JP2006084401A - Image processing method, image processor, and image processing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To prepare faithful spectral image data in a spectral waveform reflecting an original bright line spectrum having a steep bright line while having a limited channel number. <P>SOLUTION: An image processing method includes: a light source ratio calculation step for finding a light source ratio expressing a ratio of a light amount component by a sub-light source included in the spectral image data, a sub-light source spectral extraction step for finding a first sub-light source spectral data corresponding to the light amount component of the sub-light source by multiplying the light source ratio to the spectral image data, a sub-light source spectral approximate step for finding a second sub-light source spectral data approximating to the first sub-light source spectral data from the spectral data to light of the sub-light source prepared previously, and a sub-light source spectral substitution step for removing the first light source spectral data from the spectral image data and adding the second sub-light source spectral data. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image processing method, an image processing apparatus, and an image processing program for forming spectral image data from a subject irradiated with at least part of light from a sub-light source having an emission line spectrum in addition to a main light source.

  In the shooting scene, as shown in FIG. 15, in addition to the light from the main light source 3, the light from the sub-light source 5 may be irradiated to one subject 1. In this case, the spectral reflectance from the part P1 of the subject 1 mainly irradiated by the main light source 3 and the part P2 irradiated by the sub-light source 5 are different. When shooting under such circumstances, depending on the shooting system, discoloration that cannot be seen with the naked eye may occur in the area P2 irradiated with the sub-light source. Thus, when designing a photographing system that reproduces a faithful image without discoloration using a simulation technique, it is possible to provide spectral image data that accurately reflects the spectral characteristics of the main light source 3 and the sub light source 5. The construction of is important.

  2. Description of the Related Art Conventionally, a multiband camera 7 that can obtain a plurality of spectral images obtained by dividing an imaging wavelength region into a plurality of channels and imaging the same subject for each channel has been used. The multiband camera 7 includes a filter having a plurality of spectral transmittance characteristics and a CCD camera. The reflected light from the subject 1 is passed through a spectral filter having a constant spectral transmittance characteristic that is different for each channel, and By forming an image on the light receiving surface, multi-channel image data is obtained, and a spectral image having spectral reflectance for each pixel can be obtained by calculation processing of the image data. Spectral images obtained using such a multiband camera are widely used for spectral sensitivity design such as digital still cameras (DSC), TV cameras, video cameras, and silver halide photography photosensitive materials. This type of multi-band camera is disclosed in Patent Document 1, for example.

JP 2001-99710 A

In consideration of the complexity of the apparatus and the load of image processing, the conventional multi-band camera uses an apparatus using about 6 to 16 types of filters (channels) as a practical machine. For this reason, there is a limit to sharp band (wavelength band) setting. On the other hand, if the number of channels is increased, the spectral resolution is improved, but the apparatus becomes more complicated and the load of image processing increases, making it difficult to put into practical use. On the other hand, with a limited number of filters, in the case of scene shooting that includes steep bright lines such as fluorescent lamps, the original spectral characteristics are stored as the amount of light averaged in the wavelength direction at the shooting stage. The emission line information is missing. That is, as shown in the spectral reflectance characteristics at P2 in FIG. 16, the reflectance at the part P2 from the sub-light source that should be steep in nature is low due to the low resolution of the individual filters. Such spectral characteristics have become dull.
In the spectral sensitivity design of an imaging system, a color correction method based on the spectral sensitivity of various light sources at the time of indoor shooting or night scene shooting is an important issue. For this purpose, when applying a spectrum image created by a multiband camera with a limited number of channels as described above, a steep spectral characteristic region is already missing for a light source having an emission line spectrum such as a fluorescent lamp. Therefore, the problem of large errors in simulation calculation was included.

  The present invention has been made in view of the above situation, and although the number of channels is limited, the original emission line having a steep emission line even when the light of the sub-light source having the emission line spectrum is partially included. An image processing method, an image processing apparatus, and an image processing program capable of creating spectral image data that reflects the spectrum (spectral characteristics of the sub-light source) and that is faithful to the spectral waveform are provided, and faithful images without discoloration can be reproduced. An object is to obtain image data suitable for spectral sensitivity design of an imaging system.

  In order to achieve the above object, the image processing method according to claim 1 of the present invention divides a photographing wavelength band into a plurality of band bands for a subject including at least part of light of a sub-light source having an emission line spectrum. An image processing method for forming spectral image data by synthesizing spectral data of each band band obtained, wherein a ratio of a light amount component by the sub-light source included in the spectral image data to the spectral image data A light source ratio calculating step for obtaining a light source ratio representing the above, a sub light source spectrum extracting step for obtaining first sub light source spectrum data corresponding to the light amount component of the sub light source by multiplying the spectrum image data by the light source ratio. In addition, the spectral data for the light from the secondary light source is approximated to the first secondary light source spectrum data. A sub-light source spectrum approximating step for obtaining the second sub-light source spectrum data, and a sub-light source spectrum replacing step for removing the first light source spectrum data from the spectral image data and adding the second sub-light source spectrum data. It is characterized by that.

  In this image processing method, when generating spectral image data, the light source ratio of the sub-light source is obtained, and first sub-light source spectrum data corresponding to the light amount component of the sub-light source is obtained from this light source ratio. Then, the second sub-light source spectrum data approximated to the first sub-light source spectrum data is obtained from the spectral spectrum data for the light of the sub-light source, the first light source spectrum data is removed from the spectral image data, and the second sub-light source spectrum is obtained. Data is added. Thereby, even when the light of the sub-light source having the bright line spectrum is partially included, it is possible to create spectral image data faithful to the spectral waveform reflecting the original bright line spectrum having the steep bright line.

  The image processing method according to claim 2, wherein the light source ratio calculating step receives light from a wide band obtained by adding one band band including a bright line spectrum and spectral characteristics of two to four band bands before and after the band band. It is characterized by obtaining.

  In this image processing method, a corrected image can be created from the amount of received light of one band band and a broadband band obtained by adding the spectral characteristics of two to four band bands before and after this band band. The light source ratio for is obtained.

  The image processing method according to claim 3, wherein one band including the bright line spectrum has a wavelength half-width of 40 nm or less, and a half-width of the band band synthesized as the broadband band is 1.5 times or more of the narrow wavelength band. It is characterized by being.

  In this image processing method, one band including the bright line spectrum has a wavelength half-value width of 40 nm or less, the resolution of spectral characteristics is increased, and the half-width of the band band synthesized as a broadband band is 1. By being 5 times or more, a light source ratio with less bias is required from a wide band.

  The image processing method according to claim 4 is characterized in that the light source ratio calculating step obtains the light source ratio from a light reception amount obtained by averaging spectral characteristics of a plurality of band bands.

  In this image processing method, an average (median) light source ratio is obtained from the average received light amount of spectral characteristics of a plurality of band bands, and the light source ratio is not biased.

  The image processing method according to claim 5, wherein the sub-light source spectrum approximating step obtains the second sub-light source spectrum data in which the integrated light quantity with respect to the wavelength coincides with the first sub-light source spectrum data.

  In this image processing method, the second sub-light source spectrum data is obtained with the same amount of integrated light as the first sub-light source spectrum data.

  The image processing method according to claim 6, wherein the sub-light source spectrum approximating step includes a sub-light source having a sub-light source spectrum that minimizes an RMS of a light amount difference for each wavelength (a value of a square root of an average value of the sum of squares of differences). It is characterized by selecting.

  In this image processing method, when the spectral characteristics of a plurality of types of sub-light sources are known, a sub-light source having a sub-light source spectrum that minimizes the RMS of the light amount difference for each wavelength is selected.

  The image processing apparatus according to claim 7 is an image processing apparatus that divides a photographing wavelength band into a plurality of band bands and shoots, and synthesizes spectral data of each obtained band band to form spectral image data, A light source ratio calculating means for obtaining a light source ratio representing a ratio of a light quantity component by the sub-light source included in the spectral image data to the spectral image data, and corresponding to the light quantity component of the sub-light source by multiplying the light source ratio by the spectral image data Second sub-light source spectrum data approximated to the first sub-light source spectrum data is obtained from the sub-light source spectrum extracting means for obtaining the first sub-light source spectrum data and the spectral spectrum data for the light of the sub-light source prepared in advance. Sub-light source spectrum approximating means, and the first light source spectrum data from the spectrum image data. To remove the, and having an auxiliary light source spectrum replacement means for adding the second auxiliary light sources spectral data.

  In this image processing apparatus, the imaging wavelength band is imaged by dividing into a plurality of band bands, and when the spectrum data of each band band is synthesized, the emission line spectrum of the sub-light source is used instead of the dull first sub-light source spectrum data. The restored second sub-light source spectrum data is used to reflect the original emission line spectrum having a steep emission line.

  An image processing program according to an eighth aspect causes a computer to execute the procedure of the image processing method according to any one of the first to sixth aspects.

  In this image processing program, the procedure of the image processing method described in any one of claims 1 to 6 is input to a computer as a source program, and the procedure is converted into an object program and executed. The

  According to the image processing method of the present invention, when generating spectral image data from an image photographed by a solid-state imaging device (multiband camera) having a limited number of channels (practically 6 to 16 channels) The first light source spectrum data corresponding to the light quantity component of the auxiliary light source is obtained by multiplying the light source ratio by the spectral image data, and the spectral data for the light of the auxiliary light source is approximated to the first auxiliary light source spectrum data. The second sub-light source spectrum data is obtained, the first light source spectrum data is removed from the spectral image data, and the second sub-light source spectrum data is added, so that the spectral waveform reflecting the original bright line spectrum having a steep bright line is obtained. Can be created. Therefore, by using the spectral image data obtained in this way for spectral sensitivity design of imaging systems and development of partial color fog correction processing methods, it is less affected by light sources such as fluorescent lamps that contain emission line spectra. The spectral sensitivity design of the imaging system can be simulated.

  According to the image processing apparatus of the present invention, a light source ratio calculating unit that obtains a light source ratio of a sub light source, a sub light source spectrum extracting unit that obtains first sub light source spectrum data corresponding to a light amount component of the sub light source, and a light source ratio calculating unit are prepared in advance. Sub-light source spectrum approximating means for obtaining second sub-light source spectrum data approximated to the first sub-light source spectrum data from the obtained spectral spectrum data, and removing the first light source spectrum data from the spectral image data and second sub-light source spectrum data Sub-light source spectrum replacement means for adding the first and second sub-light source spectrum data that have been dulled when the obtained spectral data of each band are synthesized by dividing the imaging wavelength band into a plurality of band bands. Instead, the second sub-light source spectrum data in which the emission line spectrum of the sub-light source is restored is used. And in, reflecting the original bright line spectrum having a sharp emission line, it is possible to create a faithful spectral image data in the spectral waveform.

  According to the image processing program of the present invention, since the computer executes the procedure of the image processing method according to any one of claims 1 to 6, the computer program can be executed without using another device. By defining the operation in the image processing procedure of the invention, spectral image data faithful to the spectral waveform can be created.

Preferred embodiments of an image processing method, an image processing apparatus, and an image processing program according to the present invention will be described below with reference to the drawings.
FIG. 1 is an explanatory view of an image processing method and an image processing apparatus according to the present invention. FIG. 2 is an explanatory view of spectral transmittance characteristics of 16 types of color separation filters used in the multiband camera shown in FIG. FIG. 3 is an explanatory diagram showing the configuration of the multiband camera and the one-dimensional LUT in (a), and one one-dimensional LUT in (b).
An image processing apparatus 100 shown in FIG. 1 is used for carrying out the image processing method according to the present embodiment. The image processing apparatus 100 photographs a subject 1 that is a photographing set irradiated with light from a main light source 3 such as sunlight, a strobe light, or an incandescent lamp, and a sub-light source 5 that is a fluorescent lamp.

  The image processing apparatus 100 is roughly divided into a multiband camera 21 and a control unit (computer) 23 having a CPU. The multiband camera 21 includes an imaging optical system such as a lens, an imaging device 20 such as a CCD, a control system for controlling these, a spectral filter 25 placed on the subject 1 side, and the like. In this embodiment, a turret type spectral filter 25 is used in which a plurality of interference filters 25a, 25b, 25c, 25d,... Are arranged in a circumferential direction on a disk rotated by a motor. FIG. 2 shows an example of the spectral transmittance of the filter used. Note that a liquid crystal tunable filter may be used in the multiband camera 21 instead of the spectral filter 25.

  Image data of a plurality of channels can be obtained by switching the wavelength of incident light by the spectral filter 25 and performing imaging (photographing). The image data of the plurality of channels are converted into relative light quantity values with reference to a one-dimensional lookup table (hereinafter referred to as “1DLUT”) prepared for each channel as shown in FIG. 3A. As shown in FIG. 3B, it is a table of correlation information between the center wavelength of the spectral filter and the relative light quantity.

  In the control unit 23, an image memory 27, an operation unit 29, an internal memory 31, and an image data output unit 33 that temporarily store the spectral image data from the multiband camera 21 and send it to the control unit 23 include signal signals. Connected to send and receive. Here, the processing operation of the control unit 23 is defined by the image processing program stored in the internal memory 31.

  The image processing program includes 81-channel spectral image generation means, light source ratio calculation means, sub-light source spectrum extraction means, sub-light source spectrum approximation means, and sub-light source spectrum replacement means. The light source ratio calculating means performs a calculation process for obtaining the light source ratio representing the ratio of the light quantity components by the sub light source 5 included in the image data for the image data of a plurality of channels. The sub-light source spectrum extraction unit performs a calculation process for obtaining first sub-light source spectrum data corresponding to the light amount component of the sub-light source 5 by multiplying the spectral image data by the light source ratio.

  The sub-light source spectrum approximating unit performs arithmetic processing for obtaining second sub-light source spectrum data approximated to the first sub-light source spectrum data from spectral spectrum data for the light of the prepared sub-light source. The sub-light source spectrum replacement unit performs a calculation process of removing the first light source spectrum data from the spectrum image data and adding the second sub-light source spectrum data. In the image processing apparatus 100, spectral image data reflecting the spectral characteristics of the sub-light source 5 is created by defining the processing operation of the control unit 23 by using an image processing program for the spectral image data obtained from the multiband camera 21. .

Next, an image processing method using the image processing apparatus 100 will be described.
FIG. 4 is a diagram showing an example of a photographing target, and FIGS. 5 to 6 are flowcharts showing the procedure of the image processing method according to the present invention.

First, prior to photographing, a calibration table for each channel of the multiband camera is created. Examples of the object to be photographed include those including a gray scale 35, a Macbeth chart 36, a reference white plate 37 made of barium sulfate, and the like as shown in FIG.
In order to calibrate the amount of light received by the CCD for each channel, a gray step as shown in FIG. 4 is photographed by illuminating only with the main light source 3, and the signal amount of each density step and the spectroradiometer (Topcon Corporation) A 1DLUT (FIG. 3B) is created between the relative light quantity with respect to the reference white plate at the center wavelength of the filter from the spectral energy value measured in SR-3).

Next, a desired shooting target is shot for each channel.
In the present embodiment, an accurate accuracy that is suitably used when considering spectral sensitivity design that reduces the influence of the secondary light source 5 having an emission line spectrum, such as a fluorescent lamp, or a color correction method for an undesirable color reproduction portion. An example of creating simple spectral image data will be described.
As a shooting scene for use for such a purpose, a part of the part illuminated by the main light source 3 (sunlight, strobe, incandescent lamp, etc.) is used with another sub-light source 5 (having a bright line spectrum such as a fluorescent lamp). Include illuminated parts.
This photographing object is photographed by each channel of the multiband camera 21 (step 11: hereinafter abbreviated as “Sll”).

Then, the obtained plurality of captured images are calibrated with 1DLUT for each channel (S12). A spectrum image of a total of 81 channels is created from the 16-channel captured image after calibration by an interpolation method at intervals of 5 mm from 380 nm to 780 nm. The spectral characteristic for one pixel is as shown in FIG. 7 (S13).
A spectral curve indicated by SP1 in FIG. 7 is a spectral characteristic corresponding to one pixel of a captured image. As described above (see paragraph [0033]), when the spectral curve obtained here is different from the spectral characteristic measured by the spectroradiometer, or different from the known spectral characteristic, each pixel The spectral curve may be corrected as follows. That is, by performing convolution or deconvolution processing (for example, see Japanese Patent Laid-Open No. 2001-352559) on the spectrum curve, the color of the saturation is independently adjusted by smoothing or sharpening the curvature of the spectrum curve. .

FIG. 8 shows a graph visually representing the correction process. FIG. 8A shows a spectrum curve Sp ₁ obtained by performing the convolution process on the original spectrum curve Sp ₀ , and the spectrum curve obtained by performing the deconvolution process after the curvature of the spectrum is smoothed and the saturation is lowered. sp ₂ is curvature sharpening of the spectrum is improved chroma.

Further, as shown in FIG. 8B, the hue is independently color-adjusted by translating (shifting) the spectral curve Sp ₀ in the wavelength direction. Here, the color adjustment of the hue independently means that the hue can be adjusted without substantially affecting the color attributes other than the hue.

Further, as shown in FIG. 8C, the brightness is independently color-adjusted by multiplying the spectrum curve Sp ₀ by a predetermined coefficient. This also makes it possible to perform color adjustment for lightness without substantially affecting color attributes other than lightness.

  The spectral image data color-adjusted in this way is independently color-adjusted without affecting the color attributes of brightness, hue, and saturation. Therefore, according to this calibration, it is possible to easily and accurately reproduce the color without losing the spectral characteristics of the original subject.

Next, the mixing ratio (main light source ratio) of the main light source and the sub light source in the image is measured.
FIG. 9 is an explanatory diagram for obtaining a correction band from a plurality of band bands, and FIG. 10 is a flowchart showing a procedure for obtaining a main light source ratio.
First, the light reception amount (G2) in the channel having spectral characteristics at the wavelength closest to the emission line wavelength (545 nm) of the fluorescent lamp shown in FIG. 9 and the spectral sensitivity of the three channels including the front and rear thereof are combined and received by the spectral sensitivity. An amount (G1) is obtained (S14).
Thus, by obtaining the received light amount from a wide band obtained by adding one band band including the emission line spectrum and spectral characteristics of two band bands before and after (or two to four bands including this band band), 1 The light source ratio for each pixel is obtained from the received light amount of one band band and a wide band obtained by adding the spectral characteristics of the band bands before and after this band band.

  In this case, it is preferable that one band band including the bright line spectrum has a wavelength half width of 40 nm or less, and a half band width of the band band synthesized as a broadband band is 1.5 times or more of the narrow wavelength band band. As a result, the resolution of the spectral characteristics is increased, and a light source ratio with less bias is required from the broadband band. Further, instead of the wavelength of 545 nm, the emission line wavelength of another fluorescent lamp (for example, 435 nm) may be used, or both may be used together.

  Here, step 15 for obtaining the main light source ratio for each pixel will be described in detail. After correcting the amount of light so that G1 = G2 when gray is photographed with only the main light source 3, it is determined that G2 / G1 of all the pixels is different from the light amount ratio of G2 / G1 at the time of photographing (S21). If G2 / G1 is the same for each pixel, the main light source ratio is set to 100% (S22).

  On the other hand, if G2 / G1 is different in any of the pixels, the minimum value of G2 / G1, for example, if this value is 0.6, this value is set as the reference value (= 1) (S23), A relative value f which is a ratio of the value of G2 / G1 to the reference value is obtained (S24). The relative value f is obtained by dividing the value of G2 / G1 by the minimum value of G2 / G1. The relative value at this time is shown in FIG. As shown in the figure, a pixel having the smallest G2 / G1 is set to a relative value 1, and a value equal to or greater than 1 is assigned to the other pixels.

  After obtaining the relative value f, using the following equation (1), the mixing ratio indicating the proportion of light from the fluorescent lamp among the light from the plurality of light sources reflected in the image based on the imaging signal obtained from each pixel x (estimated value) is calculated.

Mixing ratio x = (relative value f−relative value fmin)
/ (Relative value fmax−relative value fmin) (1)

Here, the relative value fmin is the minimum value of the relative value f in all pixels, and the relative value fmax is the maximum value of the relative value f in all pixels.
The mixing ratio refers to the ratio of the amount of light from a fluorescent lamp, which is a sub-light source, out of light incident on a certain pixel. A larger value of the mixing ratio x means that more light from the fluorescent lamp is included.

Formula (1) is devised based on the following idea.
Here, the correlation diagram showing the function of the mixture ratio and the relative value is shown in FIG.
As the mixing ratio increases, the ratio G2 / G1 also increases and the relative value f also increases. Therefore, the mixing ratio x can be regarded as a linear function 40 of the relative value f as shown in FIG. However, since the mixing ratio x is unknown, in order to obtain the expression of the function 40, the mixing ratio x takes the minimum value when the relative value f is any number, and the mixing ratio x is any number when the relative value f is any number. It is necessary to infer whether to take the maximum value. In S23, since the relative value f in each pixel is obtained, the pixel having the maximum relative value f has the maximum G2 / G1, and it can be estimated that the amount of light from the fluorescent lamp is the largest. When the amount of light from the fluorescent lamp is the largest, it can be considered that the mixing ratio x in the pixel is 1. On the other hand, the pixel having the smallest relative value f has the smallest G2 / G1, and it can be estimated that the amount of light from the fluorescent lamp is the smallest, and the amount of light from the fluorescent lamp is the smallest. This means that the mixing ratio x in the pixel can be regarded as zero. Based on this idea, the function 40 is created with the pixel having the fmin value as the mixing ratio x = 0 and the pixel having the fmax value as the mixing ratio x = 1. The function 40 is expressed by the following equation (2).

f = (fmax−fmin) x + fmin
= (Fmax-1) x + 1 (2)

  From the equation (2), the above equation (1) is obtained. Thereby, the main light source ratio can be obtained (S25). Here, the pixel having the value of fmax is regarded as the mixing ratio x = 1. However, in reality, in a situation where the mixing ratio x = 1, for example, a fluorescent lamp is included as the subject 1 in the image. This is the case. If the subject 1 does not have a fluorescent lamp, it is considered that there are not many pixels that actually have a mixture ratio x = 1. For this reason, the above formulas (1) and (2) may be created by regarding a pixel having the value of fmax as a value equal to or less than x = 1 instead of regarding the mixture ratio x = 1. For example, when the subject 1 has a fluorescent lamp before photographing, the user sets the maximum value of x to 1, and when the subject 1 has no fluorescent lamp, the user sets the maximum x The value can be set between 0.6 and 0.8. In this case, according to the set numerical value, for example, a linear function 41 shown in FIG. 11 may be created to calculate the mixture ratio x in each pixel. By doing so, it is possible to estimate the mixing ratio more accurately depending on the shooting scene. By obtaining the light source ratio from the amount of light received by averaging the spectral characteristics of a plurality of band bands, an average (median) light source ratio is obtained, and the light source ratio is not biased.

After obtaining the main light source ratio for each pixel as described above (S15), based on the following equation (3), each pixel of the spectral image (SP1) is multiplied by this main light source ratio and only the main light source 3 is used. A spectrum image (SP2) is calculated (S16).
Imaging signal value of main light source only = imaging signal value × (1-x) (3)
Next, the difference spectrum (ΔSP) shown in FIG. 7 which is the difference between SP1 and SP2 is obtained (S17).

  Then, the spectral characteristic spl (λ) of the sub-light source 5 is multiplied by a coefficient k so that each wavelength integrated value of the difference spectrum (ΔSP) matches each wavelength integrated value of the spectral characteristic of the known sub-light source 5, and FIG. SP3 shown is obtained (S18). FIG. 12 is a graph showing a result of multiplying the spectral characteristic spl (λ) of the sub-light source having a steep emission line by a coefficient k. That is, SP3 is obtained as light source spectrum data in which the integrated light quantity with respect to the wavelength coincides with the following formula (4).

  Next, as shown in FIG. 13, SP2 and SP3 are added to obtain a target spectrum image (SP4) (S19). FIG. 13 is a light quantity / wavelength correlation diagram showing spectral characteristics reflecting the original emission line spectrum having a steep emission line. That is, ΔSP (first light source spectrum data) is removed from the spectrum image data, and sub-light source spectrum replacement processing is performed in which SP3 (second sub-light source spectrum data) is added.

  At this time, even when the fluorescent lamp type is not known, the spectral characteristics spl (λ) of the candidate fluorescent lamp group is applied, and, for example, the emission line spectra of a plurality of light sources known in FIG. 14 are represented by (a) to (d). As shown in Fig. 5, the combined spectral characteristics are estimated by selecting the coefficient k and the fluorescent lamp type that minimize the RMS in the wavelength direction of the difference from the plurality of sub-light sources by the following equation (5). can do.

Next, a description will be given of an embodiment performed also as verification of the image processing method.
1) A set of evaluation scenes illuminated by the main illumination light source was assembled in the studio, and a shooting set was set so that the white fluorescent light was partially illuminated from one side of the scene, and was shot with a multiband camera (photo set) 1).

2) A gray step for light quantity correction LUT creation was installed at the center of the set, and a Macbeth chart shown in FIG. 4 was installed for spectral waveform estimation.

3) Next, for the verification of the method, only the white fluorescent lamp was turned off and photographed with a multiband camera (Photo Set 2).

4) The spectral luminance of the gray step, the barium sulfate reference white plate, and the Macbeth chart was measured from the same position as the multiband camera using a spectroradiometer (Topcon Corp. SR-3). From the measured value of the gray step, the relative luminance with respect to the reference white plate at the center wavelength of each channel of the multiband camera is obtained as a reference, and the output value of each step of the gray step is obtained from the image of each channel photographed in 2). A 1DLUT for light amount correction was created between the read and relative luminance.

5) After correcting the 16-channel image obtained by shooting the multi-band camera with the light amount correction 1DLUT created in 3), the spectrum image (SP1) shown in FIG. 7 of 81 channels with a half-wavelength interval of 5 nm from 380 to 780 nm is obtained. Created. At this time, the spectral waveform of 24 colors including gray of the Macbeth chart was corrected by shift in the wavelength direction and wavelength compression or expansion so that the RMS of the difference from the spectral waveform measured by the spectroradiometer was minimized.

6) From the image of 16 channels, as shown in FIG. 9, the image (image 1) of the channel including the bright line (545 nm) of the fluorescent lamp, and the integrated value image (image 2) of the total of the channel and the preceding and following channels. It was created. After creating a corrected image so that the output values of the reference white plate of image 1 and image 2 are the same, the main light source ratio was determined for each pixel.

7) A difference spectrum image (SP2) between the spectrum image data (SP1) of 4) and a spectrum obtained by multiplying the data by the main light source ratio was obtained. The difference spectrum and the amount of light (SP3) at which the second light source with the known spectral characteristics used in 1) has the minimum RMS in the wavelength direction were calculated. SP1 and SP3 were added to obtain the desired spectral image (SP4).

  According to the above-described image processing method, the light source ratio of the sub-light source 5 is obtained when the spectral image data is created from the image captured by the multiband camera 21 having a limited number of channels (practically 6 to 16 channels). The first sub-light source spectrum data corresponding to the light quantity component of the sub-light source 5 is obtained by multiplying the light source ratio by the spectrum image data, and the second sub-light source spectrum data is approximated to the first sub-light source spectrum data from the spectral spectrum data for the light of the sub-light source 5. Sub-light source spectrum data is obtained, the first light source spectrum data is removed from the spectral image data, and the second sub-light source spectrum data is added, so that it is faithful to the spectral waveform reflecting the original emission line spectrum having a steep emission line. Spectral image data can be created. The resulting spectral image data is used for spectral sensitivity design of imaging systems and development of partial color fog correction processing methods, so that the number of channels is limited, but the image is captured with higher resolution sensitivity characteristics. As a result, spectral image data equivalent to the above can be obtained. For example, the amount of light having a narrow wavelength component such as a fluorescent lamp can be measured, and a faithful image without discoloration can be reproduced.

  According to the image processing apparatus 100, a light source ratio calculating unit that calculates a light source ratio of the sub light source 5, a sub light source spectrum extracting unit that calculates first sub light source spectrum data corresponding to the light amount component of the sub light source, and a spectroscopic prepared in advance. Sub-light source spectrum approximation means for obtaining second sub-light source spectrum data approximated to the first sub-light source spectrum data from the spectrum data, and removing the first light source spectrum data from the spectrum image data and adding the second sub-light source spectrum data And sub-light source spectrum replacement means that divides the imaging wavelength band into a plurality of band bands, and in synthesizing the obtained spectrum data of each band band, it replaces the dull first sub-light source spectrum data. Thus, the second auxiliary light source spectrum data obtained by restoring the emission line spectrum of the auxiliary light source 5 is used. In, reflecting the original bright line spectrum having a sharp emission line, it is possible to create a faithful spectral image data in the spectral waveform.

  Further, according to the image processing program for executing the above processing on a computer, spectral image data faithful to the spectral waveform can be created by causing the control unit 23 which is a computer to execute the procedure of the image processing method. .

It is explanatory drawing of the image processing method and image processing apparatus which concern on this invention. It is explanatory drawing of the spectral transmittance characteristic of 16 types of spectral filters currently used for the multiband camera shown in FIG. It is explanatory drawing which showed the structure of the multiband camera and the one-dimensional LUT in (a), and one one-dimensional LUT in (b). It is a figure which shows an example of imaging | photography object. It is a flowchart (the 1) showing the procedure of the image processing method which concerns on this invention. It is a flowchart (the 2) showing the procedure of the image processing method which concerns on this invention. It is a wavelength-light quantity correlation diagram showing spectral characteristics for one pixel of spectrum image data of 81 channels after calibration. It is explanatory drawing which represented different correction processing (a)-(c) visually. It is explanatory drawing which obtains a correction band band from a plurality of band bands. It is a flowchart showing the procedure of the image processing method which concerns on this invention. It is a correlation diagram showing the function of a mixture ratio and a relative value. It is a wavelength and light quantity correlation diagram showing the spectral characteristic of a sub-light source. It is a wavelength / light quantity correlation diagram showing spectral characteristics reflecting an original emission line spectrum having a steep emission line. It is explanatory drawing which represented the bright line spectrum in the some sublight source known by (a)-(d). It is explanatory drawing of the imaging | photography scene using the conventional multiband camera. It is explanatory drawing showing the spectral reflectance of the to-be-photographed object site | part irradiated with a main light source and a sublight source.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Subject 3 Main light source 5 Sub light source 21 Multiband camera 20 Image sensor 23 Control part (computer)
25 Spectral Filter 27 Image Memory 31 Internal Memory 33 Image Data Output Unit 100 Image Processing Device SP1, SP2, SP3, SP4 Spectral Image Data t1 to t16 Band Band

Claims (8)

For a subject that includes at least part of the light from the sub-light source having an emission line spectrum, the imaging wavelength band is divided into a plurality of band bands, and the resulting spectral data of each band band is synthesized to obtain spectral image data. An image processing method for forming,
A light source ratio calculating step for obtaining a light source ratio representing a ratio of a light amount component by the sub-light source included in the spectral image data to the spectral image data;
A sub-light source spectrum extraction step of multiplying the spectral image data by the spectral image data to obtain first sub-light source spectrum data corresponding to a light amount component of the sub-light source;
A sub-light source spectrum approximation step for obtaining second sub-light source spectrum data approximated to the first sub-light source spectrum data from spectral spectrum data for the light of the sub-light source prepared in advance;
A sub-light source spectrum replacement step of removing the first light source spectrum data from the spectral image data and adding the second sub-light source spectrum data.   2. The light source ratio calculating step obtains a received light amount from a wide band obtained by adding one band band including an emission line spectrum and spectral characteristics of two to four band bands before and after the band band. The image processing method as described.   3. One band including an emission line spectrum has a wavelength half-width of 40 nm or less, and a half-width of a band band synthesized as the broadband band is 1.5 times or more of a narrow wavelength band. Image processing method.   The image processing method according to claim 2, wherein the light source ratio calculating step obtains the light source ratio from an amount of received light obtained by averaging spectral characteristics of a plurality of band bands.   The image processing method according to claim 1, wherein the sub-light source spectrum approximating step obtains the second sub-light source spectrum data having an integrated light amount corresponding to a wavelength with respect to the first sub-light source spectrum data.   2. The sub-light source spectrum approximating step selects a sub-light source having a sub-light source spectrum that minimizes an RMS of a light amount difference for each wavelength (a value of a square root of an average value of the sum of squares of differences). The image processing method as described. An image processing apparatus that divides and shoots an imaging wavelength band into a plurality of band bands, and synthesizes spectral data of each obtained band band to form spectral image data,
A light source ratio calculating means for obtaining a light source ratio representing a ratio of a light amount component by the sub-light source included in the spectral image data to the spectral image data;
Sub-light source spectrum extraction means for multiplying the spectrum image data by the light source ratio to obtain first sub-light source spectrum data corresponding to the light amount component of the sub-light source;
Sub-light source spectrum approximation means for obtaining second sub-light source spectrum data approximated to the first sub-light source spectrum data from spectral spectrum data for the light of the sub-light source prepared in advance;
An image processing apparatus comprising: sub-light source spectrum replacement means for removing the first light source spectrum data from the spectral image data and adding the second sub-light source spectrum data.   An image processing program for causing a computer to execute the procedure of the image processing method according to any one of claims 1 to 6.

Images