JP5094066B2 - Method and apparatus for operating image processing apparatus, and electronic endoscope system - Google Patents

Description

The present invention relates to an operation method of an image processing apparatus that generates a spectral image having an arbitrary wavelength band from an image acquired by an electronic endoscope, an apparatus, and an electronic endoscope system.

  Conventionally, medical diagnosis using an electronic endoscope has been actively performed in the medical field. A solid-state imaging device such as a CCD is built in the distal end of the insertion portion that is inserted into the subject of the electronic endoscope. By performing signal processing on the imaging signal acquired by the solid-state imaging device by the processor device, it is possible to observe the image of the observation site in the subject on the monitor.

  In the field of medical diagnosis using an electronic endoscope, in order to facilitate the discovery of lesions, light of a narrow wavelength band is irradiated to the site to be observed, and the reflected light is imaged (hereinafter referred to as this). A technique called Narrow Band Imaging (hereinafter abbreviated as NBI) that observes the obtained image by distinguishing it from a normal image and calling it a spectral image has attracted attention. According to NBI, an image in which blood vessels in the submucosa are emphasized, a stomach wall, an intestine without spraying a pigment on an observation site or injecting a contrast agent such as indocyanine green (ICG). It is possible to easily obtain an image in which an organ structure such as a surface layer tissue is emphasized.

  As an example of medical diagnosis using spectral images, a spectral image analysis method has been proposed that has a database of spectral data of lesions and identifies portions similar to the spectral data of lesions in spectral images of the fundus (patent) Reference 1).

Recently, linear approximation by dimensional compression using principal component analysis, Wiener estimation, etc. for normal images obtained by irradiating the observation site with normal white light instead of light in a narrow wavelength band A technique for obtaining a spectral image by performing spectral estimation processing represented by the above has been put into practical use. In this technique, enhancement processing is performed with a certain amount of enhancement on an image obtained as a result of spectral estimation processing.
JP 2005-296434 A

  However, there are individual differences in organ colors, and the performance of electronic endoscopes varies depending on the model and individual.Therefore, if the amount of emphasis is uniform, depending on the type of patient and electronic endoscope, The resulting image may be inappropriate for diagnosis. This problem can be solved temporarily by applying the invention described in Patent Document 1 and creating a database of enhancement amounts according to the type of patient and electronic endoscope. In addition, organ color measurement and electronic endoscope performance testing must be performed, and the amount of data in the database may become enormous.

The present invention has been made in view of the above problems, and an operation method and apparatus for an image processing apparatus capable of always obtaining an appropriate spectral image regardless of the type of patient or electronic endoscope, as well as an electronic device. An object is to provide an endoscope system.

  In order to achieve the above object, the invention described in claim 1 performs spectral estimation processing on an image acquired by an electronic endoscope, and sets an arbitrary wavelength band based on the spectral estimation image obtained thereby. An image processing method for generating a spectral image having a representative spectral characteristic representative of an overall hue of the spectral estimated image, and using the calculated representative spectral characteristic for each pixel of the spectral estimated image It is characterized by performing an emphasis process. Note that “representing the overall hue of the spectrally estimated image” is synonymous with representing the hue of a normal part occupying most of the image other than the lesioned part.

  It is preferable that an average of spectral characteristics of all pixels of the spectral estimation image is obtained and used as the representative spectral characteristics. In this case, the enhancement process is preferably performed according to a difference between the spectral characteristic of each pixel of the spectral estimation image and the representative spectral characteristic.

  Alternatively, it is preferable that a principal component analysis is performed on the spectral estimation image, and the first principal component obtained thereby is used as the representative spectral characteristic. In this case, the enhancement processing is preferably performed according to the n-th principal component (n is a natural number of 2 or more) obtained by the principal component analysis, and it is preferable that n = 2.

  Furthermore, it is preferable that cluster analysis is performed using spectral characteristics of all pixels of the spectral estimation image as vectors, and the maximum cluster is set as the representative spectral characteristics. In this case, the enhancement processing is preferably performed according to the distance between the spectral characteristics of each pixel of the spectral estimation image and the center of gravity of the maximum cluster.

  It is preferable that the enhancement amount of the enhancement processing is adjusted so that the intensity of the spectral characteristics of each pixel of the spectral estimation image falls within a predetermined range.

  It is preferable that spectrum images at arbitrary three wavelengths of the spectral estimation image after the enhancement process are assigned to preset three channels of RGB.

  According to an eleventh aspect of the present invention, image processing for performing spectral estimation processing on an image acquired by an electronic endoscope and generating a spectral image having an arbitrary wavelength band based on the spectral estimation image obtained thereby. A device that calculates a representative spectral characteristic representative of the overall hue of the spectral estimation image, and performs enhancement processing on each pixel of the spectral estimation image using the calculated representative spectral characteristic And a processing means. In this case as well, as in claim 1, “representing the overall hue of the spectrally estimated image” is synonymous with representing the hue of a normal part occupying most of the image other than the lesioned part. .

  It is preferable that the calculation unit obtains an average of spectral characteristics of all pixels of the spectral estimation image as the representative spectral characteristics. In this case, it is preferable that the enhancement processing unit performs the enhancement process according to a difference between the spectral characteristic of each pixel of the spectral estimation image and the representative spectral characteristic.

  Alternatively, the calculation unit preferably performs principal component analysis on the spectral estimation image, and the first principal component obtained thereby is used as the representative spectral characteristic. In this case, the enhancement processing means preferably performs the enhancement processing according to the n-th principal component (n is a natural number of 2 or more) obtained by the principal component analysis, and it is preferable that n = 2.

  Furthermore, it is preferable that the calculation unit performs cluster analysis using the spectral characteristics of all pixels of the spectral estimation image as a vector, and sets the maximum cluster as the representative spectral characteristic. In this case, it is preferable that the enhancement processing unit performs the enhancement processing according to the distance between the spectral characteristic of each pixel of the spectral estimation image and the center of gravity of the maximum cluster.

  It is preferable to include an enhancement amount adjusting unit that adjusts the enhancement amount of the enhancement processing so that the intensity of the spectral characteristics of each pixel of the spectral estimation image falls within a predetermined range.

  It is preferable that the apparatus further comprises channel allocating means for allocating spectral images at arbitrary three wavelengths of the spectral estimation image after the enhancement processing to preset three channels of RGB.

  A twenty-first aspect of the present invention is an electronic endoscope system including the image processing apparatus according to any of the eleventh to twentieth aspects.

According to the operation method, apparatus, and electronic endoscope system of the image processing apparatus of the present invention, the representative spectral characteristic representing the overall hue of the spectrally estimated image is calculated by the calculating means, and the calculated representative spectral characteristic is used. Thus, since the enhancement processing means performs enhancement processing on each pixel of the spectral estimation image, it is possible to always obtain an appropriate spectral image regardless of the type of patient or electronic endoscope.

  In FIG. 1, the electronic endoscope system 2 includes an electronic endoscope 10 and a processor device 11. The electronic endoscope 10 includes an insertion portion 12 that is inserted into a subject and an operation portion 13 that is connected to a proximal end portion of the insertion portion 12.

  The distal end portion 12a connected to the distal end of the insertion portion 12 is provided with an objective optical system 20 for capturing image light of an observation site in the subject, a CCD 21 for imaging the image light (both see FIG. 2), and an observation target. An illumination window 22 (see FIG. 2) for irradiating illumination light from the illumination device 24 (see FIG. 2) is built in the part. Also, cleaning water and air are sprayed onto the distal end portion 12a to clean the observation window protecting the objective optical system 20 by operating the forceps outlet communicating with the forceps port 14 and the air / water supply button 13a. A nozzle to be used is provided. The image in the subject acquired by the CCD 21 is displayed on the monitor 16 of the processor device 11 connected by the code 15.

  A bending portion 12b connecting a plurality of bending pieces is provided behind the tip portion 12a. The bending portion 12b is bent in the vertical and horizontal directions when the angle knob 13b provided in the operation portion 13 is operated and the wire inserted in the insertion portion 12 is pushed and pulled. Thereby, the front-end | tip part 12a is orient | assigned to the desired direction in a subject.

  A flexible portion 12c having flexibility is provided behind the curved portion 12b. In order to keep the distal end portion 12a reach the site to be observed and to keep the distance from the patient to the extent that the operator does not hinder the operation portion 13 when operating the flexible portion 12c, It has a length of several meters.

  In FIG. 2, one end of a light guide 23 is attached to the illumination window 22. The light guide 23 is provided over the insertion portion 12, and the other end is connected to a lighting device 24 provided in the operation portion 13. The illuminating device 24 has a light source such as a halogen lamp or a white LED, and light emitted from the light source irradiates the site to be observed from the illumination window 22 through the light guide 23.

  A driver 26 controlled by the CPU 25 is connected to the CCD 21. The driver 26 outputs to the CCD 21 a timing signal (clock pulse) for controlling the charge accumulation time and charge discharge timing of the CCD 21.

  The imaging signal output from the CCD 21 is input and stored in the image memory 27 of the processor device 11. A normal image generation unit 28 and a spectral estimation processing unit 29 are connected to the image memory 27.

  The normal image generation unit 28 performs analog signal processing such as correlated double sampling, amplification, and A / D conversion on the imaging signal read from the image memory 27, and then performs gradation correction, contour enhancement, γ correction, and the like. The normal signal is generated by performing digital signal processing. The normal image generation unit 28 outputs the generated normal image to the display control unit 30.

  The spectral estimation processing unit 29 reads an imaging signal from the image memory 27, and performs spectral estimation processing by linear approximation by dimensional compression using principal component analysis, Wiener estimation, or the like, using the read imaging signal. Specifically, the spectral estimation process estimates the spectral reflectance of the site to be observed in the visible light wavelength band (wavelength 400 nm to 700 nm) for each pixel of the CCD 21 from the imaging signal. The spectral estimation processing unit 29 outputs a spectral estimation image generated as a result of the spectral estimation processing to the spectral image generation unit 31.

  The spectral image generation unit 31 generates a spectral image having an arbitrary wavelength band based on the spectral estimation image from the spectral estimation processing unit 29. As illustrated in FIG. 3, the spectral image generation unit 31 includes a calculation circuit 40, an enhancement processing circuit 41, an enhancement amount adjustment circuit 42, and a channel assignment circuit 43.

  As shown in the upper and middle stages of FIG. 4, the calculation circuit 40 obtains the average of the spectral characteristics of all the pixels of the spectral estimation image. The average of the spectral characteristics obtained by the calculation circuit 40 is a spectral characteristic (hereinafter referred to as a representative spectral characteristic) Sr that represents the overall hue of the spectrally estimated image. In other words, the representative spectral characteristic Sr represents the hue of a normal part that occupies most of the image other than the lesioned part.

  As shown in the lower part of FIG. 4, the enhancement processing circuit 41 applies each pixel of the spectral estimation image according to the difference between the spectral characteristic of each pixel of the spectral estimation image and the representative spectral characteristic Sr obtained by the calculation circuit 40. Emphasis processing is performed on the image.

  Specifically, for example, when the difference from the representative spectral characteristic Sr is large, such as the spectral characteristic Sa indicated by the dotted line in the upper part and the spectral characteristic Sc indicated by the two-dot chain line, the enhancement amount is increased, and the spectral characteristic indicated in the lower part is increased. The characteristics are Sa ′ and Sc ′. On the other hand, when the difference from the representative spectral characteristic Sr is small like the spectral characteristic Sb indicated by the one-dot chain line in the upper stage, the enhancement amount is reduced to the spectral characteristic Sb ′ shown in the lower stage.

  When the difference is positive as in the spectral characteristic Sb, the positive enhancement amount is used. When the difference is negative as in the spectral characteristic Sc, the negative enhancement amount is used. As a result, the enhanced spectral estimated image has a portion with a hue different from the overall hue more emphasized than a portion with a hue substantially the same as the overall hue. Here, for the sake of convenience, the spectral characteristics of the spectral estimation image are drawn three by three, Sa to Sc and Sa ′ to Sc ′, but in reality, the spectral characteristics of all the pixels of the spectral estimation image are representative. Spectral characteristics are calculated and enhanced.

  Here, it is needless to say that the intensity of the spectral characteristic represents the intensity of light. In the image processing, the intensity of the light is assigned to an appropriate gradation value (for example, 0 to 255) to obtain image data. Is generated. Therefore, when the intensity of the spectral characteristic of each pixel of the spectral estimation image after the enhancement process cannot be expressed by a preset gradation value, that is, like the spectral characteristics Sd and Se shown in the upper part of FIG. If it does not fall within the specified range of 0 to It, it is necessary to adjust the amount of enhancement during the enhancement process. Therefore, as shown in the lower part of FIG. 5, the intensity adjustment circuit 42 predefines the intensity of the spectral characteristic of each pixel of the spectral estimation image after the enhancement process, like the spectral characteristics Sd ′ and Se ′. The enhancement amount of the enhancement process is adjusted so that it falls within the range of 0 to It.

  As shown in FIG. 6, the channel allocation circuit 43 allocates spectral images at arbitrary three wavelengths (for example, wavelengths of 500 nm, 450 nm, and 400 nm) of the spectral estimation image after the enhancement processing to preset three channels of RGB. . As a result, the intensity of the spectral characteristic of each pixel of the spectral estimation image at a certain wavelength is represented by one of RGB gradation values, and this is output to the display control unit 30 as a spectral image. The three wavelengths to be assigned to the RGB channels are set in advance or are individually changed by operating the console 32 (see FIG. 2).

  As shown in FIG. 7, the display control unit 30 displays the normal image 51 output from the normal image generation unit 28 and the spectral image 52 output from the spectral image generation unit 31 together with the examination date and patient information 50. Displayed side by side on the monitor 16. Further, by selecting on the console 32, the normal image 51 or the spectral image 52 can be individually displayed on the monitor 16.

  Returning to FIG. 2, the CPU 25 comprehensively controls the overall operation of the processor device 11. An operation console 32 for performing various settings and operations is connected to the CPU 25. The CPU 25 operates each unit according to an operation input signal input from the console 32.

  When observing the inside of the subject with the electronic endoscope system 2 configured as described above, the electronic endoscope 10 and the processor device 11 are turned on, and the insertion unit 12 is inserted into the subject. . Then, the irradiation device 24 is turned on, and the image obtained by the CCD 21 is observed on the monitor 16 while illuminating the inside of the subject with the light irradiated from the illumination window 22.

  The image light of the observed region taken in from the objective optical system 20 is imaged on the imaging surface of the CCD 21, whereby an imaging signal is output from the CCD 21. The imaging signal output from the CCD 21 is taken into the image memory 27. The imaging signals captured in the image memory 27 are sequentially read out to the normal image generation unit 28 and the spectral estimation processing unit 29.

  In the normal image generation unit 28, various signal processes are performed on the imaging signal read from the image memory 27, thereby generating a normal image. The normal image generated by the normal image generation unit 28 is output to the display control unit 30.

  On the other hand, in the spectral estimation processing unit 29, spectral estimation processing is executed using the imaging signal read from the image memory 27. A spectral estimation image generated as a result of the spectral estimation processing by the spectral estimation processing unit 29 is output to the spectral image generation unit 31.

  As shown in the flowchart of FIG. 8, in the spectral image generation unit 31, the calculation circuit 40 calculates the average of the spectral characteristics of all the pixels of the spectral estimation image, that is, the representative spectral characteristics Sr. Then, the enhancement processing circuit 41 performs enhancement processing on each pixel of the spectral estimation image according to the difference between the spectral characteristic of each pixel of the spectral estimation image and the representative spectral characteristic Sr.

  After the enhancement processing, when the spectral characteristic intensity of each pixel of the spectral estimation image does not fall within the predetermined range 0 to It, the spectral characteristic intensity of each pixel of the spectral estimation image is within the predetermined range 0. The emphasis amount is adjusted by the emphasis amount adjustment circuit 42 so as to be within the range of ~ It. Then, the enhancement processing circuit 41 performs the enhancement processing again with the adjusted enhancement amount. If the intensity of the spectral characteristic of each pixel of the spectral estimation image is within the predetermined range 0 to It, naturally, the enhancement amount is not adjusted.

  Next, the channel assignment circuit 43 assigns spectral images at arbitrary three wavelengths of the spectral estimation image after enhancement processing to the three RGB channels. As a result, a spectral image in which the intensity of the spectral characteristics of each pixel of the spectral estimation image at the assigned wavelength is represented by one of RGB gradation values is generated. The generated spectral image is output to the display control unit 30. The normal image and the spectral image output to the display control unit 30 are displayed side by side on the monitor 16 or individually.

  As described above, the electronic endoscope system 2 uses the calculation circuit 40 to calculate the representative spectral characteristic Sr that represents the overall hue of the spectral estimation image, and uses the calculated representative spectral characteristic Sr to perform the enhancement processing circuit 41. Because each pixel in the spectroscopic estimated image is subjected to emphasis processing, it is possible to absorb the effects on the spectroscopic image due to individual differences in organ colors and variations in the performance of electronic endoscopes. It can be carried out.

  In addition, since the enhancement amount is adjusted by the enhancement amount adjustment circuit 42 so that the intensity of the spectral characteristics of each pixel of the spectral estimation image falls within a predetermined range, image quality deterioration such as so-called whiteout or blackout is prevented. can do.

  Furthermore, since spectral images at arbitrary three wavelengths of the spectral estimation image after enhancement processing are assigned to the three channels of RGB set in advance by the channel assignment circuit 43, the spectral region in which a portion to be observed such as a lesion or blood vessel is highlighted. An image can be obtained.

In the above embodiment, an example in which the average of the spectral characteristics of all the pixels of the spectral estimation image is adopted as the representative spectral characteristic Sr has been described. However, the method using principal component analysis shown in FIGS. 9 and 10 is adopted. May be. That is, as shown in the upper and middle of FIG. 9 performs principal component analysis on the spectral estimation image by calculating circuit 40, the first principal component S ₁ thereby obtained as the representative spectral characteristics Sr. Incidentally, the first principal component S ₁ to the representative spectral characteristics Sr, the proportion occupied in the entire information (contribution rate), the first principal component S ₁ is because the highest.

Then, as shown in the lower part, at enhancement processing circuit 41, the enhancement processing on the second principal component S _2, and S 2 _'. The second principal component S ₂ is subjected to enhancement processing because the second principal component S ₂ has no correlation with the first principal component S ₁ , which is the representative spectral characteristic Sr, and the entire spectral estimated image This is because it represents information on a portion of a hue different from the hue.

It was subjected to emphasis processing on the second principal component S _2, as shown in FIG. 10, using the second principal component S _{2 'after} enhancement and a first principal component S _1, suitable principal component A spectral estimation image is reconstructed with coefficients, and a spectral estimation image in which a portion of a hue different from the overall hue of the spectral estimation image is emphasized is obtained in the same manner as shown in the lower part of FIG. The main component to be emphasized is not limited to the second main component. For example, when the contribution ratio of the first main component is not high, for example, the third main component or the fourth main component is emphasized. May be applied.

  Alternatively, a method using cluster analysis shown in FIG. 11 may be adopted. In this case, first, as shown in the upper part, the index p representing the spectral characteristics of all the pixels of the spectral estimation image is expressed as an XY two-dimensional space (actually, the number of elements of the index p is an n-dimensional space, Plot in space.)

  Then, as shown in the middle stage, the calculation circuit 40 performs cluster analysis on the plotted index p. Specifically, the Euclidean distance in the n-dimensional space between the indices p is calculated, and the indices p that are closest to each other are sequentially collected as a cluster. In this method, the maximum cluster Cm obtained by cluster analysis (the cluster having the largest size among the previous clusters in which all indices p are included in one cluster) is the most typical color of the entire spectral estimation image. Therefore, the maximum cluster Cm is set as the representative spectral characteristic Sr.

  Next, as shown in the lower stage, the emphasis processing circuit 41 differs from the index p not included in the maximum cluster Cm, that is, the overall color of the spectral estimation image, according to the distance from the center of gravity G of the maximum cluster Cm. Emphasis processing is performed on the index p representing the information on the shaded portion, and is set as p ′. For example, when the distance from the center of gravity G is large, the enhancement amount is also increased, and when the distance is small, the enhancement amount is also decreased. Then, the enhanced index p ′ is placed on the extension of the straight line connecting the center of gravity G and the original index p before enhancement. As a result, similarly to the two methods of the above-described embodiment, a spectral estimation image in which a portion of a hue different from the overall hue of the spectral estimation image is emphasized is obtained.

  In the above embodiment, when the intensity of the spectral characteristic of each pixel of the spectral estimation image does not fall within a predetermined range, the enhancement amount adjustment circuit 42 adjusts the enhancement amount, and then the enhancement processing circuit 41 emphasizes again. Although the processing is performed, the emphasis amount is adjusted before the emphasis processing is performed so that the intensity of the spectral characteristics of each pixel of the spectral estimation image is within a predetermined range by one emphasis processing. Good.

  The illumination device 24 does not need to be integrated with the electronic endoscope, and may be provided outside and connected by a cord. In the above-described embodiment, the electronic endoscope 10 has been described as an example. However, the ultrasonic transducer that irradiates the observation site with ultrasonic waves and receives an echo signal from the observation site has the tip portion together with the CCD. The present invention can also be applied to an ultrasonic endoscope that is integrally disposed on the endoscope.

It is the schematic which shows the structure of an electronic endoscope system. It is a block diagram which shows the structure of an electronic endoscope system. It is a block diagram which shows the structure of a spectral image generation part. It is explanatory drawing which shows the flow of a process in a calculation circuit and an emphasis processing circuit. It is explanatory drawing which shows the flow of the process in an emphasis amount adjustment circuit. It is a figure for demonstrating the process in a channel allocation circuit. It is a figure which shows the display state of a monitor. It is a flowchart which shows the flow of the process in a spectral image generation part. It is explanatory drawing which shows the flow of a process at the time of employ | adopting a principal component analysis. It is explanatory drawing which shows the flow of a process at the time of employ | adopting a principal component analysis. It is explanatory drawing which shows the flow of a process at the time of employ | adopting cluster analysis.

Explanation of symbols

2 Electronic Endoscope System 10 Electronic Endoscope 11 Processor Unit 16 Monitor 21 CCD
25 CPU
28 normal image generation unit 29 spectral estimation processing unit 31 spectral image generation unit 40 calculation circuit 41 enhancement processing circuit 42 enhancement amount adjustment circuit 43 channel allocation circuit 51 normal image 52 spectral image

Claims (15)

An operation method of an image processing apparatus that performs spectral estimation processing on an image acquired by an electronic endoscope and generates a spectral image having an arbitrary wavelength band based on the spectral estimation image obtained thereby,
The principal component analysis is performed on the spectral estimation image, and the first principal component obtained thereby is calculated as a representative spectral characteristic representative of the overall hue of the spectral estimation image by the calculation means,
An operation method of an image processing apparatus , wherein enhancement processing means is used to perform enhancement processing on each pixel of the spectral estimation image using the calculated representative spectral characteristics. The method of operating an image processing apparatus according to claim 1, wherein the enhancement processing is performed according to an n-th principal component (n is a natural number of 2 or more) obtained by the principal component analysis. The method of operating an image processing apparatus according to claim 2, wherein n = 2. 4. The enhancement amount of the enhancement processing is adjusted by an enhancement amount adjustment unit so that the intensity of spectral characteristics of each pixel of the spectral estimation image falls within a predetermined range. An operation method of the image processing apparatus according to any one of the above. 5. The image processing according to claim 1, wherein spectrum images at arbitrary three wavelengths of the spectral estimation image after the enhancement processing are allocated to three preset RGB channels by channel allocation means. How the device works . An image processing apparatus that performs spectral estimation processing on an image acquired by an electronic endoscope and generates a spectral image having an arbitrary wavelength band based on the spectral estimation image obtained thereby,
A calculation unit that performs principal component analysis on the spectral estimation image, and calculates the first principal component obtained as a representative spectral characteristic representative of the overall hue of the spectral estimation image;
An image processing apparatus comprising: enhancement processing means for performing enhancement processing on each pixel of the spectral estimation image using the calculated representative spectral characteristics.   The image processing apparatus according to claim 6, wherein the enhancement processing unit performs the enhancement processing according to an n-th principal component (n is a natural number of 2 or more) obtained by the principal component analysis.   The image processing apparatus according to claim 7, wherein n = 2.   9. The enhancement amount adjusting means for adjusting the enhancement amount of the enhancement processing so that the intensity of the spectral characteristic of each pixel of the spectral estimation image falls within a predetermined range. An image processing apparatus according to any one of the above.   The image according to any one of claims 6 to 9, further comprising channel allocating means for allocating spectrum images at arbitrary three wavelengths of the spectral estimation image after the enhancement processing to preset three channels of RGB. Processing equipment.   An electronic endoscope system comprising the image processing apparatus according to claim 6. An operation method of an image processing apparatus that performs spectral estimation processing on an image acquired by an electronic endoscope and generates a spectral image having an arbitrary wavelength band based on the spectral estimation image obtained thereby,
The cluster analysis is performed using the spectral characteristics of all the pixels of the spectral estimation image as a vector, and the maximum cluster is calculated as a representative spectral characteristic representative of the overall hue of the spectral estimation image by the calculation means,
An operation method of an image processing apparatus , wherein enhancement processing means is used to perform enhancement processing on each pixel of the spectral estimation image using the calculated representative spectral characteristics. The method of operating an image processing apparatus according to claim 12, wherein the enhancement processing is performed according to a distance between a spectral characteristic of each pixel of the spectral estimation image and a center of gravity of the maximum cluster. An image processing apparatus that performs spectral estimation processing on an image acquired by an electronic endoscope and generates a spectral image having an arbitrary wavelength band based on the spectral estimation image obtained thereby,
Performing a cluster analysis using the spectral characteristics of all the pixels of the spectral estimation image as a vector, and calculating a maximum cluster as a representative spectral characteristic representative of the overall hue of the spectral estimation image;
An image processing apparatus comprising: enhancement processing means for performing enhancement processing on each pixel of the spectral estimation image using the calculated representative spectral characteristics.   The image processing apparatus according to claim 14, wherein the enhancement processing unit performs the enhancement processing according to a distance between a spectral characteristic of each pixel of the spectral estimation image and a center of gravity of the maximum cluster.

Images