JP4400450B2 - Stray light correction method and two-dimensional spectral luminance meter using the same - Google Patents

Description

  The present invention provides a two-dimensional spectral luminance meter that obtains the spectral intensity of each pixel of a two-dimensional light source to be measured, converts it into a two-dimensional distribution of indices such as luminance and chromaticity, and the two-dimensional The present invention relates to a stray light correction method in the case where stray light correction is performed in measurement by a spectral luminance meter.

  Conventionally, a tristimulus type two-dimensional colorimeter using a filter has been used for the brightness and chromaticity of a two-dimensional light source, but in recent years it is close to a single color such as various displays and LED application devices (narrow spectral width). The number of surface light sources having an emission spectrum is increasing, and the need for measuring these luminance and chromaticity with high accuracy is increasing the need for a spectroscopic two-dimensional luminance meter. Many of the spectroscopic two-dimensional luminance meters (two-dimensional spectral luminance meters) adopt the following multi-filter method or slit dispersion image method.

<Multi-filter method>
A plurality of band pass filters (BPFs) with different center wavelengths attached in an annular shape to a rotating disk or the like are sequentially inserted into the imaging light flux of the target optical system, and the light flux in the transmission wavelength band of each filter. A luminance measurement method in which two-dimensional spectral characteristic information is obtained by capturing an image obtained by using a two-dimensional image sensor placed on the imaging plane (see, for example, Patent Document 1).

<Slit dispersion image method>
An image formed by the objective optical system of the two-dimensional light source to be measured is cut out by a slit placed on the imaging plane. A chromatic dispersion image of the clipped slit-like image (slit light source) by the dispersion optical system is picked up by a two-dimensional image sensor placed on the imaging surface. Then, a luminance measurement method for obtaining two-dimensional spectral characteristic information by scanning the measurement object or measurement system in a direction orthogonal to the slit (see, for example, Patent Document 2).
JP-A-6-201472 JP-A-8-50057

  In both the multi-filter method and the slit dispersion image method, there is a reflection component from a band pass filter or a slit. This reflection component is re-reflected by the lens surface of the optical system, the inner surface of the housing of the apparatus, etc., and reenters as stray light. Then, the re-entered stray light has an influence on the original measurement image (actual image) obtained by imaging by the image sensor, that is, the stray light image generated by the stray light is superimposed on the measurement image, The accuracy of the observed image, particularly the accuracy of the high contrast image, is deteriorated. Even when a band-pass filter or slit is not used, stray light is unavoidably caused by reflected light from the surface of the imaging element or the lens surface of the optical system.

  The present invention has been made in view of the above circumstances, and the effect of stray light on a measurement image is removed (stray light correction is performed), and a suitable measurement image can be obtained, while stray light is maintained while maintaining sufficient accuracy. It is an object of the present invention to provide a stray light correction method and a two-dimensional spectroluminometer using the same, which can shorten the calculation processing time (correction time) required for correction.

A stray light correction method according to one aspect of the present invention is a stray light correction method for obtaining a corrected image by correcting the influence of stray light on an observation image obtained by imaging, wherein the stray light image based on the influence of the stray light is superimposed. From the observation image and the response image obtained in advance for all the pixels in the imaging area in accordance with the incident light of unit intensity to the specific pixel in the imaging area of the observation image that is measured and stored in advance A stray light correction method for estimating and calculating a stray light image using image information having a number of pixels lower than the number of pixels constituting the image, and removing the estimated and calculated stray light image from the observation image to calculate the correction image that approximates the actual image The observation image (O _ij (λ)) obtained by the imaging is converted into a low pixel observation image (O _IJ (λ)) that is an observation image having a low pixel count with respect to the total number of pixels of the observation image. The first step to The from the low pixel observation image and the low pixel number of the response image _{((F MN) IJ (λ} )), low-pixel correction image _{(Q IJ} as a low pixel number image of the corrected image (R _'ij (λ)) (Λ)) and a stray light image (B _IJ ) having a low number of pixels according to a difference (O _IJ (λ) −Q _IJ (λ)) between the low pixel observation image and the low pixel corrected image. A third step of calculating (λ)), a fourth step of converting the low stray light image into a stray light image (B _ij (λ)) for the total number of pixels, and observation of the total number of pixels A fifth step of calculating a corrected image (R ′ _ij (λ)) that approximates the actual image by subtracting the stray light image (B _ij (λ)) for the total number of pixels from the image (O _ij (λ)) ; , characterized in that it have a.

  According to the above configuration, the observation image on which the stray light image based on the influence of the stray light is superimposed, and the specific pixel in the imaging area of the observation image that is measured and stored in advance by calibration or the like to be used in the stray light correction calculation. A stray light image is estimated from a response image (referred to as a response image in this embodiment) created by incident light of unit intensity on all pixels in the imaging area, using image information having a number of pixels lower than the number of pixels constituting the observation image. The calculated and calculated stray light image is removed (for example, subtracted) from the observation image, thereby obtaining a corrected image (an image obtained by correcting the influence of stray light from the observation image). Therefore, since the stray light correction process is performed in this way, a suitable measurement image from which the influence of stray light on the measurement image (observation image) is removed can be obtained. In addition, the calculation at the time of actual measurement is efficiently performed by using the response image information obtained by measurement in advance, and the image information having the number of pixels lower than the number of pixels constituting the observation image is used. The method of performing stray light correction makes it possible to reduce the amount of data handled at the time of calculation. On the other hand, with a low pixel count for stray light images whose spatial frequency is much lower than the observed image. Even if this calculation is applied, the error from the case of not performing the calculation with a low number of pixels is small, so that the calculation processing time (correction time) required for stray light correction is significantly reduced while maintaining sufficient accuracy. can do. Note that, by removing the influence of stray light from the observed image, a great effect can be obtained in improving the accuracy of low-luminance pixels in an image having contrast.

Furthermore, since the stray light correction calculation is performed in the first to fifth steps, the observation image (with the original total number of pixels) can be obtained with sufficient accuracy while greatly reducing (shortening) the amount of data to be handled and the calculation processing time. It is possible to perform stray light correction on the.

In the above configuration , the low pixel corrected image (Q _IJ (λ)) and the response image ((F _MN ) _IJ (λ)) as a temporary actual image with respect to the low pixel observation image (O _IJ (λ)) are simulated. An observation image (O ′ _IJ (λ)) is calculated, and a low pixel correction image (Q is calculated by a difference image (O ′ _IJ (λ) −O _IJ (λ)) due to a difference between the pseudo observation image and the low pixel observation image. _IJ (λ)) is re-corrected, and a new pseudo-observation image (O ′ _IJ (λ)) is calculated from the re-corrected low pixel correction image (Q _IJ (λ)) and the response image. An asymptotic operation for further correcting the low pixel correction image with a new difference image (O ′ _IJ (λ) −O _IJ (λ)) based on the difference between the simple pseudo observation image and the low pixel observation image is an error caused by the difference image. In the second step by repeatedly performing until the value becomes equal to or less than a predetermined value. It is desirable to calculate the pixel correction image _{(Q IJ (λ)).}

According to this configuration, the low pixel correction (in the second step) is performed from the low pixel observation image (O _IJ (λ)) and the response image ((F _MN ) _IJ (λ)) having a low number of pixels by an asymptotic method. Since the method of calculating the image (Q _IJ (λ)) is used, it is possible to obtain a low-pixel corrected image with high accuracy in a short calculation processing time by using a simpler processing program, and thus, further to the entire stray light correction. The calculation processing time can be shortened and the calculation accuracy (the accuracy of the corrected image (R ′ _ij (λ))) can be improved.

In the above-described configuration, it is preferable that the image having the total number of pixels is binned, and a pixel set image including a predetermined number of pixel sets in which the predetermined number of pixels are grouped is used as the low pixel number image . According to this configuration, an image with a total number of pixels (for example, an observation image (O _ij (λ)) or a stray light image (B _ij (λ))) is easily binned, so that an image with a low number of pixels (for example, a low image) An observation image (O _IJ (λ)) or stray light image (B _IJ (λ))) of the number of pixels can be obtained.

In the above configuration , it is preferable that the response image is obtained based on an image captured by forming light from a point light source on each pixel set in the imaging region of the observation image . According to this configuration, in order to obtain a response image based on an image picked up by irradiating (irradiating) light from a point light source to each pixel set, not for each pixel, Response images (depending on the individual dimensional spectroluminometer) can be acquired experimentally and easily.

In the above-described configuration , it is desirable that the stray light correction calculation process be performed for each wavelength of incident light in imaging an observation image . According to this configuration, the stray light correction process is performed for each wavelength of incident light in the imaging of the observation image (in this embodiment, according to the center wavelength λ of the spectrally transmitted light by the wedge band pass filter). Even if there is wavelength dependency, the stray light correction can be performed effectively.

A two-dimensional spectroluminometer according to another aspect of the present invention is a two-dimensional spectroluminometer that performs a stray light correction calculation to obtain a corrected image by correcting the influence of stray light on an observation image obtained by imaging. Obtained for all the pixels in the imaging area according to the observation image on which the stray light image based on the influence is superimposed and the incident light of unit intensity to the specific pixel in the imaging area of the observation image that is measured and stored in advance. The correction is performed to estimate and calculate a stray light image using image information having a lower number of pixels than the number of pixels of the observation image, and to remove the estimated and calculated stray light image from the observation image and approximate the actual image Stray light correction calculating means for calculating an image , wherein the stray light correction calculating means is configured to observe an observation image (O _ij (λ)) obtained by the imaging with a lower number of pixels than the total number of pixels of the observation image. Low image that is an image An original observation image (O _IJ (λ)) is converted, and the corrected image (R ′ _ij (λ)) is converted from the low pixel observation image and the response image ((F _MN ) _IJ (λ)). A low pixel corrected image (Q _IJ (λ)) as a low pixel number image is calculated, and the difference between the low pixel observed image and the low pixel corrected image (O _IJ (λ) −Q _IJ (λ)) is calculated. A low stray light image (B _IJ (λ)) is calculated, the low stray light image is converted into a stray light image (B _ij (λ)) for the total number of pixels, and the total number of pixels A corrected image (R ′ _ij (λ)) that approximates the actual image is calculated by subtracting the stray light image (B _ij (λ)) for the total number of pixels from the observed image (O _ij (λ)). And

  According to the above configuration, the observation image on which the stray light image based on the influence of the stray light is superimposed by the stray light correction calculation means, and the observation image captured and stored in advance by calibration or the like to be used in the stray light correction calculation. A stray light image is estimated and calculated from image data having a pixel number lower than the number of pixels in the observation image from the response image created by incident light of unit intensity on a specific pixel in the region on all pixels in the imaging region. By removing (for example, subtracting) the calculated stray light image from the observed image, a corrected image that approximates the actual image is obtained. Therefore, since the stray light correction process is performed in this way, a suitable measurement image from which the influence of stray light on the measurement image (observation image) is removed can be obtained. In addition, calculation at the time of actual measurement is efficiently performed by using response image information obtained by measurement in advance, and the stray light correction is performed using image information having a pixel number lower than the number of pixels of the observation image. It is possible to reduce the amount of data handled at the time of calculation, while applying the calculation with the low pixel count to the stray light image whose spatial frequency is much lower than the observation image. However, since the error from the case where the calculation with the low number of pixels is not performed becomes small, the calculation processing time (correction time) required for the stray light correction can be greatly shortened while maintaining sufficient accuracy.

According to the configuration of one aspect of the present invention, the stray light image is estimated and calculated from the observation image and the response image measured in advance using the image information having the number of pixels lower than the number of pixels of the observation image, and the calculated stray light Since a stray light correction process is performed such as obtaining a corrected image that approximates the actual image by removing the image from the observation image, a suitable measurement image in which the influence of stray light on the measurement image (observation image) is removed is obtained. be able to. In addition, the calculation at the time of actual measurement is efficiently performed by using the response image information obtained by measurement in advance, and the image information having the number of pixels lower than the number of pixels constituting the observation image is used. The method of performing stray light correction makes it possible to reduce the amount of data handled at the time of calculation. On the other hand, with a low pixel count for stray light images whose spatial frequency is much lower than the observed image. Even if this calculation is applied, the error from the case of not performing the calculation with a low number of pixels is small, so that the calculation processing time (correction time) required for stray light correction is significantly reduced while maintaining sufficient accuracy. can do. Note that, by removing the influence of stray light from the observed image, a great effect can be obtained in improving the accuracy of low-luminance pixels in an image having contrast.

Further, since the stray light correction calculation is performed in the first to fifth steps, the amount of data to be handled and the calculation processing time are greatly reduced (shortened), and the observation image (with the original total number of pixels) is sufficiently accurate. It becomes possible to perform stray light correction.

In addition, since a low pixel correction image is calculated by an asymptotic method from a low pixel observation image and a low pixel response image, a simple processing program is used and a low pixel with high accuracy in a short calculation processing time. A corrected image can be obtained, and as a result, calculation time for the entire stray light correction can be further shortened and calculation accuracy can be improved.

Also, by binning an image with the total number of pixels (an observation image with a total number of pixels, a stray light image, etc.), an image with a low number of pixels (an observation image with a low number of pixels, a stray light image, etc.) can be easily obtained. .

In addition, in order to obtain a response image based on an image picked up by irradiating (irradiating) light from a point light source to each pixel set, not to each pixel, a two-dimensional spectral luminance meter using the response image Response images can be acquired experimentally and easily.

In addition, since the stray light correction process is performed for each wavelength of incident light in the imaging of the observation image, the stray light correction can be performed effectively even if the cause of the stray light has wavelength dependency.

According to the configuration according to another aspect of the present invention, the stray light image is estimated and calculated from the observation image and the response image measured in advance using the image information having the number of pixels lower than the number of pixels of the observation image, and the calculated stray light Since a stray light correction process is performed such as obtaining a corrected image that approximates the actual image by removing the image from the observation image, a suitable measurement image in which the influence of stray light on the measurement image (observation image) is removed is obtained. be able to. In addition, the calculation at the time of actual measurement is efficiently performed by using the response image information obtained by measurement in advance, and the image information having the number of pixels lower than the number of pixels constituting the observation image is used. The method of performing stray light correction makes it possible to reduce the amount of data handled at the time of calculation. On the other hand, with a low pixel count for stray light images whose spatial frequency is much lower than the observed image. Even if this calculation is applied, the error from the case of not performing the calculation with a low number of pixels is small, so that the calculation processing time (correction time) required for stray light correction is significantly reduced while maintaining sufficient accuracy. can do.

In addition, it is possible to perform stray light correction on the observation image (with the original total number of pixels) with sufficient accuracy while greatly reducing (shortening) the amount of data to be handled and the calculation processing time.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
(Overall description of the two-dimensional spectral luminance meter)
FIG. 1 is a schematic structural diagram showing an example of a two-dimensional spectral luminance meter 1 according to the present invention. The two-dimensional spectral luminance meter 1 includes an objective optical system 2, a sub-transmission band removal filter 3, a first condenser lens 4, a second condenser lens 5, a relay lens 6, an image sensor 7, an image signal processing unit 8, and a main control unit 9. , And a scanning wedge bandpass filter 10. The objective optical system 2 is an optical lens (lens group) that makes a light beam La from a two-dimensional light source L, which is a measurement target, incident thereon and creates a first image 2a on a first image surface 2b. The sub-transmission band removing filter 3 is a filter that removes sub-band transmission light of the WBPF 12 described later. Specifically, the WBPF 12 has a sub-transmission band at a wavelength position that is 1/2 and 2 times the center wavelength of incident light (for example, when the center wavelength is about 400 nm, about 200 nm and 800 nm). However, the sub-transmission band removing filter 3 blocks light having a wavelength of, for example, about 380 nm or less and 720 nm or more with respect to the WBPF 12 having such a sub-transmission band. Then, the sub-transmission band transmitted light is removed. The sub-transmission band removal filter 3 is disposed in front of the first condenser lens 4.

  The first condenser lens 4 and the second condenser lens 5 are lenses for converging, that is, condensing (converging) the light beam La. The first condenser lens 4 prevents the light beam La from the objective optical system 2 from diverging to the WBPF 12, and condenses the light beam La from the objective optical system 2 on the first image plane 2b. Similarly, the second condenser lens 5 condenses the light beam La transmitted through the WBPF 12 on the relay lens 6.

  The relay lens 6 is a lens that relays the light beam La (light image) that formed the first image 2a toward the image sensor 7 at the same magnification, and creates the second image 6a on the second image plane 6b. By the way, the focal point of the first condenser lens 4 is in the vicinity of the rear principal point of the objective optical system 2, and the principal ray of the luminous flux La that emerges from the objective optical system 2 to form each pixel of the first image 2a and converges, Regardless of the incident part, it is made substantially parallel (parallel light flux) to the optical axis, thereby suppressing the influence of the incident angle dependence of the center wavelength of WBPF 12 (interference BPF) described later. The focal point of the second condenser lens 5 is in the vicinity of the front principal point of the relay lens 6, and the light beam La that has passed through the WBPF 12 is effectively incident on the relay lens 6.

  The image pickup device 7 is arranged on the second image plane 6b, that is, arranged so that the image pickup surface (light receiving surface) of the image pickup device 7 overlaps with the second image plane 6b, and the light beam La that forms the second image 6a is generated. The second image 6a is picked up by receiving (imaging) light. The imaging element 7 has, for example, a 1000 × 1000 pixel two-dimensional imaging area (imaging surface) of about 7.4 mm square (one side is 7.4 mm), and a light beam La is imaged in the imaging area. Then, each pixel of about 7.4 μm square in the imaging area is converted into an electrical signal (photoelectric conversion), and this image signal is output to the image signal processing unit 8.

  The image signal processing unit 8 performs predetermined signal processing (analog signal processing and digital signal processing) on the image signal transmitted from the image sensor 7. The image signal processing unit 8 has a function of performing A / D conversion from an analog signal to a digital signal, and transmits the digital signal to the main control unit 9.

  The main control unit 9 reads out and executes a ROM (Read Only Memory) that stores each control program, a RAM (Random Access Memory) that stores data for arithmetic processing and control processing, and the control program. It consists of a CPU (Central Processing Unit) and the like, and controls operation of the entire two-dimensional spectral luminance meter 1.

  The main control unit 9 has a spectral sensitivity of an image (an image having different transmission wavelength information for each pixel column) captured by the image sensor 7 in accordance with each scanning step by the scanning WBPF 10 described later via the image signal processing unit 8. Based on information such as response image information, which will be described later, and predetermined coefficients (for example, calibration coefficient and weight coefficient), which are captured as a plurality of different images and stored in advance in the RAM or the like, stray light correction and luminance correction for the images are performed. And a function of performing various arithmetic processes such as obtaining tristimulus values described later. The main control unit 9 may separately include a dedicated storage unit (for example, an image memory) for storing each captured image.

  A scanning wedge band pass filter (WBPF) 10 extracts monochromatic light (spectral transmitted light) having different wavelengths (center wavelength of the WBPF 12). The scanning WBPF 10 sequentially moves (slides) a later-described WBPF 12 (filter holding plate 11) in the scanning WBPF 10 by a plurality of scanning steps in a vertical direction (x direction) with respect to the optical axis of each lens. Thus, the monochromatic light having different wavelengths is extracted by relatively scanning the WBPF 12 with the light beam La. The scanning WBPF 10 is arranged in the vicinity of the first image plane 2b (including the case where the incident surface of the scanning WBPF 10 overlaps the first image plane 2b). Further, “scanning” by the scanning WBPF 10 indicates an operation of moving the WBPF 12 (filter holding plate 11) along the scanning direction (x direction) with respect to the light beam La (fixed in position). Details of the scanning WBPF 10 will be described later.

  The light beam La emitted from the two-dimensional light source L is incident on the objective optical system 2 and passes through the sub-transmission band removal filter 3 to remove the sub-transmission band transmission light. Then, the first condenser lens 4 scans the WBPF 10 (WBPF12). The first image 2a is formed (imaged) on the first image plane 2b. The light beam La that forms the first image 2a is split through the scanning WBPF 10 (WBPF 12), and the split light beam La passes through the second condenser lens 5 and enters the relay lens 6, and the relay lens 6 (the light beam La). The first image 2a is relayed at the same magnification) to form the second image 6a on the second image surface 6b. Then, the second image 6 a is picked up by the image pickup device 7, and this captured image (signal) is transmitted to the image signal processing unit 8 and subjected to signal processing such as A / D conversion, and then the main control unit 9. Sent to. Then, various arithmetic processes related to removing the stray light image from the captured image are performed in the main control unit 9.

  Here, the scanning WBPF 10 will be described in detail with reference to FIG. As shown in FIG. 2, the scanning WBPF 10 includes a filter holding plate 11, a WBPF 12, and a scanning drive unit 13. The filter holding plate 11 is for holding the WBPF 12. The filter holding plate 11 is a plate-like body (frame body of the WBPF 12) whose outer shape is substantially rectangular, and an opening 111 is formed at a substantially central portion thereof. For example, the opening 111 is formed in a size having a length (edge) of about 20 mm in the x direction (scanning direction) and about 9 mm in the y direction orthogonal to the x direction (scanning direction). The WBPF 12 is fitted into the opening 111. It is provided by combining. A dotted line frame illustrated in the WBPF 12 is a portion through which the light beam La for imaging on the image sensor 7 passes (this portion is fixed in position), that is, the first image 2a imaged on the WBPF 12. The size and shape of the dotted frame is, for example, a square shape of about 7.4 mm square corresponding to the imaging surface of the imaging device 7.

  The filter holding plate 11 includes a first light-shielding portion 112 and a second light-shielding portion 113 in the region of both end portions (left and right end portions) sandwiching the opening 111 in the x direction. The first and second light shielding portions 112 and 113 have at least the size of the first image 2a indicated by the dotted frame, and the first light shielding portion 112 or the second light shielding portion 112 is moved by the movement of the filter holding plate 11 in the x direction. When the light shielding portion 113 comes to the position of the first image 2a (imaging area), the light shielding portion shields (shields) the light beam La from the imaging element 7. As described above, since the filter holding plate 11 is provided with the first and second light shielding portions 112 and 113, the filter holding plate 11 also has a function as a so-called shutter for shielding light from the image sensor 7. (Thus, it is not necessary to provide a separate shutter, and the configuration can be simplified).

  The WBPF 12 is a filter for obtaining transmitted light having different center wavelengths in the scanning direction. Specifically, in the WBPF 12, the center wavelength λ (center wavelength λc (x, y)) of transmitted light at each part (position of coordinates (x, y)) of the WBPF 12 is constant along the y direction, and This is a so-called interference bandpass filter (BPF) that continuously (smoothly) changes at about 17 nm / mm along the x direction. The center wavelength of the WBPF 12 is about 390 to 710 nm in the scanning range of x = 0 to 64d (0 to 18.9 mm) when the x coordinate is expressed using a movement width d (= 0.296 mm) for each scanning step described later. Change continuously. In this way, the WBPF 12 having a different center wavelength depending on the part is a so-called wedge-shaped interference filter whose film thickness changes (linearly) in the scanning direction (x direction), such as the WBPF 12 shown in FIG. Yes.

  The scanning drive unit 13 scans the WBPF 12 and the first and second light shielding units 112 and 113 with respect to the light beam La (first image 2a) by moving the filter holding plate 11 along the scanning direction (x direction). Scanning drive). The scanning drive unit 13 includes a stepping motor 14, a rotating shaft 15, a connecting member 16, and a drive circuit 17. The stepping motor 14 is a motor that rotationally drives the rotary shaft 15 and rotates by a predetermined angle at a predetermined rotational speed every time a pulse signal (digital signal) is input. The connecting member 16 is for connecting the rotary shaft 15 and the filter holding plate 11 so as to be movable in parallel, and is fixed to the filter holding plate 11. Here, the connecting member 16 is provided integrally with the filter holding plate 11 at, for example, one end portion on the first light shielding portion 112 side. The connecting member 16 has an arbitrary position with respect to the filter holding plate 11 (for example, the second light shielding portion 113 side) and an arbitrary configuration as long as the filter holding plate 11 can move (scan) in the x direction. It may be provided.

  The rotating shaft 15 is a member (shaft body) that moves the connecting member 16 in a state of being connected to the connecting member 16. For example, a screw portion is formed over the entire length of the rotating shaft 15, and thereby the screw is engaged with the connecting member 16. Has been. The rotating shaft 15 forms a rotating shaft of the stepping motor 14 and is configured to move the connecting member 16 in the scanning direction (x direction) by rotating according to the rotational driving of the stepping motor 14. The drive circuit 17 controls the drive of the stepping motor 14 and outputs a pulse signal corresponding to the stepping motor to control the rotation drive (rotation angle and rotation speed) of the stepping motor 14. Note that the rotation drive control of the stepping motor 14 by the drive circuit 17 is performed based on a scanning drive control instruction signal from the main control unit 9.

  As described above, the scanning drive unit 13 drives the stepping motor 14 by the drive circuit 17 in accordance with the scanning drive control instruction of the main control unit 9, and moves the filter holding plate 11 in the x direction via the rotating shaft 15 and the connecting member 16. Move in the positive or negative direction. However, scanning by this movement is performed in a scanning step having a movement width of about 0.296 mm (calculated based on the change of the center wavelength of the WBPF 12 described above at about 17 nm / mm) corresponding to the center wavelength of the WBPF 12 of 5 nm. To be controlled.

  Regarding the movement of the filter holding plate 11 by the scanning drive unit 13, as described above, since the sub-transmission band removal filter 3 is provided separately from the WBPF 12, the movement by the scanning driving unit 13 is performed by the WBPF 12 and the filter holding plate. 11 can be performed, the scanning load and the moving distance by the scanning drive unit 13 can be reduced, so that the scanning drive unit 13 can be reduced in size and cost, or the scanning time (measurement time) can be shortened. .

  In the present embodiment, in each scanning step, the spectral illumination characteristics of the 5 nm pitch BPF sequentially set to each pixel of the second image 6a (the imaging area of the imaging device 7) are weighted and integrated, and the International Illumination Committee. The color matching functions x (λ), y (λ), and z (λ) recommended by (CIE) for the 2 ° field of view are synthesized, but the half-value width of the center wavelength of the WBPF 12 is set to about 15 nm, which increases the synthesis accuracy. For this reason (when the half-width is somewhat large, such as 15 nm, the synthesis accuracy is improved), the WBPF 12 employs a first-order interference BPF. In general, the primary interference BPF has a sub-transmission band in a wavelength region that is 1/2 and 2 times the center wavelength as described above.

Next, an actual scanning operation by the scanning WBPF 10 (scanning drive unit 13) will be described.
First, symbols used in the following description are summarized here.
x: coordinates in the scanning direction (x direction) of the first image plane 2b and the second image plane 6b (x = 0 to 7.4 mm)
y: coordinates in the direction (y direction) orthogonal to the scanning direction of the first image plane 2b and the second image plane 6b (y = 0 to 7.4 mm)
s: Scanning position of WBPF12 expressed in scanning step (s = 0 to 65)
i, j: x and y coordinates (i, j = 1 to 1000) in pixel units of the first image plane 2b and the second image plane 6b.
j ′: x-coordinate of pixel unit of WBPF12 plane (j ′ = 1 to 2560)
I _ij : Image information (I _ij ) _S : Image information in scanning step s

  As described above, the scanning of the filter holding plate 11 (WBPF 12) in the scanning WBPF 10 is performed with a step width of d = 40 * 7.4 μm = 0.296 mm (corresponding to 40 pixel columns) corresponding to 5 nm. The wavelength range of 320 nm (= 710-390 nm) in the visible range 390 to 710 nm is covered with 64 (= 320/5) step movement (64 * 0.296 = 18.9 mm; equivalent to 64 * 40 = 2560 pixel columns). can do. However, since the imaging area of the imaging device 7 has a length of 1000 pixel rows in the x direction, that is, a length corresponding to 25 steps, all the pixels (1000 pixel rows) of the imaging device 7 have the wavelength width of 320 nm (WBPF12). ), That is, in order that all pixels in the imaging region cover all wavelength regions (390 to 710 nm), at least 89 (= 64 + 25) steps of movement including these 25 steps are required. (Scanning) is required. Note that the initial position before the start of scanning (movement) is set to scanning step s = 0 (zero).

  This will be described with reference to FIGS. FIG. 4 is a conceptual diagram illustrating a captured image (exposure state) corresponding to each scanning step s in the imaging area of the imaging device 7. FIG. 5 shows the positional relationship between each part (WBPF 12 and the first and second light shielding parts 112 and 113) of the filter holding plate 11 moved in the scanning direction (x direction) in each scanning step s and the imaging area in the image sensor 7. It is a conceptual diagram explaining about. As shown in FIG. 4, before the start of scanning (before the start of spectral luminance measurement by the two-dimensional spectral luminance meter 1), the filter holding plate 11 (WBPF 12) is positioned at the initial position (s = 0). The imaging area (imaging area) of the first image plane 2 b is in a state where it is shielded (covered) by the first light shielding part 112 of the filter holding plate 11. This state corresponds to the position of the filter holding plate 11 (s = 0) indicated by reference numeral 501 in FIG. 5, and corresponds to the imaging region T (the width of the pixel row of j = 1 to 1000) in the first light shielding portion 112. ) Is located.

When scanning is started, the foreground image 401 (= image (I _ij ) ₀ ) shown in FIG. 4 is captured as a dark image (noise image or offset image) by the image sensor 7 at the position of scanning step s = 0. It is. Subsequently, the scanning drive unit 13 translates the filter holding plate 11 in the positive direction in the x direction by one scanning step (1 s) (corresponding to the moving position of the filter holding plate 11 indicated by reference numeral 502 in FIG. 5), and the s = 1, the first image 402 (= image (I _ij ) ₁ ) is captured. In this image (I _ij ) ₁ , monochromatic light having a center wavelength of 390 nm is applied to the first pixel column (j = 1) indicated by reference numeral 403 in the image (corresponding to the pixel column at the position indicated by reference numeral 503 in FIG. 5). Incident light is applied, and the other pixel columns (j = 2 to 1000) are shielded from light.

Thereafter, an image is captured for each scanning step. The pixel row j on which monochromatic light having a center wavelength of 390 nm is incident for each scanning step moves to the right (the positive direction side in the x direction), and the first pixel The wavelength of the light beam incident on the column (j = 1) increases by about 5 nm. Along with this, the number of pixel blocks shielded by the first light shield 112 decreases, and when the scanning step reaches s = 25, the first light shield 112 goes out of the imaging area. In the image (I _ij ) ₂₅ shown in the image 404 captured at s = 25 (corresponding to the moving position of the filter holding plate 11 indicated by reference numeral 504 in FIG. 5), the first to 1000th pixel rows (j = 1 to 1000) The center wavelength of the monochromatic light incident on the light beam changes continuously from 390 nm to 515 nm.

The scanning step s advances, and in the image (I _ij ) ₆₄ shown in the image 405 captured at s = 64 (corresponding to the moving position of the filter holding plate 11 indicated by reference numeral 505 in FIG. 5), the first to 1000th pixel columns are arranged. The corresponding center wavelength is changed from 585 nm to 710 nm. In the next s = 65, the first pixel row (j = 1) indicated by reference numeral 407 of the image 406 (the same pixel row as the first pixel row indicated by reference numeral 403; in FIG. When the scanning step s further proceeds, the light-shielding area (light-shielded pixel row) by the second light-shielding unit 113 increases, and the final scanning step s = 90 (FIG. 5 corresponds to the moving position of the filter holding plate 11 indicated by reference numeral 507), and as shown in an image 408 (a dark image similar to the image 401), the entire imaging area (all pixel rows j = 1 to 1000) is the second light-shielding portion. 113 is shielded from light. In this manner, the filter holding plate 11 and the WBPF 12 are moved stepwise in the scanning direction by the scanning WBPF 10 (scanning drive unit 13), and a captured image by the image sensor 7 is captured at each scanning step (however, at the above step s = 90) Captured images need not be captured).

  As described above, in the image captured at each scanning step s, the center wavelength of the monochromatic light forming each pixel of the image depends on the scanning step s and the x coordinate of the pixel, and the center wavelength is While changing continuously along the x (column number j) direction, it is substantially constant along the y (row number i) direction.

(Spectral image information processing)
By the way, the two-dimensional spectral luminance meter 1 obtains and outputs a tristimulus value image from the image (I _ij ) _S captured at each scanning step s. In general, in a multiband colorimeter, a color matching function x (λ), y (λ) of 2 ° field of view recommended by the CIE is performed by performing a process of multiplying different spectral sensitivities by multiplying them with an appropriate weight coefficient. ) And z (λ) are synthesized, but in a two-dimensional measuring instrument such as the two-dimensional spectroluminometer 1, the processing is executed for each pixel.

The calculation of the tristimulus value image from this image (I _ij ) _S will be described below.
The image (I _ij ) _S obtained in each scanning step s is converted into a luminance image (L _ij ) _S using the luminance calibration coefficient (C _ij ) _S of each pixel that is known in the luminance axis calibration described later. To do. The tristimulus values X _ij , Y _ij , and Z _ij are given as Wx _{S, j} , Wy _{S, j} , and Wz _{S, j} , respectively, in order to obtain the tristimulus values for the luminance image (L _ij ) _S. Are calculated by the following equations (1-1) to (1-3).
X _ij = Σ _S Wx _{S, j} * (L _ij ) _S (1-1)
Y _ij = Σ _S Wy _{S, j} * (L _ij ) _S (1-2)
Z _ij = Σ _S Wz _{S, j} * (L _ij ) _S (1-3)

  In general, in spectral luminance measurement using a multi-filter method (see Background Art), since light beams in the same wavelength band are incident on all pixels in one image, the same weight coefficient is applied to all pixels. In the method of the present embodiment, the transmission wavelength band of incident light is different for each scanning step s and for each pixel column, and accordingly, the weight coefficient for each pixel is also different.

Note that, from the luminance calibration coefficient (C _ij ) _S and the weight coefficient Wx _{S, j} , Wy _{S, j} , Wz _{S, j} , the weight coefficient Wx _{S, ij for} each pixel of the acquired image in each scanning step, It is also possible to obtain Wy _{S, ij} and Wz _{S, ij} in advance and calculate the tristimulus values X _ij , Y _ij , and Z _ij by the following equations (2-1) to (2-3). Thereby, the calculation time for the calculation of the tristimulus values can be further shortened.
X _ij = Σ _S Wx _{S, ij} * (I _ij ) _S (2-1)
Y _ij = Σ _S Wy _{S, ij} * (I _ij ) _S (2-2)
Z _ij = Σ _S Wz _{S, ij} * (I _ij ) _S (2-3)

  The integration is performed at every scanning step, and a tristimulus value image is obtained at the end of scanning. This tristimulus value image is stored (stored) in the main control unit 9 (RAM or the like). In this method, all the 89 monochromatic light images (captured images at the scanning steps s = 1 to 89 shown in FIG. 4) acquired by imaging with the 5 nm pitch are stored, and after the scanning is completed, three images are collected. Compared with the method of converting to a stimulus value image, the required data storage capacity can be greatly reduced.

FIG. 11 is a flowchart illustrating an example of an operation related to measurement of a tristimulus value image by the two-dimensional spectral luminance meter 1. First, the scanning WBPF 10 (filter holding plate 11) is set (moved) to the initial position (s = 0) by the main control unit 9 (scan driving unit 13) (step S1), and tristimulus values X _ij and Y _{ij are set.} , Z _ij is set to 0 (step S2), and the dark image (I _{i, j} ) ₀ is passed through the image signal processing unit 8 in a state where the imaging area of the image sensor 7 is shielded from light at s = 0. Is taken into the main controller 9 (step S3). Next, similarly, the scanning WBPF 10 (filter holding plate 11) is moved by one scanning step (step S4), and the image (I _ij ) _{S at} s = 1 is captured (step S5). Then, dark image correction ((I _ij ) _S = (I) using the dark image (I _ij ) ₀ captured in step S3 with respect to the captured image (I _ij ) _S by the main control unit 9. _ij ) _S- ( _Iij ) ₀ ) is performed (step S6), and the dark image corrected image ( _Iij ) _S is calculated and stored by luminance axis calibration (to be described later). ) Luminance correction using the luminance calibration coefficient (C _ij ) _S ((L _ij ) _S = (C _ij ) _S * (I _ij ) _S is performed and converted into a luminance image (L _ij ) _S (step S7) ).

The luminance image (L _ij ) _S is multiplied by the weight coefficient Wx _{S, j} , Wy _{S, j} , Wz _{S, j of} each stimulus value that differs for each pixel column of the captured image for each scanning step s. ( _Xij = _Xij + WxS _{, j} * ( _Lij ) _S , _Yij = _Yij + WyS _{, j} * ( _Lij ) _S , _Zij = _Zij + WzS _{, j} * (L _ij ) _S ) (step S8). If the main control unit 9 determines that the scanning step s is not greater than 89 (NO in step S9), the process returns to step S4 and is further moved by one scanning step, and step S5 for the next scanning step s. The operation of ~ S8 is repeated. However, in step S8 in this repetition, further integration (update of the integrated value) is performed on the tristimulus values X _ij , Y _ij , and Z _ij obtained in the previous scanning step. If it is determined that the scanning step s is greater than 89 (YES in step S9), the tristimulus value images (X _ij , Y _ij , Z _ij− ) obtained by integration over the entire scanning step s. Is stored in a predetermined data storage unit (such as the RAM of the main control unit 9) (step S10), and the flow ends.

In the embodiment shown in the flowchart, the later-described stray light correction is not performed on the captured image (I _ij ) _S. However, even when this stray light correction is not performed, the image (I _ij ) _S is stored, converted into a luminance image (L _ij ) _S using the luminance calibration coefficient (C _ij ) _S of each pixel (formula (6-1) described later), and wavelength axis calibration described later. The spectral brightness image L _ij (λ), which will be described later, may be obtained by a known calculation process using the relative spectral sensitivity S _{S, j} (λ) of each pixel row j known in FIG.

By the way, the two-dimensional spectroluminometer 1 performs wavelength axis calibration and luminance axis calibration on the WBPF 12 in advance (for example, at the time of shipment), and the calculation data such as the sensitivity calibration coefficient (C _ij ) _S obtained by this calibration is mainly used. It is set (stored) in the control unit 9 (RAM or the like), and at the time of actual spectral luminance measurement, the captured image (observed image) is converted into a luminance image (luminance correction) using the calculation data. Done. The wavelength axis calibration and luminance axis calibration will be described below.

<Wavelength axis calibration>
Calibration of the wavelength axis to the WBPF12, for example obtains a reference intensity M _lambda using a calibration system 30 shown in FIG. 6, each pixel column in each scan step s of a two-dimensional spectral luminance meter 1 on the basis of the reference intensity M _lambda This is done by determining the spectral sensitivity of j. That is, in the calibration system 30, the reference single-wavelength light source 31 is controlled by a control PC (personal computer) 32. For example, a total of 201 single-wavelength light beams 31a having a sufficiently narrow half-value width from 350 to 750 nm at a pitch of 2 nm, The light is output to the incident aperture 331 of the integrating sphere 33. The single wavelength light beam 31a is subjected to multiple diffuse reflection within the integrating sphere 33 to form a uniform luminance surface at the output aperture 332 of the integrating sphere 33. The to-be-calibrated two-dimensional spectral luminance meter 34 as the two-dimensional spectral luminance meter 1 to be calibrated acquires image information (single wavelength image) for each single-wavelength light beam 31a by measuring this uniform luminance plane. Is transmitted (output) to the control PC 32. The uniform luminance surface for each of the single-wavelength light beam 31a, which is measured by the reference spectral luminometer 35 of the spot-type, the reference information of the intensity M _lambda obtained by the reference spectral luminance meter 35 measurements may control PC32 Sent to. The reference numeral 36 of the calibration system 30 indicates a white light source (outputting a white light beam 36a incident on the integrating sphere 33 in place of the single wavelength light beam 31a of the reference single wavelength light source 31 in the later-described luminance axis calibration. The white light source 36) is not used in this wavelength axis calibration (in this case, the white light source 36 and the incident aperture 333 of the integrating sphere 33 may not be provided).

  4 and 5, the effective area in the scanning direction of the WBPF 12 corresponds to 2560 pixel columns (j ′ = 1 to 2560), and the image sensor 7 having 1000 pixel columns (j = 1 to 1000). Therefore, the entire effective area of the WBPF 12 cannot be measured by imaging at one scanning step position. Therefore, scanning with the single wavelength light beam 31a (light beam from the uniform luminance surface) from the reference single wavelength light source 31 is performed at three scanning step positions of scanning steps s = 25, 45 and 65, so that all of the WBPF 12 can be scanned. The spectral sensitivity in the effective range is obtained (for each pixel).

First, in the calibrated two-dimensional spectrophotometer 34, the WBPF 12 (filter holding plate 11) is fixed at the position of the scanning step s = 25, the reference single-wavelength light source 31 is wavelength-scanned, and 201 single-wavelength images taken are captured. the integrated value of the pixel rows and scaled by the reference intensity M _lambda, in WBPF12, the range of the pixel column j '= 1 to 1000 corresponding to the pixel columns of 1000 of the imaging region (j = 1 to 1000) relative Obtain spectral sensitivity. Similarly, pixel rows j ′ = 801 to 1800 and j ′ = 1601 to 2560 of the WBPF 12 are respectively calculated from the pixel row integrated values of the single-wavelength image group scanned at a fixed wavelength at the scanning steps s = 45 and 65. The relative spectral sensitivity of the range to be calculated is obtained. Then, from the relative spectral sensitivities at the scanning step positions of s = 25, 45, and 65, relative spectral sensitivities S _{j ′} (2 nm pitch) with respect to the pixel rows (j ′ = 1 to 2560) in the entire effective area of the WBPF 12. λ). However, from the pixel column ranges in the case of s = 25, 45, and 65, that is, the pixel columns j ′ = 1 to 1000, 801 to 1800, and 1601 to 2560, one entire effective region pixel column (j ′ = 1) ˜2560) when creating relative spectral sensitivities, there are regions that overlap each other (about 1/3 each), and the two data (relative spectral sensitivity information) of the overlapping regions are May be adopted.

As shown in FIG. 7, the relative spectral sensitivity S _{S, j} (λ) of each pixel column j of the image sensor 7 (imaging area) at the scanning step s position at the time of measurement is all effective of the WBPF 12 obtained as described above. The relative spectral sensitivity S _{j ′} (λ) of the 2560 pixel array in the area is given as S _{j ′} (λ) (where j ′ = 40 * s−j + 1) translated in accordance with the position of each scanning step s. . 7, relative spectral sensitivity waveforms indicated by reference numerals 701, 702, 703,... Are for the respective scanning steps s obtained by the above-mentioned translation, and each of the relative spectral sensitivity waveforms 701, 702, 703,. The interval is 40 pixels corresponding to one scanning step. In addition, the horizontal axes indicated by reference numeral 704 are the same, and the graphs 710 to 730 are vertically displayed in order to clearly show the relative spectral sensitivity waveforms of the low wavelength region, the middle wavelength region, and the high wavelength region at the center wavelength of 350 nm to 750 nm. Are shown side by side. However, since the spectral sensitivity (spectral transmittance) is the same in the pixel column direction (y direction), the averaged value in the pixel column direction is the (relative) spectral sensitivity of the pixel column.

The weight coefficient Wx _{S, j} , Wy _{S, j} , Wz _{S, j} for the scanning step position s of the pixel row j used in the above formulas (1-1) to (1-3) is the relative spectral sensitivity S _{Using S, j} (λ), combined spectral sensitivities x _j (λ), y _j (λ), z _j (λ) calculated by the following equations (3-1) to (3-3),
x _j (λ) = Σ _S Wx _{S, j} * S _{S, j} (λ) (3-1)
y _j (λ) = Σ _S Wy _{S, j} * S _{S, j} (λ) (3-2)
z _j (λ) = Σ _S Wz _{S, j} * S _{S, j} (λ) (3-3)
The sum of squares of error Ex _j and Ey _{j for} each wavelength shown in the following equations (4-1) to (4-3) based on the theoretical values x (λ), y (λ), and z (λ) of the color matching functions. , Ez _j is obtained as a weight coefficient Wx _{S, j} , Wy _{S, j} , Wz _{S, j} that minimizes Ez _j .
_{Ex j = Σ λ [x j} (λ) -x (λ)] 2 ... (4-1)
_{Ey j = Σ λ [y j} (λ) -y (λ)] 2 ... (4-2)
_{Ez j = Σ λ [z j} (λ) -z (λ)] 2 ... (4-3)

FIG. 12 is a flowchart illustrating an example of an operation related to wavelength axis calibration. First, the scanning WBPF 10 (filter holding plate 11) of the calibrated two-dimensional spectroluminometer 34 disposed opposite to the uniform luminance surface formed by the integrating sphere 33 is set (moved) to the initial position (s = 0) (step S21). Then, the dark image (I _ij ) ₀ is captured in a state where the imaging area of the imaging device 7 is shielded from light (step S22). Next, the scanning WBPF 10 is moved to the position of s = 25 (steps S23 and S24), and the reference single wavelength (center wavelength λ) is set to a predetermined initial value (for example, 348 nm) by the reference single wavelength light source 31 (or control PC 34). (Step S25). Then, single wavelength light having a center wavelength 2 nm larger than the initially set center wavelength λ is output from the reference single wavelength light source 31 (step S26), and an integrating sphere corresponding to the center wavelength λ output in step S26. An image (I _ij ) _{S, λ of} 33 uniform luminance surfaces is captured (step S27).

On the other hand, the reference intensity _Mλ is measured by the spot type standard spectral luminance meter 35 (step S28). Then, the image (I _ij ) _{S, λ} captured in step S27 is dark image corrected using the dark image (I _ij ) ₀ ((I _ij ) _{S, λ} = (I _ij ) _{S, λ} − ( I _ij ) ₀ ) (step S 29), and further normalized ((I _ij ) _{S, λ} = (I _ij ) _{S, λ} / M _λ ) using the reference intensity M _λ obtained in step S 28. (Step S30). When the center wavelength λ of the monochromatic light output from the reference single wavelength light source 31 is not larger than 750 nm (NO in step S31), the process returns to step S26, and is further increased by 2 nm and output from the reference single wavelength light source 31. The image for the single wavelength light of the next center wavelength λ is taken in, and the measurement (detection) of the reference intensity M _λ , dark image correction, and standardization processing are performed. In this way, single wavelength light increased at a pitch of 2 nm over the center wavelength of 350 nm to 750 nm is sequentially output, and the operations of steps S27 to S30 for the single wavelength light are repeated.

If the center wavelength λ output from the reference single wavelength light source 31 is greater than 750 nm (YES in step S31), then if the scanning step s is not greater than 65 (NO in step S32), the process returns to step S24, The scanning WBPF 10 (filter holding plate 11) is further moved from the current scanning step s position by a scanning step “20” in the scanning direction (positive direction in the x direction), that is, the position of scanning step s = 25 in the first step S24. Are sequentially moved to the position of scanning step s = 45, and the operations of steps S26 to S30 are repeated until the center wavelength λ output at a pitch of 2 nm reaches 350 nm to 750 nm. Similarly, the operations in steps S26 to S30 are repeated for the scanning step s = 65 that has been further moved by 20 scanning steps. As a result, each of the scanning steps 25, 45 that are standardized in step S30, and Images (I _ij ) _{25, λ} , (I _ij ) _{45, λ} and (I _ij ) _{65, λ} for ₆₅ are obtained and stored.

When the scanning step s becomes larger than 65 in step S32 (YES in step S32), that is, after the measurement is completed, the stored image (I _ij ) _{25, λ} , (I _ij ) _{45, λ} and (I _ij ) A spectral sensitivity image (I _{ij ′} ) _λ for a pixel row (j ′ = 1 to 2560) in the entire effective range of the WBPF 12 is _calculated from _{65, λ} (step S33). Further, the spectral sensitivity image (I _{ij ′} ) _λ is integrated with respect to the pixel row i ((I _{j ′} ) _λ = Σ _i (I _{ij ′} ) _λ ), and the spectral sensitivity image (I _{j ′} ) of all the pixel columns _{j ′.} ) _Λ is obtained and stored (step S34).

<Luminance axis calibration>
Next, luminance axis calibration will be described. What is calibrated by the above-described wavelength axis calibration is the relative spectral sensitivity (relative spectral sensitivity) of each pixel row j at each scanning step position (calibration as absolute luminance is not performed, Since the sensitivity unevenness is not corrected), the luminance axis (sensitivity axis) is calibrated here. This luminance axis calibration is performed using, for example, the calibration system 30 shown in FIG. However, instead of the reference single wavelength light source 31 in the case of wavelength axis calibration, the luminous flux of a white light source 36 such as an A light source is incident on the integrating sphere 33 to form a uniform luminance surface at the output aperture 332, and the two-dimensional spectral luminance to be calibrated This uniform luminance surface is measured by a total of 34. This uniform luminance surface is simultaneously measured by the standard spectral luminance meter 35 to obtain the reference spectral luminance L ₀ (λ).

The output Σ _λ L ₀ to be obtained when the pixel P _ij of the pixel row j in the scanning step s having the relative spectral sensitivity S _{S, j} (λ) measures the light source having the reference spectral luminance L ₀ (λ). Using (λ) * S _{S, j} (λ) and the output (I _ij ) _S obtained by actual measurement in the luminance axis calibration, the sensitivity calibration coefficient (C _ij ) of the pixel P _ij at the scanning step s. _S is obtained by the following equation (5-1).
_{(C ij) S = [Σ} λ L 0 (λ) * S S, j (λ)] / (I ij) S ... (5-1)

At the time of actual measurement (spectral luminance measurement) by the two-dimensional spectral luminance meter 1, the sensitivity calibration coefficient (C _ij ) _S is used to output the pixel P _ij at the scanning step s for the two-dimensional light source L to be measured ( I _ij ) _S is converted into a luminance image (L _ij ) _S by the following equation (6-1) and output.
(L _ij ) _S = (C _ij ) _S * (I _ij ) _S (6-1)

(Stray light correction)
By the way, the WBPF 12 is a multilayer interference filter, and in general, component light in a wavelength region other than the transmission wavelength band is reflected. The component reflected by the WBPF 12 is further reflected by the lens surface of the optical system and the inner surface of the housing, and becomes so-called “stray light” and reenters the WBPF 12. The component in the transmission wavelength band at this incident position passes through the WBPF 12 and reaches the image sensor 7, and the re-incident light that does not pass through is incident again, and the above process is repeated. In this way, the stray light image due to the stray light that has reached the image sensor 7 is superimposed on the original observation image and deteriorates the accuracy of the observation image. Therefore, in order to eliminate the influence of the stray light image on the observation image (the image (I _ij ) _{S acquired in} each scanning step s described above), stray light correction may be performed as described below.

  The symbols used in the following explanation regarding the stray light correction are organized below.

m, n: the coordinates (m, n) of the measurement unit light at the incident positions of the first image plane 2b and the second image plane 6b
= 1 to 1000)
I, J: pixel set coordinates (I, J = 1 to 25) binned by 40 × 40 pixels on the first image plane 2b and the second image plane 6b
M, N: pixel set coordinates (M, N = 1 to 25) of the incident positions of the first image plane 2b and the second image plane 6b of the light to be measured
R _ij (λ) or R _IJ (λ): real image on which no stray light image is superimposed (image from which stray light has been removed)
R ′ _ij (λ): Approximate image of real image R _ij (λ) B _ij (λ) or B _IJ (λ): stray light image O _ij (λ) or O _IJ (λ): observation with stray light image superimposed Image (actually measured image), that is, O _ij (λ) = R _ij (λ) + B _ij (λ) or O _IJ (λ) = R _IJ (λ) + B _IJ (λ)
(F _mn ) _ij (λ) or (F _MN ) _IJ (λ): a response image, an image obtained by incident light of unit intensity incident on the pixel P _mn or the pixel set P _MN Q _IJ (λ): an actual image Corrected image O ′ _IJ (λ) estimated to approximate: pseudo-observation image calculated from the corrected image and the response image

FIG. 8 shows an example of the intensity distribution of the real image R _ij (λ), the observed image O _ij (λ), and the stray light image B _ij (λ). However, this intensity distribution is along the j coordinate (x direction). As shown in the figure, the observation image O _ij (λ) is obtained by superimposing the stray light image B _ij (λ) on the real image R _ij (λ). However, the distribution of the stray light image B _ij (λ) that is not imaged is gentle (substantially uniform), and its spatial frequency is less than the actual image R _ij (λ) obtained by imaging. It is a very low level.

(Response image)
Here, the response image will be described.
A light beam having a unit intensity (imaging light beam) incident on a specific pixel P _mn in the measurement area, that is, the first image plane 2 b and the second image plane 6 b of the light to be measured, on all the pixels P _ij in the imaging area of the image sensor 7. The image created in (5) is referred to as a response image (F _mn ) _ij . This response image (F _mn ) _ij includes the influence of the “stray light” on the point light source imaged at the coordinates (m, n) of the incident position. It can be said that it is a function that indicates whether to give. By the way, the level of stray light that is incident light to pixels other than the incident pixel (a certain one pixel P _mn ) is (1) the part (pixel P _mn ) where the imaging light beam is incident and (2) stray light. Since it depends on the incident part (pixel P _ij ) and (3) depends on the wavelength λ of the imaging light beam (because reflection / transmission involving stray light generally has wavelength dependence), the response image (F _mn ) _Ij is expressed as a function of wavelength λ such as (F _mn ) _ij (λ).

When the response image is defined in this way, the observation image O _ij (λ) actually measured (observed) by the two-dimensional spectroluminometer 1 is the actual image R having the intensity of the light beam to be incident on the incident pixel P _{mn. The} response image (F _mn ) _ij (λ) weighted by _mn (λ) is integrated for all incident pixels P _mn , and is expressed by the following equation (7-1).
O _ij (λ) = Σ _mn [R _mn (λ) * (F _mn ) _ij (λ)] (7-1)

(Real image estimation)
In the stray light correction, the actual image R _ij (λ) is obtained by removing the influence of the stray light from the observation image O _ij (λ). Here, the real image R _ij (λ) is obtained from the observation image O _ij (λ). _The estimation of _ij (λ) will be described.

As the estimated actual image R ij (λ), the influence of the stray light image is obtained from the observation image O ij (λ) and the response image (F mn ) ij (λ) based on the above equation (7-1). A corrected image that approximates the removed real image R ij (λ) is obtained. First, the initial value R ij approximate corrected image to the actual image R ij (λ) (λ) (0) to you and O mn (λ) (R ij (λ) (0) = O mn (λ)) . The initial value R ij (λ) (0) and the response image (F mn) ij (λ) pseudo observed image is calculated from the Σ mn [R ij (λ) (0) * (F mn) ij (λ) ] And an actual observation image O ij (λ) Σ mn [R ij (λ) (0) * (F mn ) ij (λ)] − O ij (λ) by an initial value R ij (λ) By correcting (0) , a first-order (first-order approximate solution) corrected image R ij (λ) (1) is obtained (the following equation (8-1)).
R ij (λ) (1) = R ij (λ) (0) −a * [Σ mn [R ij (λ) (0) * (F mn ) ij (λ)] − O ij (λ)] (8-1)

Further, R _ij (λ) ⁽⁰⁾ on the right side of this equation is replaced with R _ij (λ) ⁽¹⁾ , and a second-order corrected image R _ij (λ) ⁽² ) shown in the following equation (9-1) is obtained. ⁾ Is required.
R _ij (λ) ⁽²⁾ = R _ij (λ) ⁽¹⁾ −a * [Σ _mn [R _ij (λ) ⁽¹⁾ * (F _mn ) _ij (λ)] − O _ij (λ)] (9-1)

Then, such calculation is repeated, and a k-th order corrected image R _ij (λ) ^(k) is given by the following equation (10-1) using R _mn (λ) ^(k−1). .
R _ij (λ) ^(k) = R _ij (λ) ^(k−1) −a * [Σ _mn [R _ij (λ) ^(k−1) * (F _mn ) _ij (λ)] − O _ij ( λ)]… (10-1)

The symbol “a” in the second term on the right side in the above formulas (8-1) to (10-1) is a constant that gives an asymptotic characteristic, and its value is selected experimentally (so that the asymptotic formula converges). Is done. With the above process, the first k-th order corrected image R _ij (λ) ^{(k) in which} the square sum for each pixel of the second term on the right-hand side that is the correction term (the following expression (11-1)) is less than the allowable value is used. Let it be an approximate image R ′ _ij (λ). However, the k-th order corrected image R _ij (λ) ^(k) (= approximate image R ′ _ij (λ)) is a corrected image that approximates the above-described actual image R _ij (λ), that is, the estimated actual image R. _ij (λ) is shown.
E ^(k-1) = _Σij [ _Σmn [ _Rij (λ) ^(k-1) * ( _Fmn ) _ij (λ)]- _Oij (λ)] ² (11-1)

The stray light image is given by B _ij (λ) in the following expression (12-1) using the k-th order corrected image R _ij (λ) ^(k) (from the observed image O _ij (λ): A stray light image B _ij (λ) is ^obtained by subtracting the estimated actual image R _ij (λ) ^(k) ).
B _ij (λ) = O _ij (λ) −R _ij (λ) ^(k) (12-1)

(Actual calculation)
If the above calculation is performed for 10 ⁶ pixels of the image sensor 7, 10 ¹² data will be involved in the primary calculation, and the amount of data and the calculation time will be enormous. Therefore, first, a stray light image is obtained, and using the following (13-1) obtained by modifying the above equation (12-1), the stray light image G _ij (λ) (the stray light image B described above) is obtained from the observed image O _ij (λ). _ij (lambda) in equivalent) to reduce by taking a method for obtaining an approximate image R _{'ij (λ).}
R _ij (λ) ^(k) = O _ij (λ) −G _ij (λ) (13-1)

  As described in FIG. 8 above, the background of using this method is that the stray light image does not contain a high spatial frequency component, and has sufficient accuracy even with low pixel count image information, and the number of pixels is reduced (low pixel count). The amount of image information and the calculation time can be greatly reduced. In addition, the obtained stray light image with the low number of pixels is converted into a stray light image with the original number of pixels (total number of pixels) by performing known processing such as smoothing, and this is converted from the observation image (having the total number of pixels). The approximate image (corrected image approximated to the actual image) obtained by subtraction also has sufficient accuracy.

Therefore, in the present embodiment, as shown in FIG. 9, 40 × 40 pixels are binned (integrated), and stray light is used using image information based on 25 × 25 (= I × J and M × N) pixel sets. Ask for an image. That is, the pixel coordinates (i, j) and (m, n) (i, j, m, n = 1 to 1000) described above are converted into the pixel set coordinates (I, J) and (M, N) (I, J , M, N = 1 to 25), and accordingly, the observed image O _ij (λ), the actual image R _ij (λ), the response image (F _mn ) _ij (λ), and the stray light image B _ij (λ) ) Are replaced with O _IJ (λ), R _IJ (λ), (F _MN ) _IJ (λ), and B _IJ (λ) that have been subjected to binning processing of 40 × 40 pixels. By handling the binned information in each calculation, the amount of image information and the calculation time can be reduced or shortened to, for example, 10 ⁻⁴ / 256, and can be suppressed to a practical level.

(Measurement of response image)
In the calculation in the spectral luminance measurement by the two-dimensional spectral luminance meter 1, as described above, the actual image obtained by removing the influence of the stray light image from the observation image O _ij (λ) and the response image (F _mn ) _ij (λ). although determine a corrected image which approximates to the image R _{ij (lambda)} (estimated real image R _{ij (lambda)),} information in the response image used in this calculation, previously obtained in advance by (e.g. at the factory) Found . Hereinafter, actual measurement of the response image will be described.

  The actual measurement of the response image is performed using, for example, a calibration system 30a shown in FIG. However, the calibration system 30a is further provided with a pinhole plate 37 and a moving table 38 with respect to the calibration system 30 shown in FIG. As shown in FIG. 10, when actually measuring the response image, first, the luminous flux 36 a of the white light source 36 is incident on the integrating sphere 33, and the pinhole plate 37 is disposed at the position (near the output) 332 of the integrating sphere 33. Thus, a point light source is formed. A moving table 38 configured to be able to move the light (white light) from the point light source in the x direction and the y direction (in directions indicated by reference numeral 39 in FIG. (A view) when viewed in the direction of arrow A). Measurement is performed by a to-be-calibrated two-dimensional spectral luminance meter 34 installed above. By controlling the moving table 38 by the control PC 32, the image of the point light source can be converted into arbitrary coordinates (pixel set coordinates (M, N); M = 1 to 25) on the first image plane 2b and the second image plane 6b. , N = 1 to 25) (for example, refer to a point light source P in a certain pixel set 901 shown in FIG. 9), the position of the calibrated two-dimensional spectral luminance meter 34 is set (moved). be able to. The spectral intensity of the white light emitted from the pinhole of the pinhole plate 37 is measured by the reference spectral luminance meter 35, and the relative spectral intensity M (λ) is obtained from the monitor output by the measurement.

FIG. 13 is a flowchart showing an example of an operation related to the actual measurement of the response image. First, the position of the moving table 38 is controlled by the control PC 32, and the point light source is in the initial position, that is, the pixel set coordinates (M, N) of the first image plane 2b (or the second image plane 6b) are M = 1, for example. It is set (moved) to the position of N = 1 (step S41). When the point light source is at the pixel set coordinate position, the filter holding plate 11 (WBPF 12) is set to the initial position of the scanning step s = 0 by the scanning driving unit 13 in the scanning WBPF 10 (step S42), and this scanning step s = At 0, a dark image (I _ij ) ₀ is picked up by the image pickup device 7 and taken into the main control unit 9 via the image signal processing unit 8 (step S43). Next, the scanning step s is moved by one step (step S44). Similarly, in this scanning step s, the image (I _ij ) _{S, MN} is picked up and captured (step S45), and the image (I _ij ) _{S, MN is} captured. Are corrected and stored ((I _ij ) _{S, MN} = (I _ij ) _{S, MN} − (I _ij ) ₀ )) (step S 46).

If the scanning step s is not greater than 89 (NO in step S47), the process returns to step S44 and is moved to the next scanning step s, and the image (I _ij ) _{S, MN} is captured for each scanning step s. The operation of performing dark image correction and saving is repeated. If the scanning step s is greater than 89 (YES in step S47), the dark image corrected and stored image (I _ij ) _{S, MN} is converted into a spectral intensity image (I _ij ) _MN (λ) ( In step S48), this spectral intensity image (I _ij ) _MN (λ) is a spectral intensity image (I _IJ ) _MN (λ) of pixel set coordinates (I, J) in which the number of data is reduced by binning of 40 × 40 pixels. ) And stored (step S49).

Next, the M coordinate of the point light source is advanced by one (M = M + 1, N = N) (step S50). If the M coordinate value is not larger than 25 (NO in step S51), this pixel set is set. At the coordinate position, the process returns to step S42, and the filter holding plate 11 (WBPF 12) is set to the initial position of the scanning step s = 0. Similarly, image capture, dark image correction, and binning are performed. If the coordinate value of M is greater than 25 (YES in step S51), the M coordinate of the point light source is returned to 1 and the N coordinate is advanced by 1 (M = 1, N = N + 1) (step S52). When the coordinate value of N is not larger than 25 (NO in step S53), the same operation is performed by returning to step S42 at the pixel set coordinate position. In this way, the pixel set coordinates (M, N) are sequentially moved from the initial position (1, 1) to (25, 25), and the operations of steps S42 to S49 are repeated at each coordinate position. When the coordinate value of N becomes a value larger than 25 and the spectral intensity image (I _IJ ) _MN (λ) with respect to the incidence of the point light source on all the pixel set coordinates (M, N) is obtained (in step S53) YES), the relative spectral intensity M (λ) of the point light source is measured by the spot-type reference spectral luminance meter 35 (step S54). The spectral intensity image (I _IJ ) _MN (λ) is normalized by the relative spectral intensity M (λ) for each wavelength ((F _MN ) _IJ (λ) = (I _IJ ) _MN (λ) / M (λ )) And stored as a response image (F _MN ) _IJ (λ) (step S55), and the flow ends.

(Stray light correction during measurement)
Here, an example of calculation when stray light correction is performed at the time of measurement by the two-dimensional spectral luminance meter 1 will be described (in the embodiment shown in the flowchart shown in FIG. 11, stray light correction is not performed, but here. In the embodiment, stray light correction is performed). When stray light correction is performed at the time of the measurement, first, a light image of the two-dimensional light source L is picked up at every scanning step s of the scanning WBPF 10 (WBPF 12), and dark image correction ((I _ij ) _S = (I _ij ) _S − The image (I _ij ) _S that has been _subjected to (I _ij ) ₀ ) is stored as an observation image (O _ij ) _S. After scanning end, is the observation image _{(O ij) S} the spectral image (as a function of wavelength lambda) was converted to the observation image _{O ij} (λ), 40 × further binned observed image _{O ij} (lambda) The image is converted into an observation image O _IJ (λ) by a pixel set of 40 pixels. Then, according to Equation (7-1) mentioned above, as shown in the following equation (14-1), the observed image O _IJ that the binning _(lambda) assuming corrected image Q _{IJ (lambda),} the correction Q _MN (λ) corresponding to the image Q _IJ (λ) (that is, Q _MN (λ) which is the intensity of the light beam to be incident on the incident pixel set in the M coordinate), and the measured and stored response image A pseudo observation image O ′ _IJ (λ) is calculated from (F _MN ) _IJ (λ).
O ′ _IJ (λ) = Σ _MN [Q _MN (λ) * (F _MN ) _IJ (λ)] (14-1)

Next, as shown in the following formula (15-1), the difference between the pseudo-observed image O ′ _IJ (λ) and the observed image O _IJ (λ) [O ′ _IJ (λ) −O _IJ (λ)] Thus, the corrected image Q _IJ (λ) is re-corrected, and this is newly set as a corrected image Q _IJ (λ).
Q _IJ (λ) = Q _IJ (λ) −a * [O ′ _IJ (λ) −O _IJ (λ)] (15-1)

Then, the sum of squares E (λ) of the differences between the pseudo observation image O ′ _IJ (λ) and the observation image O _IJ (λ) shown in the following formula (16-1) is set as a limit value set in advance. The first corrected image Q _IJ (λ) that is smaller than Et is adopted as the final corrected image,
E (λ) = Σ _IJ [O ′ _IJ (λ) −O _IJ (λ)] ² (16-1)
When E (λ) takes a value equal to or greater than the limit value Et, the above formulas (14-1), (15-1), and (16-1) are repeated until the value becomes smaller than the limit value Et.

When the final corrected image Q _IJ (λ) is obtained, the stray light image B _IJ (λ) is obtained from the difference from the observed image O _IJ (λ) (the following equation (17-1)).
B _IJ (λ) = O _IJ (λ) −Q _IJ (λ) (17-1)

Since this B _IJ (λ) is a stray light image by the binned pixel set, it is converted into a stray light image B _ij (λ) consisting of all original pixels (before binning) by performing a smoothing process. As described above, since the stray light image has a low spatial frequency, there is a large error between the stray light image B _ij (λ) obtained by smoothing the stray light image B _IJ (λ) and the stray light image of the equation (12-1) using all pixels. without therefore the observed image _{O ij} (lambda) by subtracting the stray light image _{B ij} (lambda) obtaining an approximate image R _'ij (lambda) (the following formula (18-1)).
R ′ _ij (λ) = O _ij (λ) −B _ij (λ) (18-1)

As described above, from the image (I _ij ) _S (observed image O _ij (λ)) obtained by imaging the two-dimensional light source L for each scanning step s and the response image (F _MN ) _IJ (λ) stored in advance. Then, the stray light image B _ij (λ) is calculated by the calculation, and the stray light correction is performed by subtracting the stray light image B _ij (λ) from the observation image O _ij (λ), thereby obtaining the approximate image R ′ _ij (λ).

Incidentally, a spectral luminance image L _ij (λ), which will be described later, can be obtained by performing the above-described luminance correction on the approximate image R ′ _ij (λ). The calibration coefficient used in the brightness correction needs to be obtained in advance by the same brightness axis calibration as described above. Specifically, in the same manner as the calculations in (14-1) to (18-1) above, a measurement image of a uniform luminance surface by the white light source 36 is acquired (in FIG. 6), and this measurement image is corrected for stray light. Thus, an approximate image R ′ _ij (λ) similar to the above is calculated. On the other hand, the measurement image is simultaneously measured by the spot-type reference spectral luminance meter 35 to obtain the reference spectral luminance L ₀ (λ), and these approximate images R ′ are obtained as shown in the following equation (19-1). A calibration coefficient C _ij (λ) as a function of the wavelength λ is obtained from _ij (λ) and the reference spectral luminance L ₀ (λ), and is stored in the main controller 9 or the like.
C _ij (λ) = L ₀ (λ) / R ′ _ij (λ) (19-1)

At the time of actual measurement by the two-dimensional spectrophotometer 1, the measurement image of the two-dimensional light source L is corrected by stray light by the above formulas (14-1) to (18-1) to obtain the approximate image R ′ _ij (λ). The stored calibration coefficient C _ij (λ) is used to convert to a spectral luminance image L _ij (λ) and output (the following equation (20-1)).
L _ij (λ) = C _ij (λ) * R ′ _ij (λ) (20-1)

FIG. 14 is a flowchart showing an example of the operation when stray light correction is performed at the time of measurement by the two-dimensional spectral luminance meter 1 described above. First, an image (I _ij ) _{S that} has been captured at each scanning step s of the scanning WBPF 10 (WBPF 12) and corrected for dark image ((I _ij ) _S = (I _ij ) _S − (I _ij ) ₀ ) is an observation image ( O _ij ) _S is stored (step S61), and the observed image (O _ij ) _S is converted into an observed image O _ij (λ) (step S62). Next, the observed image O _ij (λ) is binned and converted into an observed image O _IJ (λ) based on a 40 × 40 pixel set (step S63). Then, the observed image O _IJ (λ) is set as a corrected image Q _IJ (λ) (step S64), and Q _MN (λ) corresponding to the corrected image Q _IJ (λ) is obtained by the above equation (14-1). And a response image (F _MN ) _IJ (λ) actually measured and stored, a pseudo observation image O ′ _IJ (λ) is calculated (step S65).

Next, a sum of squares E (λ) of differences between the pseudo observation image O ′ _IJ (λ) and the observation image O _IJ (λ) shown in the above equation (16-1) is obtained (step S66). If the value of E (λ) is not smaller than the preset limit value Et (NO in step S67), the pseudo-observation image O ′ _IJ (λ) and the observation image O in the above equation (15-1). _The corrected image Q _IJ (λ) is re-corrected by the difference (O ′ _IJ (λ) −O _IJ (λ)) from _IJ (λ), and this is used as a new corrected image Q _IJ (λ) (step S68). ), The process returns to step S65, and the operations of steps S65 to S68 are repeated until E (λ) becomes smaller than the limit value Et. When the value of E (λ) is smaller than the limit value Et (YES in step S67), the first corrected image Q _IJ (λ) that is smaller than the limit value Et is determined as the final corrected image, and As shown in Expression (17-1), the correction image Q _IJ (λ) is subtracted from the observation image O _IJ (λ) to calculate the stray light image B _IJ (λ) (step S69). Thereafter, a smoothing process is performed on the stray light image B _IJ (λ), and the stray light image B _ij (λ) including all original pixels is converted (step S70). Then, as shown in the above equation (18-1), this stray light image B _ij (λ) is subtracted from the observed image O _ij (λ) to obtain an approximate image R ′ _ij (λ) (step S71).

As described above, according to the stray light correction method of the present embodiment, an observation image (O _ij (λ)) on which a stray light image based on the influence of stray light is superimposed and, for example, the stray light correction calculation shown in FIG. Specific pixels (for example, pixel set P) in the imaging area (second image 6a or first image 2a) of the observation image, which is measured (actually measured) in advance by calibration using the calibration system 30a and stored in the main controller 9 or the like. _{From the} response image (for example, (F _MN ) _IJ (λ); response image) generated by the incident light of unit intensity on _MN ) on all the pixels in the imaging area of the image sensor 7, the observation image (O _ij (λ)) The stray light image (B _ij (λ)) is estimated and calculated using image information (for example, O _IJ (λ) or B _IJ (λ)) that is lower than the number of pixels constituting the, and this estimated and calculated stray light image observation field a _{(B ij (λ)) (O ij} (lambda)) by removing (subtracting) from the actual image _{(R ij} (lambda)) corrected image which approximates the (R _'ij (λ); images from the observation image obtained by correcting the influence of stray light) is Therefore, since the stray light correction process is performed in this way, a suitable observation image (measurement image) from which the influence of the stray light on the observation image is removed can be obtained. The calculation at the time of actual measurement is executed efficiently by using response image information, etc., and the calculation is performed by using the method of correcting the stray light using the image information of the number of pixels lower than the number of pixels constituting the observation image. It is possible to reduce the amount of data sometimes handled, while applying the calculation with the low number of pixels to the stray light image whose spatial frequency is much lower than the observed image (see FIG. 8). Even low Since the error from the case of not performing the arithmetic operation with a number is small, the calculation processing time (correction time) required for the stray light correction can be greatly shortened while maintaining sufficient accuracy. According to the method, since blur and fog due to the influence of stray light can be removed from the observed image, a great effect can be obtained particularly in improving the accuracy of low-luminance pixels in an image having contrast (this is because the stray light correction method is used). The same applies to the two-dimensional spectral luminance meter 1).

Further, as shown in FIG. 14, an observation image (O _ij (λ)) obtained by imaging is converted into a low pixel observation image (O _IJ ) that is an observation image having a low number of pixels with respect to the total number of pixels of the observation image. (Λ)), and the corrected image (R ′ _ij (λ)) is reduced from the low pixel observation image and the response image ((F _MN ) _IJ (λ)) with the low pixel count. The second step of calculating the low pixel correction image (Q _IJ (λ)) as the pixel number image and the difference between the low pixel observation image and the low pixel correction image (O _IJ (λ) −Q _IJ (λ) ) To calculate a stray light image (B _IJ (λ)) having a low pixel count, and a fourth step of converting the stray light image having the low pixel count into a stray light image (B _ij (λ)) for all the pixel counts. steps and stray light images from all pixels number of observation images _{(O ij (λ))} with respect to the total number of pixels _{(B ij (λ)} Since the stray light correction operation is performed by the fifth step of calculating the a by subtracting actual image corrected image _{(R 'ij} (λ)) which approximates _{(R ij} (λ)), significant amount of data to be handled during the operation It is possible to perform stray light correction on the observation image (O _ij (λ)) with sufficient accuracy while reducing the processing time and greatly reducing the calculation processing time.

Further, the low pixel corrected image (Q _IJ (λ)) as a temporary actual image with respect to the low pixel observation image (O _IJ (λ)) (or Q _MN (λ) corresponding to Q _IJ (λ); 14-1) reference) and the response image _{((F MN) IJ (λ} )) from the pseudo observed image (O _'IJ (λ)) is calculated, and the difference due to the difference between the pseudo observed image and the low pixel observation image The low pixel corrected image (Q _IJ (λ)) is re-corrected with the image (O ′ _IJ (λ) −O _IJ (λ)), the re-corrected low pixel corrected image (Q _IJ (λ)) and the response image And a new pseudo-observation image (O ′ _IJ (λ)) is calculated, and a new difference image (O ′ _IJ (λ) −O _IJ (λ) based on the difference between the new pseudo-observation image and the low-pixel observation image is calculated. )), The asymptotic operation for further correcting the low pixel correction image is repeated until the error due to the difference image becomes a predetermined value or less. By performing returns, the calculated low pixel correction image _{(Q IJ (λ)),} i.e., the low-pixel observation image _{(O IJ (λ)} and the low pixel number of responses image _{((F MN) IJ (λ} )) Therefore, a low pixel corrected image (Q _IJ (λ)) is calculated by an asymptotic method, and therefore, a low-pixel corrected image (Q _IJ (Q _IJ ( λ)) can be obtained, and as a result, the calculation time for the entire stray light correction can be further shortened and the calculation accuracy (the accuracy of the corrected image (R ′ _ij (λ))) can be improved.

Further, in the stray light correction calculation, an image with the total number of pixels (for example, an observation image (O _ij (λ)) or a stray light image (B _ij (λ))) is binned (see FIG. 9). A number of images (for example, an observation image (O _IJ (λ)) or a stray light image (B _IJ (λ)) having a low number of pixels) can be obtained.

Further, the response image ((F _MN ) _IJ (λ)) is not a point light source for each pixel (instead of each pixel (m, n) or (i, j)) in the imaging area of the observation image. Is practically difficult), and is based on an image obtained by imaging a light beam 36a from a point light source (white light source 36) with respect to each pixel set (each pixel set coordinate (M, N)). Therefore, a response image corresponding to each two-dimensional spectroluminometer can be acquired experimentally and easily.

  In addition, since the stray light correction processing is performed at the wavelength of incident light (according to the center wavelength λ of the spectrally transmitted light by the WBPF 12) in the imaging of the observation image, stray light correction is effectively performed even if the cause of stray light has wavelength dependency. Can be implemented.

Furthermore, according to the two-dimensional spectral luminance meter 1 of the present embodiment, an observation image (O _ij (λ)) on which a stray light image based on the influence of stray light is superimposed by the stray light correction calculation means (main control unit 9), For use in the stray light correction calculation, for example, an imaging area (second image 6a or second image 6a or the second image 6a) that is measured (measured) in advance by calibration or the like using the calibration system 30a shown in FIG. A response image (for example, (F _MN ) _IJ (λ); a response image generated by incident light of unit intensity on a specific pixel (for example, pixel set P _MN ) in the first image 2a) on all the pixels in the imaging area of the image sensor 7; ) from the observation image _{(O ij} (lambda)) image information less number of pixels than the number of pixels of the ((e.g. _{O iJ} (lambda) and _{B iJ} (lambda)) stray image using _{(B ij} (lambda)) Is estimated and this estimated The corrected image (R ′ _ij (λ) approximated to the real image (R _ij (λ)) is obtained by removing (subtracting) the stray light image (B _ij (λ)) from the observed image (O _ij (λ)). Therefore, since the stray light correction process is performed in this way, the influence of stray light on the observation image (measurement image) by the two-dimensional spectral luminance meter 1 is removed, and a suitable observation image can be obtained. The calculation at the time of actual measurement is efficiently performed by using response image information obtained by measurement in advance, and the stray light correction is performed using image information having a number of pixels lower than the number of pixels of the observation image. By doing so, it is possible to reduce the amount of data handled at the time of calculation, while applying the calculation with the low number of pixels to the stray light image whose spatial frequency is much lower than the observation image. Even with a low pixel count Since becomes error between case without calculation is small, the processing time required for the stray light correction while maintaining a sufficient accuracy (correction time) can be shortened significantly.

  In addition, this invention can take the following aspects.

  (A) In the above embodiment, since the spatial frequency of the stray light image is extremely lower than that of the observed image, the amount of data handled and the processing time are reduced to a practical level by performing binning. The present invention is not limited to the two-dimensional spectroluminance meter 1 using the scanning WBPF 10 described above, and can be applied to general devices that measure two-dimensional luminance and color from image information. That is, when the interference filter (WBPF 12) is used as in the two-dimensional spectroluminometer 1, the light flux component in the wavelength region that is not transmitted through the filter is reflected to generate stray light, so that the technique (stray light correction method) is applied. However, even in a device that does not use an interference filter, stray light is generated due to reflected light from the lens surface of the optical system or the image sensor surface that captures the measurement image, and this is superimposed on the measurement image (particularly high Since the accuracy of the image (contrast) is degraded, the effect of such stray light can be improved by the present technology.

  (B) In the above embodiment, by providing the objective optical system 2 and the relay optical system (relay lens 6 or the like), two image planes of the first image plane 2b and the second image plane 6b are provided, and each of these images is provided. Although the WBPF 12 (scanning WBPF 10) and the image sensor 7 are arranged on the surface and separated (separated) from each other, the image sensor 7 and the WBPF 12 are arranged with respect to a single image plane without the relay optical system. It is good also as a structure which arranges.

  (C) In the above embodiment, after the dark image is captured by the first light shielding unit 112 (s = 0), the filter holding plate 11 (WBPF 12) is moved in the positive direction of the x direction for each scanning step s. However, conversely, the second light-shielding portion 113 may be set to the scanning step s = 0 position to capture a dark image, and the filter holding plate 11 may be scanned and moved in the negative direction of the x direction.

(D) In the above embodiment, the response image ((F _MN ) _IJ (λ)) is obtained by actually measuring in advance, but is not limited thereto, and is not limited to this (it is not necessary to obtain by actual measurement). What has been obtained in advance by, for example, a guess calculation may be stored.

  (E) The shape of the WBPF 12 does not necessarily have to be a wedge shape in which the thickness changes linearly (in the scanning direction) (front and rear transmission surfaces are flat). For example, the thickness changes in a curve. The shape may be such that the transmission surface is curved.

  (F) In the above embodiment, scanning is performed by moving the wedge-shaped WBPF 12 along one direction. For example, an annular shape (or a non-annular shape) whose thickness (transmission wavelength) changes along the circumferential direction is used. For example, an arc-shaped BPF such as a semicircle may be used, and scanning may be performed by rotating the BPF with respect to the light beam La (optical path) (in a rotational scanning step).

It is a schematic structure figure showing an example of the two-dimensional spectral luminance meter concerning the present invention. It is a schematic block diagram which shows an example of the scanning WBPF in the two-dimensional spectral luminance meter shown in FIG. It is a perspective view which shows an example of WBPF. It is a conceptual diagram explaining the picked-up image corresponding to each scanning step in the imaging area of an image sensor. It is a conceptual diagram explaining the positional relationship between each part of the filter holding plate moved in the scanning direction (x direction) in each scanning step and the imaging area in the imaging device. It is a schematic block diagram which shows an example of the calibration system for wavelength axis calibration. It is a graph which shows an example of the relative spectral sensitivity of WBPF by wavelength axis calibration. It is a graph which shows an example of intensity distribution of real image _Rij ((lambda)), observation image _Oij ((lambda)), and stray light image _Bij ((lambda)). It is a conceptual diagram explaining the pixel set coordinate of the image by binning. It is a schematic block diagram which shows an example of the system for response image measurement. It is a flowchart which shows an example of the operation | movement regarding the measurement of the tristimulus value image by a two-dimensional spectral luminance meter. It is a flowchart which shows an example of the operation | movement regarding wavelength axis calibration. It is a flowchart which shows an example of the operation | movement regarding actual measurement of a response image. It is a flowchart which shows an example of an operation | movement in the case of performing a stray light correction at the time of the measurement by a two-dimensional spectral luminance meter.

DESCRIPTION OF SYMBOLS 1 2D spectral luminance meter 2a 1st image 6a 2nd image 7 Image pick-up element 8 Image signal processing part 9 Main control part (stray light correction calculating means)
10 Scanning WBPF
12 WBPF
13 Scanning drive unit 30, 30a Calibration system 31 Reference single wavelength light source 34 Two-dimensional spectral luminance meter to be calibrated 35 Reference spectral luminance meter 36 White light source (point light source)
112 1st light-shielding part 113 2nd light-shielding part 901 Pixel set L Two-dimensional light source La Light beam T Imaging area

Claims (6)

A stray light correction method for correcting the influence of stray light on an observation image obtained by imaging to obtain a corrected image,
The observation image on which a stray light image based on the influence of the stray light is superimposed;
From the response image obtained for all the pixels in the imaging area according to the incident light of unit intensity to the specific pixel in the imaging area of the observation image, which is measured and stored in advance,
Estimating and calculating a stray light image using image information having a number of pixels lower than the number of pixels constituting the observation image,
In the stray light correction method for calculating the corrected image that approximates an actual image by removing the estimated and calculated stray light image from the observation image ,
An observation image (O _ij (λ)) obtained by the imaging is converted into a low pixel observation image (O _IJ (λ)) that is an observation image having a low pixel count with respect to the total number of pixels of the observation image . 1 process,
From the low-pixel observation image and the low-pixel response image ((F _MN ) _IJ (λ)), the low-pixel correction image (Q _IJ ) as the low-pixel-number image of the correction image (R ′ _ij (λ)) A second step of calculating (λ));
A third step of calculating a stray light image (B _IJ (λ)) having a low number of pixels based on a difference (O _IJ (λ) −Q _IJ (λ)) between the low pixel observation image and the low pixel correction image ;
A fourth step of converting the stray light image of the low number of pixels into a stray light image (B _ij (λ)) for the total number of pixels ;
A corrected image (R ′ _ij (λ)) approximating the actual image is calculated by subtracting the stray light image (B _ij (λ)) for the total number of pixels from the observation image (O _ij (λ)) of the total number of pixels. stray light correction method, characterized by chromatic fifth step of, a. From the low pixel corrected image (Q _IJ (λ)) and the response image ((F _MN ) _IJ (λ)) as a provisional actual image with respect to the low pixel observation image (O _IJ (λ)), the pseudo observation image (O ' _IJ (λ)) is calculated, and the low pixel corrected image (Q _IJ (λ)) is calculated by the difference image (O ′ _IJ (λ) −O _IJ (λ)) due to the difference between the pseudo observation image and the low pixel observation image. ) Is re-corrected, a new pseudo-observation image (O ′ _IJ (λ)) is calculated from the re-corrected low pixel correction image (Q _IJ (λ)) and the response image, and the new pseudo-observation image is calculated. And an asymptotic operation for further correcting the low pixel correction image with a new difference image (O ′ _IJ (λ) −O _IJ (λ)) based on the difference between the low pixel observation image and the low pixel observation image, an error due to the difference image is less than a predetermined value by repeating the performing until the low pixel correction image in said second step (Q Stray light correction method according to claim 1, wherein the calculating the _{J (λ)).} 3. The pixel set image including a predetermined number of pixel sets obtained by binning the total number of pixels and grouping a predetermined number of pixels is set as the low pixel number image. Stray light correction method. 4. The stray light correction method according to claim 3, wherein the response image is obtained based on an image picked up by imaging light from a point light source on each pixel set in the imaging region of the observation image. 5. The stray light correction method according to claim 1 , wherein the calculation process of the stray light correction is performed for each wavelength of incident light in imaging of an observation image. A two-dimensional spectroluminometer that performs a stray light correction calculation to obtain a corrected image by correcting the influence of stray light on an observation image obtained by imaging,
The observation image on which a stray light image based on the influence of the stray light is superimposed;
From the response image obtained for all the pixels in the imaging area according to the incident light of unit intensity to the specific pixel in the imaging area of the observation image, which is measured and stored in advance,
Estimating and calculating a stray light image using image information having a number of pixels lower than the number of pixels of the observation image;
Stray light correction calculating means for calculating the corrected image that approximates the actual image by removing the estimated and calculated stray light image from the observation image ,
The stray light correction calculation means includes:
An observation image (O _ij (λ)) obtained by the imaging is converted into a low pixel observation image (O _IJ (λ)) that is an observation image having a low pixel number with respect to the total number of pixels of the observation image ;
From the low-pixel observation image and the low-pixel response image ((F _MN ) _IJ (λ)), the low-pixel correction image (Q _IJ ) as the low-pixel-number image of the correction image (R ′ _ij (λ)) (Λ))
A stray light image (B _IJ (λ)) having a low number of pixels is calculated by a difference (O _IJ (λ) −Q _IJ (λ)) between the low pixel observation image and the low pixel correction image ;
Converting the stray light image of the low number of pixels into a stray light image (B _ij (λ)) for the total number of pixels ;
A corrected image (R ′ _ij (λ)) approximating the actual image is calculated by subtracting the stray light image (B _ij (λ)) for the total number of pixels from the observation image (O _ij (λ)) of the total number of pixels. To
A two-dimensional spectral luminance meter.

Images