectral radiance from the light (measurement light) emitted from aperture H1. The radiance measured by the spectroradiometer 5 is the central wavelength λ 2 dispersed by the spectrometer 3 as shown in FIG. C The spectral distribution D3 is shown as follows: The horizontal axis is the wavelength (nm) and the vertical axis is the spectral radiance (W sr -1 m -2 nm -1 ) is shown. The spectroradiometer 5 outputs the spectral distribution of radiance, which is the measurement result, to the spectral sensitivity estimation device 1.

[0020] The device under test T is provided opposite the position of the aperture H2 of the integrating sphere 4, receives the light (measurement light) emitted from the aperture H2, and outputs a response value. If the device under test T is a camera, the response value is a pixel value. If the device under test T is a photometer (illuminance meter), the response value is luminous intensity (illuminance), and if it is a colorimeter, the response value is a tristimulus value.

[0021] With this configuration, in the spectral sensitivity measurement system 100, the same measurement light is irradiated onto the spectroradiometer 5 and the device under test T. Then, the spectral sensitivity estimation device 1 becomes able to estimate the spectral sensitivity of the device under test T from the spectral distribution of the measurement light of the spectroradiometer 5 and the response value of the device under test T to the same measurement light. The configuration and operation of the spectral sensitivity estimating device 1 will be described below.

[0022] [Configuration of the spectral sensitivity estimation device] The configuration of a spectral sensitivity estimation device 1 according to an embodiment of the present invention will be described with reference to Fig. 5 (and also with reference to Fig. 1 as appropriate). The spectral sensitivity estimation device 1 includes a collection section 10, a storage section 20, and an estimation section 30.

[0023] The collecting unit 10 collects data from the spectroradiometer 5 and the device under test T. The collecting unit 10 includes a wavelength switching unit 11, a spectral distribution acquiring unit 12, and a response value acquiring unit 13.

[0024] The wavelength switching unit 11 instructs the spectrometer 3 to switch the wavelengths sampled at predetermined wavelength intervals in the wavelength range to be measured to the center wavelength. For example, if the wavelength range is 380 nm to 780 nm and the central wavelength is sampled at 10 nm intervals, the wavelength switching unit 11 sequentially switches the central wavelength λ 1 to 380 nm, 390 nm, 400 nm, . . . , 770 nm, and 780 nm for the spectrometer 3. C The sampling intervals of the center wavelength do not necessarily have to be equal.

[0025] The wavelength switching unit 11 selects a center wavelength λ C After instructing the spectral distribution acquisition unit 12 and the response value acquisition unit 13 to switch to the central wavelength λ Cand at the stage where notification that data acquisition has been completed is received from the spectral distribution acquisition unit 12 and the response value acquisition unit 13, C Switch to the next wavelength.

[0026] The spectral distribution acquisition unit 12 acquires the spectral distribution, which is a measurement value measured by the spectroradiometer 5 . The spectral distribution acquisition unit 12 acquires the measurement value (spectral distribution) from the spectroradiometer 5 when instructed by the wavelength switching unit 11 to acquire data. The spectral distribution acquisition unit 12 acquires the spectral distribution at a central wavelength λ C and stores the spectral distribution data in the storage unit 20. The spectral distribution data is a matrix (spectral distribution matrix) in which the row components are measurements taken by the spectroradiometer 5 at predetermined wavelength intervals (for example, 1 nm intervals), and the number of rows is the same as the number of samplings of the center wavelength. The spectral distribution acquisition unit 12 detects the specified center wavelength λ C After acquiring the spectral distribution of the light, the wavelength switching unit 11 is notified that acquisition of the data has been completed.

[0027] The response value acquiring section 13 acquires the response value output by the device under test T. The response value acquiring unit 13 acquires a response value from the device under test T when the wavelength switching unit 11 instructs the acquisition of data. The response value acquisition unit 13 acquires the response value at the central wavelength λ C and stores the response value data in the storage unit 20 in correspondence with the response value data. The response value acquisition unit 13 detects the specified center wavelength λ C After acquiring the response value, the wavelength switching unit 11 is notified that acquisition of the data has been completed.

[0028] This enables the collection unit 10 to collect the spectral distribution (spectral distribution data) of the spectroradiometer 5 associated with wavelengths sampled at predetermined wavelength intervals in the wavelength range of the measurement target, and the response values ​​(response value data) of the device under test T associated with wavelengths at the same sampling positions, and store them in the memory unit 20.

[0029] The storage unit 20 stores the spectral distribution data and response value data collected by the collection unit 10, and can be configured with a storage medium such as a general semiconductor memory. The spectral distribution data and the response value data stored in the storage unit 20 are referenced by the estimation unit 30.

[0030] The estimation section 30 estimates the spectral sensitivity of the device under test T from the spectral distribution data and the response value data stored in the storage section 20. The estimation unit 30 may start operation when a notification that data collection has been completed is input from the collection unit 10, or may start operation when instructed to do so from outside. When the spectral sensitivity of the device under test T is the column vector c, the matrix specified by the spectral distribution data (spectral distribution matrix) is A, and the response value of the device under test T is the row vector d, the relationship of the following equation (1) holds.

[0031]

number

[0032] The estimation unit 30 estimates the spectral sensitivity by determining the column vector c in the formula (1). Here, the estimation section 30 includes a singular value calculation section 31, a pseudo inverse matrix generation section 32, a spectral sensitivity candidate calculation section 33, and a spectral sensitivity selection section .

[0033] The singular value calculation unit 31 calculates a plurality of singular values ​​(scalars) from the matrix of the spectral distribution specified by the spectral distribution data stored in the storage unit 20. That is, the singular value calculation unit 31 calculates singular values ​​for a matrix (spectral distribution matrix) in which the number of rows is equal to the number of sampled central wavelengths, with the measurement values ​​measured by the spectroradiometer 5 at predetermined wavelength intervals for light having a central wavelength that is sampled in the wavelength range to be measured as the row direction components. For example, in the case of spectral distribution data obtained by measuring light having a wavelength range of 380 nm to 780 nm and sampled at central wavelengths at 10 nm intervals at wavelength intervals of 1 nm, the size of the matrix is ​​41 rows by 401 columns.

[0034] The singular value calculation unit 31 calculates a matrix A with m rows and n columns (m×n) as follows: A=U*S*V T Calculates a sequence of values ​​greater than "0" of matrix S when it is subjected to singular value decomposition into (T is transpose). Note that U is an orthogonal matrix with m rows and m columns, and V is an orthogonal matrix with n rows and n columns. S is an m-row, n-column matrix shown in the following formula (2).

[0035]

number

[0036] where s1,…,s m (s1 ≥ … ≥ s m ) is a singular value. The singular value calculation unit 31 calculates the singular values ​​s1, . . . , s m and output to the pseudo-inverse matrix generating unit 32. Since the method of calculating the singular values ​​is common, the explanation will be omitted here. For example, the singular value vector s=(s1s2...s) can be calculated by using the svd (singular value decomposition) function of "MATLAB (registered trademark)" by The MathWorks, Inc. m ) can be calculated using the following formula (3).

[0037]

number

[0038] The pseudo-inverse matrix generating section 32 generates a pseudo-inverse matrix for the matrix of the spectral distribution data. The pseudo-inverse matrix generation unit 32 generates, for each singular value obtained by subtracting a predetermined number from the largest of the multiple singular values ​​calculated by the singular value calculation unit 31, a plurality of pseudo-inverse matrices (Moore-Penrose generalized inverse matrices) with respect to the matrix of the spectral distribution data, with each singular value being the allowable error.

[0039] The pseudo-inverse matrix generating unit 32 sets the number of singular values ​​to be excluded as p (for example, p=10) and calculates the pth singular values ​​(s1, . . . , s 10 ), the remaining singular values ​​(s 11 ,…,s m ) to the tolerance of each. In addition, the singular values ​​to be excluded are not treated as part of the allowable error since they largely represent the characteristics of matrix A.

[0040] The pseudo-inverse matrix generating unit 32 sets singular values ​​smaller than the allowable error to "0" and generates a pseudo-inverse matrix B (n rows and m columns) of the matrix A of the spectral distribution data. This generates mp pseudo-inverse matrices B1, B2, ..., B m-p will be generated. The method of generating the pseudo-inverse matrix B is common, and therefore the description thereof will be omitted here. For example, the pinv function of "MATLAB (registered trademark)" is used to generate the pseudo-inverse matrix B with a tolerance of t=s 11 ,…,s m For each, multiple pseudo-inverse matrices B (B1, B2, ..., B m-p ) can be obtained.

[0041]

number

[0042] In this way, the pseudo-inverse matrix generating unit 32 can generate a pseudo-inverse matrix that takes into account measurement noise by setting a plurality of allowable errors and generating a plurality of pseudo-inverse matrices. The pseudo-inverse matrix generating section 32 outputs the generated pseudo-inverse matrices to the spectral sensitivity candidate calculating section 33 .

[0043] The spectral sensitivity candidate calculation section 33 calculates a plurality of spectral sensitivity candidates for the device T under test. The spectral sensitivity candidate calculation unit 33 calculates multiple spectral sensitivity candidates by treating the response value output by the device under test T as a column vector for each light having the same central wavelength measured by the spectroradiometer 5, and multiplying it by multiple pseudo-inverse matrices generated by the pseudo-inverse matrix generation unit 32. The response value output by the device under test T is stored in the storage unit 20 as response value data.

[0044] The spectral sensitivity candidate calculation unit 33 applies the pseudo inverse matrix B to the above formula (1) to calculate the spectral sensitivity (column vector) c from the pseudo inverse matrix B and the response value (row vector) d according to the following formula (5).

[0045]

number

[0046] As a result, the spectral sensitivity candidate calculation unit 33 calculates a plurality of pseudo inverse matrices B1, B2, . . . , B m-p For each of the multiple spectral sensitivity candidates c1, c2, …, c m-p Generate. The spectral sensitivity candidate calculation section 33 outputs the calculated spectral sensitivity candidates to the spectral sensitivity selection section .

[0047] The spectral sensitivity selection section 34 selects the spectral sensitivity of the device under test T from a plurality of spectral sensitivity candidates. The spectral sensitivity selecting section 34 selects, as the spectral sensitivity of the device under test T, the spectral sensitivity candidate that has the smallest sum of differences between adjacent elements from among the multiple spectral sensitivity candidates calculated by the spectral sensitivity candidate calculating section 33. Here, the spectral sensitivity selection unit 34 selects each of the spectral sensitivity candidates c i For (i=1, 2, ..., mp), the difference (diff) between adjacent elements is calculated, and the likelihood L is calculated as the inverse of the sum (sum) of those differences.

[0048]

number

[0049] The spectral sensitivity selection unit 34 selects the spectral sensitivity candidate c that maximizes the likelihood L. i is selected as the spectral sensitivity of the device under test T. The spectral sensitivity selection unit 34 outputs the selected spectral sensitivity as an estimation result to the outside. For example, the spectral sensitivity selection unit 34 graphs the spectral sensitivity and outputs it to a display device (not shown). Alternatively, the spectral sensitivity selection section 34 may output the spectral sensitivity data directly to the outside.

[0050] With the above configuration, the spectral sensitivity estimation device 1 can estimate the spectral sensitivity of the device under test T by using the spectral distribution measured by the spectroradiometer 5 at the same wavelength as that of the device under test T. The spectral sensitivity estimating device 1 can be operated by a program (spectral sensitivity estimating program) for causing a computer (not shown) to function as each of the above-mentioned units.

[0051] (Measurement results) The results of measurements performed by the spectral sensitivity estimating device 1 will be described with reference to FIGS. In Fig. 6, the horizontal axis represents the allowable error of the singular value, and the vertical axis represents the likelihood. In Fig. 7, the horizontal axis represents the wavelength, and the vertical axis represents the spectral response. FIG. 6 shows singular values ​​s1, s2, . . . , s calculated by the singular value calculation unit 31 of the spectral sensitivity estimation device 1. m The figure shows the likelihood L of the spectral response generated by the pseudo-inverse matrix with a tolerance of (m=41).

[0052] As shown in FIG. 6, the spectral sensitivity estimation device 1 estimates singular values ​​s1, s2, . . . , s m Among the spectral sensitivity candidates generated by the pseudo-inverse matrix with the singular values ​​excluding the top p (here, 10) singular values ​​as the tolerance, the tolerance (singular value s 26 ) is estimated as the spectral sensitivity of the device under test T.

[0053] In the graph of Fig. 7, the spectral sensitivity of the device under test T estimated by the spectral sensitivity estimation device 1 is indicated by a dotted line. Also, in the graph of Fig. 7, the spectral sensitivity of the device under test T measured by a conventional method is indicated by a solid line. In this way, the spectral sensitivity estimated by the spectral sensitivity estimating device 1 is almost the same as the spectral sensitivity measured strictly as in the conventional method.

[0054] As described above, the spectral sensitivity estimation device 1 uses the spectral distribution, which is the actual measurement value of the spectroradiometer 5, and estimates the spectral sensitivity of the device under test T in relation to the spectral distribution. Therefore, compared to the conventional method of measuring the device under test T alone, the spectral sensitivity estimation device 1 can reduce the number of samples of the spectral distribution and central wavelength used, and can also suppress the output of the light source used.

[0055] 8, the spectral sensitivity estimation device 1 can use a spectrum having a wavelength interval D of about 10 nm and a half width W of about 30 nm as measurement light dispersed by the spectroscope 3. Moreover, the spectral sensitivity estimation device 1 does not need to be calibrated to have equal energy at each central wavelength. As a result, compared to the conventional measurement light shown in FIG. 11, which has a wavelength interval D of 1 nm or 5 nm, a half-width W of 5 nm or 10 nm, and is spectroscopy calibrated to equal energy, the spectral sensitivity estimation device 1 can shorten the exposure time to the device under test T, suppress the accumulation of noise, and use a light source with a lower output. Therefore, the spectral sensitivity estimating device 1 can estimate the spectral sensitivity more accurately in a shorter time than before.

[0056] [Operation of the spectral sensitivity estimation device] Next, the operation of the spectral sensitivity estimating device 1 according to the embodiment of the present invention will be described with reference to Fig. 9 and Fig. 10. Here, the data collection operation in the collecting unit 10 and the spectral sensitivity estimation operation in the estimating unit 30 will be described separately.

[0057] (Data collection operation) First, the data collection operation in the collection unit 10 of the spectral sensitivity estimating device 1 will be described with reference to FIG. 9 (and for the configuration, refer to FIG. 5 as appropriate).

[0058] In step S1, the wavelength switching unit 11 sets a center wavelength (for example, an initial value is 380 nm) in the wavelength range to be measured (for example, 380 nm to 780 nm) in the spectrometer 3. Here, the spectrometer 3 splits the light emitted from the light emitting device 2 into light having the set wavelength as the center wavelength.

[0059] In step S2, the spectral distribution acquisition unit 12 acquires the measurement value (spectral distribution) measured by the spectroradiometer 5, and stores it in the storage unit 20 as spectral distribution data associated with the central wavelength set in step S1. In step S3, the response value acquisition unit 13 acquires a response value from the device under test T, and stores it in the storage unit 20 as response value data associated with the central wavelength set in step S1.

[0060] In step S4, the wavelength switching unit 11 determines whether or not the measurement for the wavelength range of the measurement target has been completed. If the measurement for the wavelength range to be measured is not completed (No in step S4), in step S5, the wavelength switching unit 11 shifts the center wavelength in the wavelength range to be measured by a predetermined wavelength interval (for example, 10 nm) and sets it in the spectrometer 3. As a result, the spectrometer 3 splits the light emitted from the light emitting device 2 into light having the newly set wavelength as the center wavelength. On the other hand, if the measurements for the wavelength range of the measurement target are completed (Yes in step S4), the collection unit 10 ends the operation.

[0061] As a result, the spectral distribution (spectral distribution data) of the spectroradiometer 5 associated with wavelengths sampled at predetermined wavelength intervals in the wavelength range of the measurement target, and the response value (response value data) of the device under test T associated with the wavelengths at the same sampling positions are stored in the memory unit 20.

[0062] (Spectral sensitivity estimation operation) Next, the spectral sensitivity estimation operation in the estimating section 30 of the spectral sensitivity estimating device 1 will be described with reference to FIG. 10 (and for the configuration, refer to FIG. 5 as appropriate).

[0063] In step S10, the singular value calculation unit 31 calculates singular values ​​from the matrix of the spectral distribution specified by the spectral distribution data stored in the storage unit 20 (see formula (3) above). In step S11, the pseudo-inverse matrix generation unit 32 generates, for each singular value calculated in step S10, excluding a predetermined number from the largest singular value, a plurality of pseudo-inverse matrices with respect to the matrix of the spectral distribution data, with each singular value being the allowable error (see equation (4) above).

[0064] In step S12, the spectral sensitivity candidate calculation unit 33 calculates multiple spectral sensitivity candidates by treating the response value data stored in the memory unit 20 as a column vector and multiplying it by the multiple pseudo-inverse matrices generated in step S11 (see equation (5) above). In step S13, the spectral sensitivity selection unit 34 calculates the difference between adjacent elements in the plurality of spectral sensitivity candidates calculated in step S12, and calculates the likelihood by the inverse of the sum of the differences (see formula (6) above).

[0065] In step S14, the spectral sensitivity selection unit 34 selects, as the spectral sensitivity of the device under test T, the spectral sensitivity candidate having the maximum likelihood calculated in step S13 from among the multiple spectral sensitivity candidates calculated in step S12. Through the above operations, the spectral sensitivity estimating device 1 can estimate the spectral sensitivity of the device under test T.

[0066] Although the configuration and operation of the spectral sensitivity estimating device 1 according to the embodiment of the present invention have been described above, the present invention is not limited to this embodiment. For example, although the spectral sensitivity estimation device 1 is configured with the collecting unit 10, the storage unit 20, and the estimating unit 30 here, these may be configured separately. That is, the collecting unit 10 may be a data collecting device, the storage unit 20 may be a storage device, and the estimating unit 30 may be a spectral sensitivity estimation device. In this case, the collection unit 10 (data collection device) and the estimation unit 30 (spectral sensitivity estimation device) can be operated by programs (data collection program, spectral sensitivity estimation program) that cause a computer (not shown) to function as each part in the above-mentioned device. [Explanation of symbols]

[0067] 100 Spectral sensitivity measurement system 1. Spectral sensitivity estimation device 10 Collection Department 11 Wavelength switching section 12 Spectral distribution acquisition section 13 Response value acquisition unit 20 Memory section 30 Estimation part 31 Singular value calculation unit 32 Pseudo-inverse matrix generator 33 Spectral sensitivity candidate calculation unit 34 Spectral sensitivity selection section 2. Light-emitting device 3 Spectrometer 4 Integrating sphere 5 Spectroradiometer T Device under test

Claims

1. A spectral sensitivity estimation device that estimates a spectral sensitivity of a device under test that outputs a response value to input light, comprising: a singular value calculation unit that calculates a plurality of singular values ​​for a matrix in which the number of rows is equal to the number of sampled central wavelengths, the measurement values ​​being obtained by measuring light having a central wavelength that is a sampled wavelength in the wavelength range of the measurement target at a predetermined wavelength interval using a spectroradiometer as row direction components; a pseudo-inverse matrix generating unit configured to generate a plurality of pseudo-inverse matrices for the matrix, the pseudo-inverse matrices being set to respective singular values, the singular values ​​being the maximum singular values ​​minus a predetermined number; a spectral sensitivity candidate calculation unit that calculates a plurality of spectral sensitivity candidates by treating a response value output by the device under test as a column vector for each light having the same center wavelength measured by the spectroradiometer and multiplying the response value by the plurality of pseudo inverse matrices; a spectral sensitivity selection unit that selects, from the plurality of spectral sensitivity candidates, a spectral sensitivity candidate that has a minimum sum of differences between adjacent elements as the spectral sensitivity of the device under test; A spectral sensitivity estimation device comprising:

2. A spectral sensitivity estimation device that estimates a spectral sensitivity of a device under test that outputs a response value to input light, comprising: a singular value calculation unit that calculates a plurality of singular values ​​for a matrix in which the number of rows is equal to the number of sampled central wavelengths, the measurement values ​​being obtained by measuring light having a central wavelength that is a sampled wavelength in the wavelength range of the measurement target at a predetermined wavelength interval using a spectroradiometer as row direction components; a pseudo-inverse matrix generating unit configured to generate a plurality of pseudo-inverse matrices for the matrix, the pseudo-inverse matrices being set to respective singular values, the singular values ​​being the maximum singular values ​​minus a predetermined number; a spectral sensitivity candidate calculation unit that calculates a plurality of spectral sensitivity candidates by treating a response value output by the device under test as a column vector for each light having the same center wavelength measured by the spectroradiometer and multiplying the response value by the plurality of pseudo inverse matrices; a spectral sensitivity selection unit that calculates a reciprocal of a sum of differences between adjacent elements from among the plurality of spectral sensitivity candidates as a likelihood, and selects the spectral sensitivity candidate with the maximum likelihood as the spectral sensitivity of the device under test; A spectral sensitivity estimation device comprising:

3. a wavelength switching unit that instructs a spectroscope that splits light emitted by a light emitting device to sample and switch a center wavelength in the wavelength range; a spectral distribution acquisition unit that acquires measurement values ​​measured at predetermined wavelength intervals by the spectroradiometer for each light of a central wavelength sampled by the wavelength switching unit; a response value acquiring unit that acquires a response value output by the device under test for each light of the central wavelength sampled by the wavelength switching unit; The spectral sensitivity estimating device according to claim 1 , further comprising:

4. 3. The spectral sensitivity estimating device according to claim 1, wherein the measurement value measured by the spectroradiometer is radiance.

5. 3. The spectral sensitivity estimating device according to claim 1, wherein the response value output by the device under test is luminance, luminous intensity or tristimulus value.

6. A program for causing a computer to function as the spectral sensitivity estimating device according to claim 1 or 2.

Images