Image generation device, image generation method, and program

Abstract

An image generation device according to one aspect of the present invention is provided with: a determination unit (323) that, with respect to a first pixel of a hyperspectral image, determines an increase or decrease of a first optical spectral intensity of a first wavelength and a second optical spectral intensity of a second wavelength that is adjacent to the first wavelength; an addition unit (324) that, on the basis of the determination result by the determination unit (323), adds an addition value to any among a first parameter, a second parameter, and a third parameter; and a generation unit (325) that generates an RGB image in which the values of the first parameter, the second parameter, and the third parameter obtained via the addition by the addition unit (324) are the R, G, and B values of a pixel corresponding to the first pixel.

Description

Image generation device, image generation method, and program

[0001] One aspect of the present invention relates to an image generating device, an image generating method, and a program.

[0002] Unlike RGB cameras, which capture color information of a subject using the three parameters R, G, and B, hyperspectral cameras can capture hyperspectral images and videos that capture color information using dozens or even hundreds of parameters as spectral intensity information measured for each wavelength of light. For this reason, hyperspectral cameras are used, for example, on factory production lines to detect and distinguish two objects of the same color that are difficult for the human eye to distinguish using object recognition algorithms, allowing machines to automatically remove one of the two objects of the same color.

[0003] Non-Patent Document 1 describes a method for object recognition in which hyperspectral images are classified into classes on a pixel-by-pixel basis by using a Support Vector Machine (SVM), a regression model, and deep learning, respectively.

[0004] Kunihito Kato, "Application of Hyperspectral Images to General Object Recognition," Journal of the Japan Society for Precision Engineering, 2018, Vol. 84, No. 12, 958-962

[0005] However, all of the above classification methods require training of a classification model, which requires training data, labeling of the learning target, and training time.

[0006] Furthermore, classification using a model based on the learning results can require a large amount of calculation depending on the algorithm used, so processing hyperspectral data, which also has a dimension in the wavelength direction, is often not easily achieved.

[0007] Even when using Partial Least Squares (PLS) regression, which realizes data dimensional compression in the wavelength direction to reduce the processing load, learning is required to perform the PLS regression.

[0008] On the other hand, if you try to distinguish and visualize two different objects of the same color from the wavelength-specific optical spectra of hyperspectral data without prior learning, the optical spectra are significantly affected by shading, and it may not be possible to distinguish and visualize them properly.

[0009] This invention was made with the above-mentioned circumstances in mind, and provides a technology that can quickly distinguish and visualize two objects of different types but the same color that are difficult to distinguish between in a hyperspectral image or video, without prior learning, and without being affected by shading (robust to shading).

[0010] In order to solve the above problem, an image generating device of one aspect of the present invention includes: a determination unit that determines, for a first pixel of a hyperspectral image, an increase or decrease between a first optical spectrum intensity of a first wavelength and a second optical spectrum intensity of a second wavelength that is adjacent to the first wavelength; an addition unit that adds an additional value to any of a first parameter, a second parameter, and a third parameter based on the result of the determination by the determination unit; and a generation unit that generates an RGB image in which the values ​​of the first parameter, the second parameter, and the third parameter obtained by the addition by the addition unit are the R, G, and B values ​​of a pixel corresponding to the first pixel.

[0011] An image generating method according to one aspect of the present invention includes determining, for a first pixel of a hyperspectral image, an increase or decrease between a first optical spectrum intensity of a first wavelength and a second optical spectrum intensity of a second wavelength adjacent to the first wavelength; adding an additional value to any of a first parameter, a second parameter, and a third parameter based on a result of the determination; and generating an RGB image in which the values ​​of the first parameter, the second parameter, and the third parameter obtained by the addition are used as the R, G, and B values ​​of a pixel corresponding to the first pixel.

[0012] According to one aspect of the present invention, two objects of different types but the same color that are difficult to distinguish can be distinguished and visualized quickly, robust to shading, and without prior learning.

[0013] FIG. 1 is a block diagram showing an example of the configuration of an image generating device according to a first embodiment. FIG. 2 is a block diagram showing an example of the hardware configuration of the image generating device according to the first embodiment. FIG. 3 is a diagram showing an example of a wavelength-specific optical spectrum in an unshaded portion of two objects captured in a hyperspectral image. FIG. 4 is a diagram showing an example of a wavelength-specific optical spectrum in an unshaded portion and a shaded portion of an object captured in a hyperspectral image. FIG. 5 is a block diagram showing an example of the functional configuration of an image generating device according to the first embodiment. FIG. 6 is a diagram showing an example of a hyperspectral image used in the image generating device according to the first embodiment. FIG. 7 is a diagram showing an example in which the entire wavelength range of a hyperspectral image used in the image generating device according to the first embodiment is divided into three regions. FIG. 8 is a diagram showing an example of a user interface included in the image generating device according to the first embodiment. FIG. 9 is a flowchart showing an example of a noise removal process of the image generating device according to the first embodiment. FIG. 10 is a flowchart showing an example of an RGB image generation process of the image generating device according to the first embodiment. FIG. 11 is a diagram showing an example of an RGB image generated by the image generating device according to the first embodiment. FIG. 12 is a block diagram showing an example of the functional configuration of an image generating device according to a first modification of the first embodiment. Fig. 13 is a block diagram showing an example of the functional configuration of an image generation device according to a second modified example of the first embodiment. Fig. 14 is a block diagram showing an example of the functional configuration of an image generation device according to a third modified example of the first embodiment. Fig. 15 is a block diagram showing an example of the functional configuration of an image generation device according to a fourth modified example of the first embodiment. Fig. 16 is a block diagram showing an example of the functional configuration of an image generation device according to a fifth modified example of the first embodiment. Fig. 17 is a block diagram showing an example of the functional configuration of an image generation device according to the second embodiment.

[0014] Hereinafter, embodiments will be described with reference to the drawings. In the following description, components having substantially the same functions and configurations will be designated by the same reference numerals. When particularly distinguishing between elements having similar configurations, different letters or numbers may be added to the end of the same reference numerals.

[0015] 1. First Embodiment An image generating device according to a first embodiment will be described.

[0016] 1.1 Configuration 1.1.1 Configuration of Image Generation Device The configuration of the image generation device according to this embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing an example of the configuration of the image generation device.

[0017] The image generating device 1 is a device that determines, for each pixel of a hyperspectral image, whether the intensity of two spectral components with adjacent wavelengths increases or decreases, adds a value to one of the parameters R, G, and B based on the result of the determination, and generates an RGB image based on the values ​​of the parameters R, G, and B obtained by the addition. The parameters R, G, and B are prepared, for example, for each pixel of the hyperspectral image.

[0018] As shown in FIG. 1 , the image generating device 1 includes a processing unit 300 and a display unit 400 .

[0019] The processing unit 300 is, for example, a computer. The processing unit 300 receives a hyperspectral image from an external device (not shown) (for example, a hyperspectral camera). The hyperspectral image is, for example, an image acquired (photographed) by the hyperspectral camera or an image of one frame of video acquired by the hyperspectral camera.

[0020] Hereinafter, the hyperspectral image received by the processing unit 300 will be referred to as a "hyperspectral (HS) image IMh." Each pixel of the HS image IMh will be referred to as a "pixel PXh(x, y, s)" or a "pixel PXh." x is a scalar value of the X coordinate of the HS image IMh. y is a scalar value of the Y coordinate of the HS image IMh. s is a vector value composed of the spectral intensity for each wavelength. The vector value s is the spectral intensity s for each wavelength. １ , s ２ , s ３ , ..., s ｎ (n is an integer of 1 or more). Spectral intensity s １ ~s ｎ are arranged in order of shortest wavelength.

[0021] The processing unit 300 receives the additional value t and the threshold value vth from an external device (not shown) (e.g., a data server). The additional value t is a value to be added to the parameters R, G, and B. The threshold value vth is a threshold for determining whether the intensity of two optical spectra having adjacent wavelengths increases or decreases.

[0022] The processing unit 300 determines, for each pixel PXh of the HS image IMh, whether the intensity of two spectral components with adjacent wavelengths increases or decreases, based on the HS image IMh and the threshold value vth. For each pixel PXh, the processing unit 300 adds an additional value t to the value of one of the parameters R, G, and B based on the result of the determination. The processing unit 300 generates an RGB image in which the values ​​of the parameters R, G, and B obtained by the addition are the R, G, and B values ​​of the pixel corresponding to each pixel PXh. The processing unit 300 transmits the generated RGB image to the display unit 400.

[0023] Hereinafter, the RGB image generated by the processing unit 300 will be referred to as "RGB image IMrgb." Each pixel of the RGB image IMrgb will be referred to as "pixel PXrgb(x, y, R, G, B)" or "pixel PXrgb." x is the scalar value of the X coordinate of the RGB image IMrgb. y is the scalar value of the Y coordinate of the RGB image IMrgb. R is the R value of the RGB image IMrgb. G is the G value of the RGB image IMrgb. B is the B value of the RGB image IMrgb. The X and Y coordinates of the RGB image IMrgb correspond to the X and Y coordinates of the HS image IMh.

[0024] The display unit 400 is, for example, an RGB display, and displays the RGB image IMrgb received from the processing unit 300.

[0025] 1.1.2 Hardware Configuration of Image Generation Device The hardware configuration of the image generation device 1 will be described with reference to Fig. 2. Fig. 2 is a block diagram showing an example of the hardware configuration of the image generation device 1.

[0026] As shown in FIG. 2, the processing unit 300 includes a processor 301 , a memory 302 , a communication interface 303 , and a user interface 304 .

[0027] The processor 301 is, for example, a central processing unit (CPU), a micro processing unit (MPU), a graphics processing unit (GPU), or a field programmable gate array (FPGA). The processor 301 executes, for example, a process for generating an RGB image IMrgb. Hereinafter, the process for generating the RGB image IMrgb will be referred to as an "image generation process."

[0028] The image generation process includes noise removal processing and RGB image generation processing. The noise removal processing is processing for removing noise from the HS image IMh. The RGB image generation process is processing for generating an RGB image IMrgb based on the HS image IMh, the added value t, and the threshold value vth. The noise removal processing and the RGB image generation process will be described in detail later.

[0029] The memory 302 includes, for example, a read-only memory (ROM) and a random access memory (RAM). The ROM stores, for example, a program for causing the processor 301 to execute the image generation process. The RAM is used as a work area for the processor 301. The RAM temporarily stores, for example, the above-mentioned program executed by the processor 301 and data used when the image generation process is executed.

[0030] The communication interface 303 controls communication between the processing unit 300 and external devices and the display unit 400. The communication interface 303 receives the HS image IMh from the external device. The communication interface 303 receives the sum t and the threshold value vth from the external device. The communication interface 303 transmits the RGB image IMrgb obtained by the image generation process to the display unit 400.

[0031] The user interface 304 is a means operated by the user to input various information, and is, for example, a mouse, a keyboard, or a touch panel.

[0032] 1.1.3 Algorithm for RGB Image Generation Processing The algorithm for RGB image generation processing will be described.

[0033] FIG. 3 shows an example of the spectral distribution by wavelength in the unshaded areas of object 1 and object 2 captured in the HS image IMh. Hereinafter, the "unshaded areas" will also be referred to as "non-shaded areas." The vertical axis of FIG. 3 represents spectral intensity. The intensity shown on the vertical axis is, for example, a relative value based on the maximum value of the spectral distribution. The horizontal axis of FIG. 3 represents wavelength [nm]. Object 1 is, for example, an object of color 1 and material 1. Object 2 is, for example, an object of color 1 and material 2 (a material different from material 1). In other words, object 1 and object 2 are different objects of the same color. In FIG. 3, the solid line represents the spectral spectrum of object 1. The dashed line represents the spectral spectrum of object 2.

[0034] In this specification, "different" means different materials. For example, two human faces are different. "Same color" does not only mean that two objects have exactly the same color, but also means that the two objects are not exactly the same color, but have similar RGB values ​​when observed with a general RGB camera, i.e., are similar in color (similar colors).

[0035] As shown in FIG. 3, when two objects are of different types and the same color, the spectral intensities of the two objects at different wavelengths differ relatively greatly.

[0036] FIG. 4 is a diagram showing an example of the spectral distribution by wavelength in the non-shaded and shaded portions of object 1 captured in the HS image IMh. Hereinafter, the "shaded portion" will also be referred to as the "shaded portion." The vertical axis of FIG. 4 represents the spectral intensity. The intensity shown on the vertical axis is, for example, a relative value based on the maximum value of the spectral distribution. The horizontal axis of FIG. 4 represents the wavelength [nm]. The solid line represents the spectral distribution in the non-shaded portion of object 1. The dashed-dotted line represents the spectral distribution in the shaded portion of object 1. Object 1 is the same object as object 1 shown in FIG. 3.

[0037] As shown in Figure 4, the spectral intensity for each wavelength differs relatively greatly between the shaded and non-shaded areas in the HS image IMh. While Figure 4 shows one object, even for two objects of the same type and color, the spectral intensity for each wavelength differs relatively greatly between the shaded and non-shaded areas. Therefore, if you try to simply distinguish and visualize two objects of different types and the same color based on their spectral spectra, the effects of shading may make it difficult to distinguish and visualize them.

[0038] The inventors have discovered that focusing on the increase or decrease in intensity of two adjacent spectral wavelengths is an effective method for distinguishing and visualizing two objects of different types but the same color quickly, robustly against shadows, and without prior learning.

[0039] For example, as shown in Figure 4, when comparing the spectral intensity at a wavelength of 450 nm with the spectral intensity at the adjacent wavelength of 455 nm, the spectral intensity at 455 nm is greater than the spectral intensity at 450 nm, regardless of whether shading is present. The same is true for wavelengths 455 nm and 460 nm. In this way, the increase or decrease in the spectral intensity of two adjacent wavelengths is less affected by the presence or absence of shading than when the spectral intensity is directly observed.

[0040] An example of an algorithm for RGB image generation processing based on the above findings is the following Algorithm 1. Algorithm 1 is an algorithm that converts the vector value s of each pixel PXh(x, y, s) of the HS image IMh into an RGB value based on the increase or decrease in intensity of two spectral spectra with adjacent wavelengths, in order to quickly distinguish and visualize two objects of different types but the same color without prior learning and in a manner robust to shadows.

[0041] (First Algorithm) In this algorithm, first, for a pixel PXh(x, y, s) to be processed in the HS image IMh, parameters R, G, and B are each set to an initial value of 0. Hereinafter, the pixel to be processed is referred to as a "target pixel." Next, for the target pixel PXh(x, y, s) in the HS image IMh, two spectral intensities s whose wavelengths are adjacent to each other and whose vector value s isｍ－１ , s ｍ (m is an integer of 2 or more) and add an additional value t to the parameters R, G, and B based on the following condition 1. <Condition 1> s ｍ－１ -s ｍ If <-vth, add t to parameter R. ｍ－１ -s ｍ If > vth, add t to parameter G - vth ≦ s ｍ－１ -s ｍ If vth≦vth, add t to parameter B. Here, vth is a threshold value. The threshold value vth is, for example, a fixed value. t is an additional value. The additional value t is, for example, a predetermined constant.

[0042] The values ​​of the parameters R, G, and B obtained by applying condition 1 to the target pixel PXh(x, y, s) of the HS image are set as the R, G, and B values ​​of the pixel PXrgb(x, y, R, G, B) corresponding to the target pixel PXh(x, y, s) in the generated RGB image IMrgb.

[0043] Two spectral intensities s ｍ－１ , s ｍ The comparison of the values ​​of the parameters R, G, and B and the addition of the values ​​of the parameters R, G, and B are used to calculate the spectral intensity s ｍ－１ , s ｍ That is, for a target pixel PXh(x, y, s) in the HS image IMh, two spectral intensities s ｍ－１ , s ｍ The number of times that the comparison of values ​​for and the addition of values ​​of parameters R, G, and B are performed is (m-1).

[0044] By performing the above process on all pixels PXh(x, y, s) of the HS image IMh, an RGB image IMrgb is generated in which the values ​​of the parameters R, G, and B obtained by applying condition 1 are the R, G, and B values ​​of the pixel PXrgb(x, y, R, G, B) corresponding to each pixel PXh(x, y, s).

[0045] Furthermore, the inventors have found that changing the added value t for each wavelength region is effective in improving the accuracy of distinction and visualization. For example, in a region with a relatively short wavelength, when two adjacent spectral intensities s ｍ－１ , s ｍ A relatively small sum is applied to the two spectral intensities s ｍ－１ , s ｍ For example, a relatively large additional value is applied to

[0046] An example of an algorithm for RGB image generation processing based on the above knowledge is the following second algorithm. In order to improve the accuracy of distinguishing and visualizing two objects of different types but the same color, the second algorithm divides the entire wavelength range, which is the measurement range of the HS image IMh, into k parts (k is an integer of 2 or more), and calculates two spectral intensities s ｍ－１ , s ｍ This is an algorithm that changes the value of the added value t depending on which of the k wavelength regions the wavelength belongs to.

[0047] (Second Algorithm) In this algorithm, first, as in the first algorithm, for the target pixel PXh(x, y, s) of the HS image IMh, the parameters R, G, and B are each set to an initial value of 0. Next, for the target pixel PXh(x, y, s) of the HS image IMh, two spectral intensities s whose wavelengths are adjacent to each other and whose vector value s is ｍ－１ , s ｍ The values ​​of all the sets are compared, and the parameters R, G, and B are added with an additional value t ｋ Add. <Condition 2> s ｍ－１ is the wavelength region a ｋ That is, L ｋ，ｍｉｎ ≦L ｍ－１ ≦L ｋ，ｍａｘ In the case of s ｍ－１ -s ｍ <-vth, parameter R is set to t ｋ Add s ｍ－１ -s ｍ In the case of >vth, parameter G is set to t ｋAdd -vth≦s ｍ－１ -s ｍ If vth≦t, parameter B is set to t ｋ Add where L ｍ－１ is, s ｍ－１ is the wavelength of ｋ，ｍｉｎ is the divided wavelength region a ｋ The wavelength is on the short wavelength side of L ｋ，ｍａｘ is the divided wavelength region a ｋ The wavelength is the longer wavelength side of t ｋ is the wavelength region a ｋ is the sum of the above.

[0048] The values ​​of the parameters R, G, and B obtained by applying condition 2 to the target pixel PXh(x, y, s) of the HS image IMh are set as the R, G, and B values ​​of the corresponding pixel PXrgb(x, y, R, G, B) of the generated RGB image IMrgb, as in the first algorithm.

[0049] Two spectral intensities s ｍ－１ , s ｍ The comparison of the values ​​of the parameters R, G, and B and the addition of the values ​​of the parameters R, G, and B are performed in the same manner as the first algorithm, when the spectral intensity s ｍ－１ , s ｍ This is done for all pairs of

[0050] By performing the above process on all pixels PXh(x, y, s) of the HS image IMh, an RGB image IMrgb is generated in which the values ​​of the parameters R, G, and B obtained by applying condition 2 are the R, G, and B values ​​of the pixel PXrgb(x, y, R, G, B) corresponding to each pixel PXh(x, y, s).

[0051] Furthermore, the inventors have found that changing the threshold value vth for each wavelength region is effective in improving the accuracy of distinction and visualization. For example, in a region with a relatively short wavelength, when two adjacent spectral intensities s ｍ－１ , s ｍ A relatively small threshold is applied to the two spectral intensities s ｍ－１ , sｍ For example, a relatively large threshold is applied to

[0052] An example of an algorithm for RGB image generation processing based on the above knowledge is the following third algorithm. In order to improve the accuracy of distinguishing and visualizing two objects of different types but the same color, the third algorithm divides the entire wavelength range, which is the measurement range of the HS image IMh, into k parts, and calculates the spectral intensity s between two adjacent wavelengths. ｍ－１ , s ｍ is an algorithm that changes the value of the threshold vth depending on which of the k wavelength regions the wavelength belongs to.

[0053] (Third Algorithm) In this algorithm, first, as in the first algorithm, for the target pixel PXh(x, y, s) of the HS image IMh, the parameters R, G, and B are each set to an initial value of 0. Next, for the target pixel PXh(x, y, s) of the HS image IMh, two spectral intensities s whose wavelengths are adjacent to each other and whose vector value s is ｍ－１ , s ｍ The values ​​of all the pairs are compared, and an additional value t is added to the parameters R, G, and B based on the following condition 3. <Condition 3> s ｍ－１ is the wavelength region a ｋ That is, L ｋ，ｍｉｎ ≦L ｍ－１ ≦L ｋ，ｍａｘ In the case of s ｍ－１ -s ｍ <-vth ｋ In the case of , add t to the parameter R. ｍ－１ -s ｍ >vth ｋ In this case, add t to the parameter G. ｋ ≦s ｍ－１ -s ｍ ≦vth ｋ In this case, add t to parameter B. ｋ is the wavelength region a ｋ is the threshold value.

[0054] The values ​​of the parameters R, G, and B obtained by applying condition 3 to the target pixel PXh(x, y, s) of the HS image IMh are set as the R, G, and B values ​​of the corresponding pixel PXrgb(x, y, R, G, B) of the generated RGB image IMrgb, as in the first algorithm.

[0055] Two spectral intensities s ｍ－１ , s ｍ The comparison of the values ​​of the parameters R, G, and B and the addition of the values ​​of the parameters R, G, and B are performed in the same manner as the first algorithm, when the spectral intensity s ｍ－１ , s ｍ This is done for all pairs of

[0056] By performing the above process on all pixels PXh(x, y, s) of the HS image IMh, an RGB image IMrgb is generated in which the values ​​of the parameters R, G, and B obtained by applying condition 3 are the R, G, and B values ​​of the pixel PXrgb(x, y, R, G, B) corresponding to each pixel PXh(x, y, s).

[0057] In addition, the sum t of condition 2 and condition 3 is t ｋ By doing so, the second and third algorithms can be applied simultaneously. In other words, the second and third algorithms can be combined. Hereinafter, the algorithm that combines the second and third algorithms will be referred to as the "fourth algorithm."

[0058] 1.1.4 Noise Removal Processing Algorithm The noise removal processing algorithm will be explained.

[0059] When any of the first to fourth algorithms described above is applied to a state in which measurement noise is included in the HS image IMh, there is a possibility that the accuracy of distinction and visualization will decrease due to the influence of the measurement noise.

[0060] The inventors have discovered that performing spatial noise removal on the HS image IMh before applying any of the above-mentioned first to fourth algorithms to the HS image IMh is effective in improving the accuracy of distinction and visualization.

[0061] An example of a noise removal processing algorithm based on the above findings is the following fifth algorithm.

[0062] (Fifth Algorithm) This algorithm is an algorithm for removing noise in the spatial direction. In this algorithm, a certain wavelength L ｉ (i is an integer equal to or greater than 1) ｉ ) is considered. At this time, a certain wavelength L ｉ All pixels PXh(x, y, s ｉ ) can be regarded as one monochrome image. Hereinafter, this image will be referred to as a "monochrome image IMg." In addition, in the entire HS image IMh, each pixel PXh(x, y, s ｉ ) vector value s is n spectral intensities s in the wavelength direction １ , s ２ , s ３ , ..., s ｎ When the HS image IMh has the wavelength L １ , L ２ , L ３ , ..., L ｎ The image can be considered as n monochrome images IMg corresponding to the wavelengths. Therefore, in this algorithm, first, n monochrome images IMg for each wavelength are extracted. Then, a smoothing filter is applied to each of the extracted monochrome images IMg for each wavelength. The kernel size of the smoothing filter is, for example, 3x3 or 5x5. The smoothing filter to be applied may be, for example, an averaging filter, a Gaussian filter, a median filter, a maximum filter, or a minimum filter.

[0063] Furthermore, the inventors have discovered that performing wavelength-direction noise removal on the HS image IMh before applying any of the above-mentioned first to fourth algorithms to the HS image IMh is effective in improving the accuracy of distinction and visualization.

[0064] An example of a noise removal processing algorithm based on the above findings is the following sixth algorithm.

[0065] (Sixth Algorithm) This algorithm is an algorithm for removing noise in the wavelength direction. In this algorithm, the spectral intensity s of the vector value s of each pixel PXh(x, y, s) of the HS image IMh is calculated as １ , s ２ , s ３ , ..., s ｎ A smoothing filter is applied to the vector value s′ after smoothing. The smoothing filter to be applied may be, for example, an averaging filter, a median filter, a maximum value filter, or a minimum value filter. For example, the vector value s′ after smoothing is expressed as s′ １ , s' ２ , s' ３ , ..., s' ｎ and the window size is w (w is an odd number), the spectral intensity s ｉ It consists of (w-1) / 2 elements before and after Regarding s ｉ The average value of s' １ averaging filter, s ｉ The median of s' １ A median filter, s ｉ The maximum value of s' １ A maximum filter, s ｉ The minimum value of s' １ Examples include a minimum filter where

[0066] Furthermore, by applying the fifth algorithm and the sixth algorithm in any order, it is possible to perform noise removal processing using both the fifth algorithm and the sixth algorithm.

[0067] 1.1.5 Functional Configuration of Image Generation Device The functional configuration of the image generation device 1 will be described with reference to Fig. 5. Fig. 5 is a block diagram showing an example of the functional configuration of the image generation device 1.

[0068] 5, the processing unit 300 includes, as functional blocks, a noise removal unit 311 and an RGB image generation unit 321. The processor 301 of the processing unit 300 functions as the noise removal unit 311 and the RGB image generation unit 321. Note that in FIG. 5, functional blocks corresponding to the memory 302, the communication interface 303, and the user interface 304 of the processing unit 300 are omitted from the illustration.

[0069] In this embodiment, we will explain an example in which the above-mentioned fifth and sixth algorithms are applied sequentially to the HS image IMh shown in Figure 6, and then the entire wavelength range of the HS image IMh is divided into three regions and the above-mentioned fourth algorithm is applied.

[0070] Fig. 6 is a diagram showing an example of an HS image IMh. In Fig. 6, two faces are shown in the HS image IMh. Face 1 and face 2 have colors similar to the skin color of Asian people. Face 1 is darker than face 2. In other words, face 1 and face 2 are of different but the same color (similar colors).

[0071] 6 also shows in table form the vector value s of one pixel PXh(x, y, s) enclosed by a square in the HS image IMh. The number n of spectral intensities of the vector value s is 120. １ Wavelength L １ = 380 [nm], spectral intensity s ２ Wavelength L ２ = 385 [nm], spectral intensity s ３ Wavelength L ３ = 390 [nm], spectral intensity s ４ Wavelength L ４ = 395 [nm], ..., spectral intensity s １２０ Wavelength L １２０ = 975 [nm]. Spectral intensity s １ = 0.000244, spectral intensity s ２ = 0.000244, spectral intensity s ３ = 0.000244, spectral intensity s ４ = 0.001709, ..., spectral intensity s １２０ = 0.001221. Spectral intensity s１ ~s １２０ is, for example, a relative value based on the maximum value of the optical spectrum distribution.

[0072] 7 is a diagram showing an example in which the entire wavelength range of the HS image IMh is divided into three regions. In FIG. 7, the entire wavelength range of the HS image IMh is divided into a first wavelength region a １ , second wavelength region a ２ , and the third wavelength region a ３ The first wavelength region a is divided into three regions. １ has a range from 380 [nm] to 575 [nm]. That is, the first wavelength region a １ The wavelength L on the short wavelength side of １，ｍｉｎ is 380 [nm]. １ The wavelength L on the long wavelength side of １，ｍａｘ is 575 [nm]. ２ has a range from 575 [nm] to 775 [nm]. That is, the second wavelength region a ２ The wavelength L on the short wavelength side of ２，ｍｉｎ is 575 [nm]. ２ The wavelength L on the long wavelength side of ２，ｍａｘ is 775 [nm]. ３ The third wavelength range a has a range from 775 [nm] to 975 [nm]. ３ The wavelength L on the short wavelength side of ３，ｍｉｎ is 775 [nm]. ３ The wavelength L on the long wavelength side of ３，ｍａｘ is 975 [nm]. Hereinafter, the first wavelength region a １ is the "short wavelength region", and the second wavelength region a ２ is the "medium wavelength region", and the third wavelength region a ３ is also referred to as the "long wavelength region."

[0073] 8 is a diagram showing an example of the user interface 304 of the processing unit 300. The user can set the additional value t and the threshold value vth using a slide bar displayed on the user interface 304. For example, the user can set the additional value t of the short wavelength region by sliding the position of the slide bar for "additional value of short wavelength region" within the range from 0 to 3. １The value corresponding to the position of the slide bar can be set as the additional value t ２ , and the added value t ３ The user can also set the threshold value vth for the short wavelength region by sliding the slider bar for "threshold value for short wavelength region" between 0 and 30. １ The value corresponding to the position of the slide bar can be set as the threshold value vth of the medium wavelength region. ２ , and the threshold value vth in the long wavelength region ３ The set additional value t １ , t ２ , and t ３ , and threshold value vth １ , vth ２ , and vth ３ is stored in an external device, for example.

[0074] In FIG. 8, the position of the slide bar for "additional value of short wavelength region" is closer to 0 than the positions of the slide bar for "additional value of medium wavelength region" and the slide bar for "additional value of long wavelength region". The positions of the slide bar for "additional value of medium wavelength region" and the slide bar for "additional value of long wavelength region" are the same and are closer to 3 than the position of the slide bar for "additional value of short wavelength region". That is, ２ and t ３ is the added value t １ The additional value t ２ is the added value t ３ is the same as

[0075] Furthermore, the position of the slide bar for "short wavelength region threshold" is closer to 0 than the positions of the slide bar for "medium wavelength region threshold" and the slide bar for "long wavelength region threshold". The positions of the slide bar for "medium wavelength region threshold" and the slide bar for "long wavelength region threshold" are the same, and are closer to 30 than the position of the slide bar for "short wavelength region threshold". That is, the threshold vth ２ and vth ３ is the threshold value vth １ greater than the threshold value vth ２ is the threshold value vth ３ is the same as

[0076] Each functional block of the processing unit 300 will be described below.

[0077] (Noise Removal Unit 311) The noise removal unit 311 performs noise removal processing.

[0078] 5 , the noise removal unit 311 includes, as functional blocks, a first noise removal unit 312 and a second noise removal unit 313. The processor 301 of the processing unit 300 functions as the first noise removal unit 312 and the second noise removal unit 313.

[0079] The noise removal unit 311 receives the HS image IMh from an external device and transmits the received HS image IMh to the first noise removal unit 312.

[0080] (First noise removal unit 312) The first noise removal unit 312 receives the HS image IMh from the noise removal unit 311. The first noise removal unit 312 performs spatial noise removal on the HS image IMh by applying the fifth algorithm described above. That is, the first noise removal unit 312 removes noise by applying the smoothing filter described above to the monochrome image IMg for each wavelength of the HS image IMh. The first noise removal unit 312 transmits the HS image IMh1 obtained by the spatial noise removal to the second noise removal unit 313. Although the spectral intensity values ​​of each pixel PXh of the HS image IMh1 change from those of the HS image IMh due to the noise removal, the HS image IMh1 has the same data structure as the HS image IMh.

[0081] (Second noise removal unit 313) The second noise removal unit 313 receives the HS image IMh1 from the first noise removal unit 312. The second noise removal unit 313 performs wavelength direction noise removal on the HS image IMh1 by applying the sixth algorithm described above. That is, the second noise removal unit 313 removes noise by applying the smoothing filter described above to each pixel PXh of the HS image IMh. The second noise removal unit 313 transmits the HS image IMh2 obtained by the wavelength direction noise removal to the RGB image generation unit 321. Although the spectral intensity values ​​of each pixel PXh of the HS image IMh2 change from those of the HS image IMh1 due to the noise removal, the HS image IMh2 also has the same data structure as the HS image IMh.

[0082] (RGB Image Generator 321) The RGB image generator 321 executes an RGB image generation process.

[0083] 5 , the RGB image generation unit 321 includes, as functional blocks, an initialization unit 322, a determination unit 323, an addition unit 324, and a generation unit 325. The processor 301 of the processing unit 300 functions as the initialization unit 322, the determination unit 323, the addition unit 324, and the generation unit 325.

[0084] The RGB image generation unit 321 receives the HS image IMh2 from the second noise removal unit 313. The RGB image generation unit 321 transmits the received HS image IMh2 to the determination unit 323.

[0085] The RGB image generating unit 321 receives a threshold value vth (vth １ , vth ２ , and vth ３ ), and the sum t(t １ , t ２ , and t ３ The RGB image generation unit 321 transmits the received threshold value vth to the determination unit 323. The RGB image generation unit 321 transmits the received sum value t to the addition unit 324.

[0086] (Initializing Unit 322) The initializing unit 322 initializes the values ​​of the parameters R, G, and B to 0. The initializing unit 322 transmits the initialized parameters R, G, and B to the adding unit 324.

[0087] (Determination Unit 323) The determination unit 323 receives the HS image IMh2 and the threshold value vth from the RGB image generation unit 321. The determination unit 323 determines the wavelength L ｍ－１ The spectral intensity s ｍ－１ and wavelength L ｍ－１ and the adjacent wavelength L ｍ The spectral intensity s ｍ Determine the increase or decrease.

[0088] More specifically, the wavelength L ｍ－１ and wavelength L ｍ is the first wavelength region a １ If the threshold value vth is included in the threshold value Vth, the determination unit 323 determines the threshold value Vth as the threshold value Vth. １ The determination is made using the wavelength L ｍ－１ and wavelength L ｍ is the second wavelength region a ２ If the threshold value vth is included in the threshold value Vth, the determination unit 323 determines the threshold value Vth as the threshold value Vth. ２ The determination is made using the wavelength L ｍ－１ and wavelength L ｍ is the third wavelength region a ３ If the threshold value vth is included in the threshold value Vth, the determination unit 323 determines the threshold value Vth as the threshold value Vth. ３ The judgment is made using

[0089] The determination unit 323 determines the spectral intensity s ｍ－１ and the spectral intensity s ｍ and the threshold value vth ｋ By comparing the spectral intensity s ｍ－１ and the spectral intensity s ｍ Determine the increase or decrease with s ｍ－１ -s ｍ <-vth ｋ In this case, the determining unit 323 determines whether the wavelength L ｍ The spectral intensity s ｍ is the wavelength L ｍ－１ The spectral intensity s ｍ－１ In other words, it is judged as "increased." ｍ－１ -s ｍ >vth ｋIn this case, the determining unit 323 determines whether the wavelength L ｍ The spectral intensity s ｍ is the wavelength L ｍ－１ The spectral intensity s ｍ－１ -vth is determined to be decreased. ｋ ≦s ｍ－１ -s ｍ ≦vth ｋ In this case, the determining unit 323 determines whether the wavelength L ｍ The spectral intensity s ｍ is the wavelength L ｍ－１ The spectral intensity s ｍ－１ The difference between the two is relatively small, i.e., it is judged to be "flat."

[0090] In this specification, "determining whether the intensities of two spectral regions adjacent in wavelength are increased or decreased" means determining whether the intensities are "increasing," "decreasing," or "staying steady."

[0091] The determination unit 323 transmits the determination result to the addition unit 324. Hereinafter, the determination result will be referred to as a "determination result RES." The determination result RES includes, for example, "increase," "decrease," or "remain unchanged."

[0092] (Adder 324) The adder 324 receives the parameters R, G, and B from the initialization unit 322. The adder 324 receives the determination result RES from the determination unit 323. The adder 324 receives the addition value t from the RGB image generation unit 321. The adder 324 adds the addition value t to one of the parameters R, G, and B based on the determination result RES.

[0093] More specifically, the wavelength L ｍ－１ and wavelength L ｍ is the first wavelength region a １ If the value is included in the value of s, the judgment result RES is "increased", that is, s ｍ－１ -s ｍ <-vth ｋ In this case, the adder 324 adds the additional value t １ If the determination result RES is "decreased", that is, s ｍ－１ -s ｍ >vth ｋ In this case, the adder 324 adds the additional value t１ If the determination result RES is "flat", that is, -vth ｋ ≦s ｍ－１ -s ｍ ≦vth ｋ In this case, the adder 324 adds the additional value t １ In this way, the wavelength L ｍ－１ and wavelength L ｍ is the first wavelength region a １ If the sum is included in the sum, the adder 324 calculates the sum t １ Addition is performed using

[0094] Wavelength L ｍ－１ and wavelength L ｍ is the second wavelength region a ２ If the determination result RES is "increased", the adder 324 adds an additional value t ２ If the determination result RES is "decreased", the adder 324 adds the additional value t ２ If the determination result RES is "flat", the adder 324 adds the additional value t ２ In this way, the wavelength L ｍ－１ and wavelength L ｍ is the second wavelength region a ２ If the sum is included in the sum, the adder 324 calculates the sum t ２ Addition is performed using

[0095] Wavelength L ｍ－１ and wavelength L ｍ is the third wavelength region a ３ If the determination result RES is "increased", the adder 324 adds an additional value t ３ If the determination result RES is "decreased", the adder 324 adds the additional value t ３ If the determination result RES is "flat", the adder 324 adds the additional value t ３ In this way, the wavelength L ｍ－１ and wavelength L ｍ is the third wavelength region a ３ If the sum is included in the sum, the adder 324 calculates the sum t３ Addition is performed using

[0096] When the addition is completed for all pixels PXh of the HS image IMh2, the adder 324 transmits the parameters R, G, and B of all pixels PXh obtained by the addition to the generator 325.

[0097] (Generation unit 325) The generation unit 325 receives the parameters R, G, and B of all pixels PXh of the HS image IMh2 from the addition unit 324. For each pixel PXh, the generation unit 325 generates an RGB image IMrgb in which the values ​​of the received parameters R, G, and B are set as the R, G, and B values ​​of the pixel PXrgb corresponding to each pixel PXh of the HS image IMh2.

[0098] The generation unit 325 transmits the generated RGB image IMrgb to the display unit 400 .

[0099] 1.2 Noise Removal Processing The noise removal processing will be described with reference to Fig. 9. Fig. 9 is a flowchart showing an example of the noise removal processing of the image generating device 1.

[0100] First, the noise removal unit 311 receives an HS image IMh from an external device (S101), and transmits the received HS image IMh to the first noise removal unit 312.

[0101] Next, the first noise removal unit 312 performs spatial noise removal on the monochrome image IMg for each wavelength of the HS image IMh extracted from the HS image IMh as described above (S102). The first noise removal unit 312 transmits the HS image IMh1 obtained by the spatial noise removal to the second noise removal unit 313.

[0102] Next, upon receiving the HS image IMh1 from the first noise removal unit 312, the second noise removal unit 313 performs wavelength-direction noise removal on each pixel PXh of the HS image IMh1 as described above (S103).The second noise removal unit 313 transmits the HS image IMh2 obtained by the wavelength-direction noise removal to the RGB image generation unit 321 (S104).

[0103] After that, when the noise removal unit 311 receives the next HS image IMh from an external device (S101), steps S102 to S104 are carried out.

[0104] 1.3 RGB Image Generation Processing The RGB image generation processing will be described with reference to Fig. 10. Fig. 10 is a flowchart showing an example of the RGB image generation processing of the image generation device 1.

[0105] First, the RGB image generation unit 321 receives the HS image IMh2 from the second noise removal unit 313, and also receives the sum t and threshold vth from an external device (S201). The RGB image generation unit 321 transmits the HS image IMh2 and the threshold vth to the determination unit 323. The RGB image generation unit 321 transmits the sum t to the addition unit 324. The RGB image generation unit 321 also sets a target pixel PXh of the HS image IMh2.

[0106] Next, when the RGB image generation unit 321 sets the target pixel PXh of the HS image IMh2, the initialization unit 322 initializes the values ​​of the parameters R, G, and B to 0 (S202). The initialization unit 322 transmits the initialized parameters R, G, and B to the addition unit 324.

[0107] Next, upon receiving the HS image IMh2 and the threshold value vth from the RGB image generation unit 321, the determination unit 323 determines, as described above, whether the intensities of two spectral components with adjacent wavelengths increase or decrease for the target pixel PXh in the HS image IMh2 (S203). The determination unit 323 transmits the determination result RES to the addition unit 324.

[0108] Next, the addition unit 324 receives the addition value t from the RGB image generation unit 321 and the judgment result RES from the judgment unit 323, and adds the addition value t to one of the parameters R, G, and B based on the judgment result RES as described above (S204).

[0109] Next, the RGB image generating unit 321 determines whether steps S203 and S204 have been performed for all pairs of two optical spectrum intensities having adjacent wavelengths (S205).

[0110] If it is determined in step S205 that the process has not been performed (S205_No), step S203 is performed on the next pair.

[0111] On the other hand, if it is determined in step S205 that the steps have been performed (S205_Yes), the RGB image generating unit 321 determines whether steps S202 to S204 have been performed for all pixels PXh (S206).

[0112] If it is determined in step S206 that the pixel PXh has not been processed (S206_No), the RGB image generating unit 321 sets the target pixel PXh to the next pixel PXh. Then, step S202 is processed for the next pixel PXh.

[0113] On the other hand, if it is determined in step S206 that the addition has been performed (S206_Yes), the adder 324 transmits the R, G, and B parameters of all pixels PXh obtained by the addition to the generator 325. Thereafter, the generator 325 generates, as described above, an RGB image IMrgb in which the values ​​of the R, G, and B parameters obtained by the addition for each pixel PXh are the R, G, and B values ​​of the pixel PXrgb corresponding to the pixel PXh (S207).

[0114] Next, the generation unit 325 transmits the generated RGB image IMrgb to the display unit 400 (S208).

[0115] After that, when the RGB image generating unit 321 receives the next HS image IMh2 from an external device (S201), steps S202 to S208 are carried out.

[0116] 1.4 Effects of the Present Embodiment FIG. 11 is a diagram showing an example of an RGB image IMrgb generated by the image generating device 1 according to the present embodiment. As shown in FIG. 11 , face 1, which has a color similar to the skin color of yellow people and is darker than face 2, is displayed in black. On the other hand, face 2, which has a color similar to the skin color of yellow people and is lighter than face 1, is displayed in white. In other words, face 1 and face 2, which are of different but the same color (similar colors), are displayed in contrasting colors. Therefore, face 1 and face 2 can be distinguished and visualized by the RGB image IMrgb.

[0117] The fourth algorithm described above, which is applied to the image generating device 1 according to this embodiment, does not require prior learning. Furthermore, the fourth algorithm described above uses the increase and decrease in intensity of two adjacent wavelengths of the spectrum, making it possible to generate an RGB image that is robust to shading. Furthermore, the fourth algorithm described above is a relatively simple algorithm, making it possible to generate an RGB image quickly. Therefore, according to this embodiment, two objects of different types but the same color that are difficult to distinguish can be distinguished and visualized quickly, robust to shading, and without prior learning.

[0118] Furthermore, in the fourth algorithm described above, the entire wavelength range of the HS image IMh is divided into three regions, and the added value t and the threshold value vth are changed for each wavelength region, so that two objects of different types but the same color that are difficult to distinguish can be displayed in colors that are easier to distinguish. Thus, according to this embodiment, the accuracy of distinguishing and visualizing two objects of different types but the same color that are difficult to distinguish can be improved.

[0119] Furthermore, the fifth algorithm described above, which is applied to the image generating device 1 according to this embodiment, performs noise removal in the spatial direction, thereby enabling the removal of salt-and-pepper noise in the HS image IMh. Furthermore, the sixth algorithm described above, which is applied to the image generating device 1 according to this embodiment, performs noise removal in the wavelength direction, thereby smoothing out the jagged edges in the spectrum of each pixel PXh in the HS image IMh, making the peaks and valleys in the spectrum easier to observe. Therefore, this embodiment can improve the accuracy of distinguishing and visualizing two objects of different types but the same color that are difficult to distinguish.

[0120] 1.5 First Modification An image generating device according to a first modification of the first embodiment will be described. The image generating device 1A according to the first modification of the first embodiment differs from the first embodiment in that it performs RGB image generation processing by applying the second algorithm described above. The following description will focus on the differences from the first embodiment.

[0121] 1.5.1 Functional Configuration of Image Generation Device The functional configuration of the image generation device 1A will be described with reference to Fig. 12. Fig. 12 is a block diagram showing an example of the functional configuration of the image generation device 1A.

[0122] In this modification, the functional configuration other than the determining unit 323A is the same as the functional configuration of the first embodiment shown in FIG.

[0123] In this modification, the threshold value vth is １ , second wavelength region a ２ , and the third wavelength region a ３ The threshold value vth is constant regardless of, for example, the threshold value vth shown in the first embodiment. ２ and vth ３ is the same as

[0124] (Determination Unit 323A) The determination unit 323A receives the HS image IMh2 and the threshold value vth from the RGB image generation unit 321. The determination unit 323A determines the wavelength L ｍ－１ The spectral intensity s ｍ－１ and wavelength L ｍ－１ and the adjacent wavelength L ｍ The spectral intensity s ｍ Determine the increase or decrease.

[0125] More specifically, the wavelength L ｍ－１ and wavelength L ｍ is the first wavelength region a １ , second wavelength region a ２ , and the third wavelength region a ３ In either case, the determination unit 323A makes a determination using the threshold value Vth.

[0126] The determination unit 323A determines the spectral intensity s ｍ－１ and the spectral intensity s ｍ The spectral intensity s ｍ－１ and the spectral intensity s ｍ Determine the increase or decrease with s ｍ－１ -s ｍ If <-vth, the determining unit 323A determines that the value is "increased."ｍ－１ -s ｍ If -vth≦s, the determining unit 323A determines that the value is "decreased." ｍ－１ -s ｍ If vth≦vth, the determining unit 323A determines that the current is “horizontal”.

[0127] The determination unit 323A transmits the determination result RES to the addition unit 324.

[0128] 1.5.2 Effects of this Modification In this modification, two objects of different types but the same color that are difficult to distinguish can also be distinguished and visualized quickly, robustly against shading, and without prior learning, and the accuracy of distinguishing and visualizing two objects of different types but the same color that are difficult to distinguish can be improved.

[0129] 1.6 Second Modification An image generating device according to a second modification of the first embodiment will be described. The image generating device 1B according to the second modification of the first embodiment differs from the first embodiment in that it performs RGB image generation processing by applying the third algorithm described above. The following description will focus on the differences from the first embodiment.

[0130] 1.6.1 Functional Configuration of Image Generation Device The functional configuration of image generation device 1B will be described with reference to Fig. 13. Fig. 13 is a block diagram showing an example of the functional configuration of image generation device 1B.

[0131] In this modification, the functional configuration other than the addition unit 324B is the same as the functional configuration of the first embodiment shown in FIG.

[0132] In this modification, the additional value t is １ , second wavelength region a ２ , and the third wavelength region a ３ The additional value t is constant regardless of, for example, the additional value t shown in the first embodiment. ２ and t ３ is the same as

[0133] (Adder 324B) The adder 324B receives the parameters R, G, and B from the initialization unit 322. The adder 324B receives the determination result RES from the determination unit 323. The adder 324B receives the addition value t from the RGB image generation unit 321. The adder 324B adds the addition value t to one of the parameters R, G, and B based on the determination result RES.

[0134] More specifically, the wavelength L ｍ－１ and wavelength L ｍ is the first wavelength region a １ , second wavelength region a ２ , and the third wavelength region a ３ In any case where the result of the judgment RES is "increased", that is, s ｍ－１ -s ｍ <-vth ｋ In this case, the adder 324B adds the additional value t to the parameter R. If the determination result RES is "decreased", that is, s ｍ－１ -s ｍ >vth ｋ In this case, the adder 324B adds the additional value t to the parameter G. If the determination result RES is "flat", that is, -vth ｋ ≦s ｍ－１ -s ｍ ≦vth ｋ In this case, the adder 324B adds the additional value t to the parameter B. ｍ－１ and wavelength L ｍ is the first wavelength region a １ , second wavelength region a ２ , and the third wavelength region a ３ In either case, the addition unit 324B performs addition using the additional value t.

[0135] When the addition is completed for all pixels PXh of the HS image IMh2, the adder 324B transmits the parameters R, G, and B of all pixels PXh obtained by the addition to the generator 325.

[0136] 1.6.2 Effects of this Modification In this modification, two objects of different types but the same color that are difficult to distinguish can also be distinguished and visualized quickly, robustly against shading, and without prior learning, and the accuracy of distinguishing and visualizing two objects of different types but the same color that are difficult to distinguish can be improved.

[0137] 1.7 Third Modification An image generating device according to a third modification of the first embodiment will be described. The image generating device 1C according to the third modification of the first embodiment differs from the first embodiment in that it performs noise removal processing by applying only the fifth algorithm described above. The following description will focus on the differences from the first embodiment.

[0138] 1.7.1 Functional Configuration of Image Generation Device The functional configuration of image generation device 1C will be described with reference to Fig. 14. Fig. 14 is a block diagram showing an example of the functional configuration of image generation device 1C.

[0139] In this modification, the functional configuration is the same as that of the first embodiment shown in FIG. 5, except that the second noise removal unit 313 is eliminated from the noise removal unit 311.

[0140] 1.7.2 Noise Removal Processing In this modification, the flowchart of the noise removal processing is the same as the flowchart of FIG. 9 shown in the first embodiment, except that step S103 is omitted.

[0141] 1.7.3 Effects of this Modification In this modification, two objects of different types but the same color that are difficult to distinguish can also be distinguished and visualized quickly, robustly against shading, and without prior learning, and the accuracy of distinguishing and visualizing two objects of different types but the same color that are difficult to distinguish can be improved.

[0142] 1.8 Fourth Modification An image generating device according to a fourth modification of the first embodiment will be described. The image generating device 1D according to the fourth modification of the first embodiment differs from the first embodiment in that it performs noise removal processing by applying only the sixth algorithm described above. The following description will focus on the differences from the first embodiment.

[0143] 1.8.1 Functional Configuration of Image Generation Device The functional configuration of the image generation device 1D will be described with reference to Fig. 15. Fig. 15 is a block diagram showing an example of the functional configuration of the image generation device 1D.

[0144] In this modification, the functional configuration is the same as that of the first embodiment shown in FIG. 5, except that the first noise removal unit 312 is eliminated from the noise removal unit 311.

[0145] 1.8.2 Noise Removal Processing In this modification, the flowchart of the noise removal processing is the same as the flowchart of FIG. 9 shown in the first embodiment, except that step S102 is omitted.

[0146] 1.8.3 Effects of this Modification In this modification, two objects of different types but the same color that are difficult to distinguish can also be distinguished and visualized quickly, robustly against shading, and without prior learning, and the accuracy of distinguishing and visualizing two objects of different types but the same color that are difficult to distinguish can be improved.

[0147] 1.9 Fifth Modification An image generating device according to a fifth modification of the first embodiment will be described. The image generating device 1E according to the fifth modification of the first embodiment differs from the first embodiment in that it does not perform noise removal processing. The following description will focus on the differences from the first embodiment.

[0148] 1.9.1 Functional Configuration of Image Generation Device The functional configuration of the image generation device 1E will be described with reference to Fig. 16. Fig. 16 is a block diagram showing an example of the functional configuration of the image generation device 1E.

[0149] In this modification, the functional configuration is the same as that of the first embodiment shown in FIG. 5, except that the noise removal unit 311 is eliminated from the processing unit 300.

[0150] 1.9.2 Effects of this Modification In this modification, two objects of different types but the same color that are difficult to distinguish can also be distinguished and visualized quickly, robustly against shading, and without prior learning, and the accuracy of distinguishing and visualizing two objects of different types but the same color that are difficult to distinguish can be improved.

[0151] 2. Second Embodiment An image generating device according to a second embodiment will be described. The image generating device 1F according to the second embodiment differs from the first embodiment in that it performs RGB image generation processing by applying the first algorithm described above. The following description will focus on the differences from the first embodiment.

[0152] 2.1 Functional Configuration of Image Generation Device The functional configuration of the image generation device 1F will be described with reference to Fig. 17. Fig. 17 is a block diagram showing an example of the functional configuration of the image generation device 1F.

[0153] In this modification, the functional configuration other than the determining unit 323F and the adding unit 324F is the same as the functional configuration of the first embodiment shown in FIG.

[0154] In this modification, the entire wavelength region of the HS image IMh is not divided. Therefore, the threshold value vth and the added value t are constant. The threshold value vth is, for example, the threshold value vth shown in the first embodiment. ２ and vth ３ The additional value t is the same as the additional value t shown in the first embodiment, for example. ２ and t ３ is the same as

[0155] (Determination Unit 323F) The determination unit 323F receives the HS image IMh2 and the threshold value vth from the RGB image generation unit 321. The determination unit 323F determines the wavelength L ｍ－１ The spectral intensity s ｍ－１ and wavelength L ｍ－１ and the adjacent wavelength L ｍ The spectral intensity s ｍ Determine the increase or decrease.

[0156] More specifically, the determination unit 323F makes the determination using a threshold value Vth.

[0157] The determination unit 323F determines the spectral intensity s ｍ－１ and the spectral intensity s ｍ The spectral intensity s ｍ－１ and the spectral intensity s ｍDetermine the increase or decrease with s ｍ－１ -s ｍ If <-vth, the determining unit 323F determines that the value is "increased." ｍ－１ -s ｍ If -vth≦s, the determining unit 323F determines that the value is "decreased." ｍ－１ -s ｍ If vth or less, the determining unit 323F determines that the current is "horizontal."

[0158] The determination unit 323F transmits the determination result RES to the addition unit 324.

[0159] (Adder 324F) The adder 324F receives the parameters R, G, and B from the initialization unit 322. The adder 324F receives the determination result RES from the determination unit 323F. The adder 324F receives the addition value t from the RGB image generation unit 321. The adder 324F adds the addition value t to one of the parameters R, G, and B based on the determination result RES.

[0160] More specifically, if the determination result RES is "increased", that is, s ｍ－１ -s ｍ If <-vth, the adder 324F adds the additional value t to the parameter R. If the determination result RES is "decreased", that is, s ｍ－１ -s ｍ If the value of the determination result RES is "flat", that is, -vth≦s, the adder 324F adds the additional value t to the parameter G. ｍ－１ -s ｍ If vth≦vth, the adder 324F adds the additional value t to the parameter B. In this way, the adder 324F performs addition using the additional value t.

[0161] When the addition is completed for all pixels PXh of the HS image IMh2, the adder 324F transmits the parameters R, G, and B of all pixels PXh obtained by the addition to the generator 325.

[0162] 2.2 Effects of this embodiment In this embodiment, too, two objects of different types but the same color that are difficult to distinguish can be distinguished and visualized quickly, robustly against shading, and without prior learning, and the accuracy of distinguishing and visualizing two objects of different types but the same color that are difficult to distinguish can be improved.

[0163] 3. Modifications The above-described embodiments may be combined to the extent possible. For example, the second embodiment may be combined with the third, fourth, and fifth modifications of the first embodiment as modifications.

[0164] Furthermore, in the flowcharts described in the above embodiments, the order of the processes can be changed as much as possible.

[0165] It should be noted that the present invention is not limited to the above-described embodiments, and various modifications can be made in the implementation stage without departing from the spirit of the invention. Furthermore, the embodiments may be implemented in appropriate combinations, in which case the combined effects can be obtained. Furthermore, the above-described embodiments include various inventions, and various inventions can be extracted by selecting and combining the multiple disclosed constituent elements.

[0166] 1, 1A, 1B, 1C, 1D, 1E, 1F... Image generating device 300... Processing unit 301... Processor 302... Memory 303... Communication interface 304... User interface 311... Noise removal unit 312... First noise removal unit 313... Second noise removal unit 321... RGB image generating unit 322... Initialization unit 323, 323A, 323F... Determination unit 324, 324B, 324F... Addition unit 325... Generation unit 400... Display unit

Claims

1. An image generating device comprising: a determining unit that determines, for a first pixel of a hyperspectral image, an increase or decrease between a first spectral intensity of a first wavelength and a second spectral intensity of a second wavelength adjacent to the first wavelength; an adding unit that adds an additional value to one of a first parameter, a second parameter, and a third parameter based on the result of the determination by the determining unit; and a generating unit that generates an RGB image in which the values of the first parameter, the second parameter, and the third parameter obtained by the addition by the adding unit are the R, G, and B values of a pixel corresponding to the first pixel.

2. The image generating device according to claim 1, wherein the determination unit determines the increase or decrease by comparing the intensity difference between the first optical spectrum intensity and the second optical spectrum intensity with a threshold, and the addition unit adds the additional value to the first parameter when the intensity difference is smaller than a first value obtained by multiplying the threshold by -1, adds the additional value to the second parameter when the intensity difference is larger than the threshold, and adds the additional value to the third parameter when the intensity difference is equal to or larger than the first value and equal to or smaller than the threshold.

3. The image generating device according to claim 2, wherein when the first wavelength and the second wavelength are included in a first wavelength region ranging from a third wavelength to a fourth wavelength longer than the third wavelength, the adder performs the addition using a first sum as the sum, and when the first wavelength and the second wavelength are included in a second wavelength region ranging from the fourth wavelength to a fifth wavelength longer than the fourth wavelength, the adder performs the addition using a second sum larger than the first sum as the sum.

4. The image generating device according to claim 2, wherein when the first wavelength and the second wavelength are included in a first wavelength region ranging from a third wavelength to a fourth wavelength longer than the third wavelength, the judgment unit makes a judgment using a first threshold value as the threshold value, and when the first wavelength and the second wavelength are included in a second wavelength region ranging from the fourth wavelength to a fifth wavelength longer than the fourth wavelength, the judgment unit makes a judgment using a second threshold value larger than the first threshold value as the threshold value.

5. The image generating device of claim 2, wherein, when the first wavelength and the second wavelength are included in a first wavelength region ranging from a third wavelength to a fourth wavelength longer than the third wavelength, the determination unit makes a determination using a first threshold value as the threshold value, and the addition unit makes an addition using a first additional value as the additional value; and when the first wavelength and the second wavelength are included in a second wavelength region ranging from the fourth wavelength to a fifth wavelength longer than the fourth wavelength, the determination unit makes a determination using a second threshold value larger than the first threshold value as the threshold value, and the addition unit makes an addition using a second additional value larger than the first additional value as the additional value.

6. An image generating device as described in claim 1, further comprising a removal unit that removes noise from the hyperspectral image, wherein the removal unit: applies a first smoothing filter to a monochrome image for each wavelength of the hyperspectral image; or applies a second smoothing filter to each pixel of the hyperspectral image; or applies the first smoothing filter to the monochrome image for each wavelength of the hyperspectral image and also applies the second smoothing filter to each pixel of the hyperspectral image.

7. An image generation method comprising: determining, for a first pixel of a hyperspectral image, an increase or decrease between a first spectral intensity of a first wavelength and a second spectral intensity of a second wavelength adjacent to the first wavelength; adding an additional value to one of a first parameter, a second parameter, and a third parameter based on the result of the determination; and generating an RGB image in which the values of the first parameter, the second parameter, and the third parameter obtained by the addition are used as the R, G, and B values of a pixel corresponding to the first pixel.

8. A program that causes a processor included in the image generating device to execute the processing performed by the image generating device according to any one of claims 1 to 6.

Images