JP3126996B2 - Endoscope image processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of processing an endoscope image in which a pixel is decomposed into a plurality of color signals and an appropriate process is performed on the decomposed color signals to thereby improve recognition performance. .

[0002]

2. Description of the Related Art In recent years, by inserting an elongated insertion portion into a body cavity, it is possible to observe a diseased part or the like in the body cavity without performing an incision, and to perform treatment using a processing tool as necessary. Endoscopes have become widely used.

[0003] In addition to an optical endoscope (fiberscope) using an image guide for an image transmission procedure, an electronic endoscope (a CCD or the like) using a solid-state imaging procedure has recently been used as the endoscope. Hereinafter, an electronic endoscope or an electronic scope) has been put to practical use.

[0004] In addition, CC is used for the eyepiece of the fiberscope.
There is also a device in which an imaging camera using a solid-state imaging procedure such as D is connected to enable color display on a monitor.

[0005] Recently, attempts have been made to improve the diagnostic ability by assisting human visual recognition by performing various processes on video signals obtained from such endoscopes.

For example, according to US Pat. No. 4,819,077,
The video signals RGB representing the three primary colors are converted into signals relatively close to human color perception called HSI (hue, saturation, luminance) by relatively simple coordinate conversion, and appropriate emphasis is applied to each signal of HSI. An apparatus is disclosed in which after processing is performed, the image signal is returned to the video signal RGB again by inverse conversion to display an image.

[0007] For example, according to Japanese Patent Application No. 62-296143, there is provided a device which converts an image signal into a Lu ^* v ^* color space which is closer to human color perception, performs various enhancements, and performs an inverse change. It has been disclosed.

The present applicant also discloses in Japanese Patent Application No. 61-169057 an apparatus for converting a video signal into a signal in the La ^* b ^* or Lu ^* v ^* color space and performing an emphasis process on the luminance L. are doing.

In these conventional examples, in order to emphasize the outline or fine pattern of a color image, a process of emphasizing high-frequency components with respect to luminance is performed. This is because, by separating luminance, hue, and saturation, emphasis can be performed without changing the color tone. On the other hand, when an independent emphasis process is performed on each of the RGB planes, the color tone changes and an unnatural image may occur.

[0010] By the way, in the living body observation using an endoscope, depending on the object to be observed, a sufficient resolution, brightness, contrast, etc. may be required depending on the object to be observed due to restrictions on equipment such as reduction in size and diameter to improve insertion properties. It also happens that no performance is obtained. Observation of microscopic patterns on the mucosal surface (formed by glandular cavities such as gastric and intestinal glands , anonymous sulcus, capillaries, or stains), which are important information for differential diagnosis between benign and malignant lesions. For this reason, even if an endoscope having the highest resolution at present is used, a sufficient image for diagnosis cannot be obtained in many cases.

For this reason, it is natural that an endoscope having better resolution is desired. However, an image processing method and apparatus for compensating for lack of contrast and resolution and making diagnosis easier. The development of is desired.

[0012]

For example, in the case of a fine pattern on the mucous membrane surface in a normal endoscopic image (without using a staining method), most of the fluctuation information forming the pattern is based on fluctuations of G and B signals. . This is probably reflected by the nature of the absorption by hemoglobin in the blood.

Further, in an image using a staining method, light absorption due to a stain is added in addition to reflection or light absorption inherent in a living body, and thus data fluctuations for forming an image are determined by these properties.

In order to enhance these images, high-frequency enhancement processing in a color space, which has been conventionally attempted, has not been successful. The reason is presumed to be as follows.

The data distribution formed by the data fluctuation forming the fine pattern does not naturally correspond to any fluctuation in luminance, hue, and saturation in human color perception. The observer first grasps global information from the image. Next, by observing the fine pattern clearly, an attempt is made to obtain more detailed information. Then, when trying to obtain detailed information, the observer's attention is focused intensively on any information derived from the living body observed in the image. Whether this information is represented by luminance, saturation or hue,
It doesn't make any sense to the observer whether they are or both.

The most well-accepted enhancement method that can be used to assist observation is that data fluctuations of interest in the region of interest must be amplified and otherwise suppressed. Is not always satisfied with this condition.

For this reason, the emphasis processing in the color space does not always provide an optimum emphasis result for an observer of a fine pattern.

Furthermore, data fluctuations in the entire endoscope image are often governed by global fluctuations due to illumination, the shape of the object, and the like. The emphasis on the brightness tends to make the contours and the like formed by the global fluctuation more clear, so that a fine pattern may be difficult to see.

The present invention has been made in view of the above points, and an object of the present invention is to provide a method for processing an endoscope image that allows a fine pattern on a mucous membrane surface to be clearly observed.

[0020]

SUMMARY OF THE INVENTION According to the present invention,
The method of processing an endoscope image is a method of processing an endoscope image that performs predetermined image processing on each signal of an endoscope image divided into a plurality of color signals. definitive the pixel information, based on the coordinate axis determining <br/> procedure for determining the coordinate axes, determined by the coordinate determination procedure axes for the coordinate conversion based on the statistical characteristics of the distribution of frequency components
Can, wherein a coordinate transformation procedure for each signal to coordinate transformation, frequency of definitive to each pixel information after the coordinate conversion by the coordinate transformation procedure
A filter processing procedure for performing a desired filter processing on several components; and a plurality of original color signals for each pixel information after the coordinate conversion filtered by the filter processing procedure.
Characterized in that it comprises a reverse coordinate transformation procedure for inverse coordinate transformation coordinate consisting of No., also according to the invention
Method of processing an endoscopic image, among divided into a plurality of color signals
Endoscope image that performs predetermined image processing on each signal of the endoscope image
In the image processing method, each image of each signal of the endoscope image is provided.
Statistical features related to the distribution of frequency components in elementary information
A step of calculating a coordinate transformation matrix based on the coordinates strange
Using the coordinate transformation matrix obtained in the procedure for calculating the permutation matrix
A coordinate transformation procedure for coordinate converting the respective signal Te, the coordinates
Frequency in each pixel information after coordinate transformation by transformation procedure
Filtering means for performing desired filtering on components
Order and filtered by the filtering procedure
For each pixel information after the coordinate conversion, the coordinate conversion matrix
An inverse coordinate conversion procedure for performing an inverse coordinate conversion using an inverse matrix.
It features a has, then the filtering procedure
The filtering performed by the filter is emphasis and / or suppression.
It is characterized by that.

[0021]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the endoscope apparatus, and FIG.
Is a block diagram showing the configuration of the image processing apparatus, FIG. 3 is a side view showing the entire endoscope apparatus, FIGS. 4 and 5 are flowcharts for explaining the operation of the image processing apparatus, and FIG. FIG. 7 is a flowchart showing an example of the setting, FIG. 7 is a flowchart for explaining the operation of the band-pass filtering process, FIG. 8 is a conceptual diagram showing the distribution of the correlation between the values of the R and G data on the image, and FIG. FIG. 10 is a conceptual diagram showing an example of correlation distribution after band-pass filtering with respect to distribution, FIG. 10 is an explanatory diagram of weighting using a cos function, FIGS. 11 and 12 are explanatory diagrams of band-pass filtering processing, and FIG. FIG. 14 is an explanatory diagram showing a weighting function used to perform the above-described operation, and FIG. 14 is an explanatory diagram showing an example of automatic setting of a characteristic region.

The endoscope apparatus according to the present embodiment includes an electronic endoscope 1 as shown in FIG. The electronic endoscope 1 has a slender, for example, flexible insertion section 2, and a large-diameter operation section 3 is connected to the rear end of the insertion section 2. A flexible universal cord 4 extends laterally from the rear end of the operation unit 3, and a connector 5 is provided at the tip of the universal cord 4. The electronic endoscope 1 is connected via the connector 5 to a video processor 6 having a light source device and a signal processing circuit built therein. Further, the video pro set 6 includes a monitor 7
Are connected.

On the distal end side of the insertion portion 2, a rigid distal end portion 9 and a bending portion 10 which can be bent rearward adjacent to the distal end portion 9 are sequentially provided. By rotating a bending operation knob 11 provided on the operation section 3, the bending section 10 can be bent in the left-right direction or the up-down direction. Further, the operation section 3 is provided with an insertion port 12 which communicates with a treatment instrument channel provided in the insertion section 2.

As shown in FIG. 1, a light guide 14 for transmitting illumination light is inserted into the insertion section 2 of the electronic endoscope 1. The distal end surface of the light guide 14 is
, And the illumination light can be emitted from the tip 9. In addition, the light guide 14
Is inserted through the universal cord 4 and connected to the connector 5. In addition, the tip portion 9 includes:
An objective lens system 15 is provided, and a solid-state imaging device 16 is provided at an image forming position of the objective lens system 15. The solid-state imaging device 16 has sensitivity in a wide wavelength range from the ultraviolet region to the infrared region including the visible region. Signal lines 26 and 27 are connected to the solid-state imaging device 16, and these signal lines 26 and 27 are inserted through the insertion section 2 and the universal cord 4 and connected to the connector 5.

On the other hand, the video processor 6 is provided with a lamp 21 which emits light in a wide band from ultraviolet light to infrared light. As the lamp 21, a general xenon lamp, a strobe lamp, or the like can be used.
The xenon lamp and the strobe lamp emit a large amount of ultraviolet light and infrared light as well as visible light. This lamp 2
1 is supplied with power from a power supply 22. A rotary filter 50 driven by a motor 23 is disposed in front of the lamp 21. In the rotary filter 50, filters for transmitting light in each wavelength region of red (R), green (G), and blue (B) for normal observation are arranged along the circumferential direction.

The rotation of the motor 23 is controlled by a motor driver 25 to be driven.

R, G, R, G,
Light separated in time series into light of each wavelength region of B is incident on the incident end of the light guide 14, guided to the tip 9 via the light guide 14, and emitted from the tip 9. Thus, the observation site is illuminated.

The return light from the observation site due to the illumination light is focused on the solid-state image pickup device 16 by the objective lens system 15 and is photoelectrically converted. A drive pulse from a drive circuit 31 in the video processor 6 is applied to the solid-state imaging device 16 via the signal line 26, and reading and transfer are performed by the drive pulse. The video signal read from the solid-state imaging device 16 is input to a preamplifier 32 provided in the video processor 6 or the electronic endoscope 1 via the signal line 27. The image signal amplified by the preamplifier 32 is input to a process circuit 33, subjected to signal processing such as γ correction and white balance, and converted into a digital signal by an A / D converter 34. The digital image signal is supplied to a memory (1) 36a, a memory (2) 36b, and a memory (3) 36 corresponding to, for example, each of red (R), green (G), and blue (B) by the selection circuit 35.
c is selectively stored. The memory (1) 36a, the memory (2) 36b, and the memory (3) 36
c is read out at the same time, converted into an analog signal by a D / A converter 37, and input to the color monitor 7 as R, G, B signals via an input / output interface (I / O) 38. By the color monitor 7,
The observation site is displayed in color.

In the video processor 6, a timing generator 42 for generating the timing of the entire system is provided. The timing generator 42 synchronizes the motor driver 25, the drive circuit 31, the select circuit 35 and other circuits. Has been taken.

In this embodiment, the memories (1-3) 36
Output terminals a to 36c are connected to the image processing device 104. The image processing apparatus 104 is connected to a monitor 10 via an input / output interface (I / O) 105.
6 is connected, and the result of the arithmetic processing by the image processing device 104 is displayed on the monitor 106.

The image processing device 104 has a configuration as shown in FIG. That is, the image processing apparatus 10
Reference numeral 4 denotes a CPU 108, an information input device 109, a main storage device 110 including a RAM, and an image input interface (I
/ F) 111 and a display interface 112,
These are connected to each other by a bus. The information input device 109 is a keyboard or the like, and can input data such as the type of the electronic endoscope 1. The image input interface 111 has a memory (1) 3
6a, the memory (2) 36b, and the memory (3) 36c, and receives image data from each memory. The display interface (I / F)
112 is connected to the input / output interface 105,
Image data to be input to the monitor 106 is sent.

In the present embodiment, the image of the target part obtained by the electronic endoscope 1 is processed by the image processing device 104, and the processing result is output to the monitor 106.

Next, the processing of the image processing device 104 will be described with reference to a flowchart. The image processing device 104 stores each of the memories 36a, 36a in step S1 of FIG.
RGB video signals are input from 36b and 36c, respectively. Then, the image processing device 104 performs a coordinate axis conversion on the RGB image as shown in FIG.
Step S for performing a two- or two-dimensional discrete Fourier transform process
4. A processing group including a step S5 for performing so-called filtering, a step S7 for performing a two-dimensional discrete Fourier inverse transform, a step S9 for performing a coordinate axis inverse transform, and 3 × for performing a coordinate axis transform as shown in FIG. A processing group including steps S11 to S16 for deriving three matrices and their inverses is provided. The image processing device 104 processes R, G, and B image signals from the memories (1 to 3) 36a to 36c. It is assumed that the two-dimensional discrete Fourier transform and the inverse transform in this embodiment have a coefficient arrangement corresponding to a so-called optical Fourier transform in which a DC component is arranged at the center.

Generally, the fluctuation of density value information constituting an unstained endoscope image is large in the R image, but mainly in the low frequency component. And the information on the fine features of the actual organ mucosal surface,
There is a strong tendency to rely on G and B images. Further, the density value information shows a different distribution depending on each image. For example, in the case where indigo / carmine staining is performed, there are some information in which the R image contains a large amount of information on the feature.

FIG. 8 shows the concept of the distribution of the correlation between the R and G data in each pixel of the 256 gradation image as an example. Here, the density value is treated as two-dimensional data for simplicity, but actual (not shown) B data has similarity to G data. The distribution of the density information shown in FIG. 8 includes all the low-frequency components and high-frequency components in the image, but includes information indicating the characteristics of the surface of the mucous membrane of the organ (hereinafter, referred to as “characteristic”)
Characteristic component) mainly exists in a high-frequency component in a certain band. In the present embodiment, attention is paid to this point, and a band that seems to include the characteristic component is extracted by applying band-pass filtering.

FIG. 9 shows the concept of the distribution of correlation after band-pass filtering is applied to the R and G data of FIG. In conventional endoscopic captured image, the high-frequency component constituting the feature is mainly G (and B) for abundant in the image, after the bandpass biased to variations in the G data as compared to the R data correlation Is shown. In addition, the distribution of this correlation differs from image to image, and particularly in a captured image using a stain (eg, indigo carmine),
Since the stain has a property of being absorbed by R, the R image contains more characteristic components than the G and B images. In order to effectively emphasize the feature components on each of the original images to be processed, the statistical distribution of the extracted feature components is obtained, three new axes are set in place of RGB, and the original image is set with respect to each of these coordinate axes. Is emphasized on the result of the conversion. Hereinafter, examples of specific processing contents will be presented.

A region having a size of, for example, 256 × 256 (hereinafter, referred to as a characteristic region) including, in the center, a characteristic region of the RGB original image or a region where enhancement processing is desired in the RGB original image is shown in FIG. This is set in step S11. Note that the center point of the image having the size of 256 × 256 is a pixel corresponding to the point (128, 128). The setting of the characteristic region can be automatically performed in addition to the selection based on the subjective opinion of the operator. The automatic setting process of the characteristic region according to the flowchart shown in FIG. 6 is an example thereof.

The characteristic component in a normal endoscopic image is G
By utilizing the property of being included most in the image, a characteristic region used for deriving a coordinate transformation matrix that enables effective enhancement processing is set. First, as shown in FIG. 6, the G image in the original image is converted
In step 1, the block is divided into blocks of, for example, 64 × 64. A process of extracting a certain block from these block groups as the center of the characteristic region is performed. That is, in step S32, a block including halation at a certain ratio or more is excluded from the central region candidates of the characteristic region. Then, in step S33, a density average value excluding the halation portion in each block that has not been excluded is obtained. Subsequently, in step S34, blocks that do not satisfy a certain average density value, that is, blocks that are dark portions of the image and that can be determined to be inappropriate for extraction as a feature region, are excluded from the central region candidates. In the example of FIG. 14, a block group D indicated by a thick line is a target of a dark part. After the above series of processing, the average density value in each block not excluded is obtained in step S35, and the block having the average density value closest to the obtained value is defined as the central part of the feature region. In step S36, a 256 × 256 characteristic region including the block at the center is extracted. Such as the edge of the image,
When a block in which an area satisfying the size of 256 × 256 cannot be obtained is selected, a block having a density average value close to the above-described value and from which a characteristic area can be extracted is set as a central part.

Next, for each of the RGB data of each pixel in the characteristic region in step S12 of FIG. 5, the center of the image is set to 1 in step S13, and the distance from the center (corresponding to the number of pixels) n is reduced. , N = 128, and multiplication of the weight function by the cos function shown in the following equation is applied.

Wc (n) = 0.5 × cos (nπ / 128) +0.5 (1) FIG. 10A shows a weighting value for each pixel on a line segment passing through the center of the characteristic region image. Indicates a change. FIG.
In (b), all the weighting values are set to 0 for an area where n> 128 indicated by a hatched portion. That is, the same value is applied to the weighting function wc (n) shown in Expression (1) concentrically with respect to the center point of the image. Tenth
Circles shown in FIG. (B) shows the points of n = 128, the hatched portion represents a region to be n> 128
You. In other words, the weight according to equation (1) is applied when n ≦ 128, and points that do not apply are unconditionally multiplied by zero. This processing has the function of reducing the occurrence of distortion of the band-pass image due to discontinuity in the peripheral part in the processing in band-pass filtering described later.

Further, at step S13 in FIG.
For the feature area image weighted by the function,
In step S14, band pass filtering is performed. FIG. 7 is a flowchart showing a specific example of the band-pass filtering process. After performing a two-dimensional discrete Fourier transform on each of the RGB images in step S21 of FIG.
In step S22, for example, filtering having the pass characteristics shown in FIG. 12 is performed. Here, the band to be passed is set to the above-mentioned frequency band that is considered to contain the most characteristic component in the image captured by the endoscope . Next, in step S23 of FIG. 7, two-dimensional discrete Fourier inverse transform is performed. To derive the coordinate transformation matrix, as shown in FIG. 11, 128 × 128 including the center of the image obtained by the above-described processing
(Hereinafter referred to as a feature image). The obtained image is obtained by extracting only the characteristic components from the original image. The concept of the distribution of the correlation between the RGB data is, for example, as shown in FIG.
It is based on. Then, in step S15 of FIG. 5, it is determined whether or not the processing has been performed on all the RGB images of the characteristic region. When the processing is completed, the process proceeds to the next step.

Then, in order to obtain a statistical distribution of RGB data in the characteristic components of the characteristic image obtained by applying the above-described processing, a coordinate transformation / inverse transformation matrix using the KL transformation in step S16 of FIG. Apply decision processing. The KL transform is a process for mechanically deriving a new suitable orthogonal function system based on the statistical distribution of input image data. Note that the KL conversion is detailed in a literature (Digital Image Processing Makoto Nagao Modern Science Co., Ltd.), so the details are omitted. Through step S16, three three-dimensional vectors used for coordinate axis conversion are obtained. The X, Y, and Z axes that are the coordinate axes after the conversion are defined as those with the largest eigenvalue of these three vectors as the X axis and those with the smallest eigenvalue as the Z axis. Each vector defining each of the XYZ axes is defined as a vector ax (a11, 12, a1).
3), vector ay (a21, a22, a23) and vector az (a31, a32, a33), RGB
The coordinate axis conversion from the space to the XYZ space using the new three axes is realized by a matrix operation represented by the following equation (2).

[0043]

Here, assuming that the 3 × 3 vector matrix in the equation (2) is a matrix M1, the coordinate axis inverse transformation is realized by multiplying the XYZ data by the inverse matrix of the matrix M1.

In step S2 of FIG. 4, step S16
(2) is performed using the coordinate axis conversion matrix generated in step (1) to convert (r, g, b) data of each pixel in the RGB original image into (x, y, z) data.

In step S4 of FIG. 4, a two-dimensional discrete Fourier transform is performed on each of the XYZ pixels in step S3.

The filtering in step S5 is as follows.
The real term ar (t) and the imaginary term ai of the data generated by the two-dimensional discrete Fourier transform in step S4
This is realized by multiplying (t) by a weighting function in step S6 described below. This weighting function is
For X, Y, and Z pixels, an appropriate one is given to each.
Specifically, for example, a weighting function that performs filtering having an emphasis effect on an X image including many feature components and applies filtering having a noise suppression effect on Y and Z images is applied. After the filtering is completed in step S5, the two-dimensional discrete Fourier transform in step S7 can obtain the X, Y, and Z pixels after the application of the enhancement processing result.

The process of creating a weighting function used for filtering in step S6 is performed as follows. In the so-called optical Fourier transform, a DC component that determines the overall density value of an image is located at the center on a closed spatial frequency surface, and a low frequency component that is referred to as a spatial frequency exists near the DC component. The filter to be created is such that a point separated by a certain distance from the center is set as the highest point of emphasis, and the point is used as a boundary to reduce the value of the weight for the low frequency component and the high frequency component. Various types of weighting functions having such a property can be considered, and the following filter is an example.

Let the coordinates of the center O of the spatial frequency plane be (u0, v
0), assuming that the coordinates of the target point P on the same plane are (u, v), the distance x of the OP is

[0050]

Is represented by When the weighting value is w, w is expressed as a function w (x) for x, and the value satisfies 0 ≦ w (x) ≦ α. α is the maximum value of the weight. The filter created here is centered on O and w
Let p be the diameter of a circle consisting of a set of points giving the maximum value of (x)
Then, for each case of 0 ≦ x <p / 2 and p / 2 ≦ x, there are separate functions, where 0 ≦ x <p /
In 2, the cos function is used, and when p / 2 ≦ x, the normal distribution function is used.

In the cos function part of the filter to be created, its amplitude is assumed to be A. The cos function part is continuous with the normal distribution function part at x = p / 2, and w (p / 2) = α
It is assumed that In addition, the minimum value must be taken at x = 0. The following equation (3) is obtained as a function satisfying these conditions.

W (x) = α−A−A · cos {x · π / (p / 2)}, (0 ≦ x <p / 2) (3) The function of equation (3) is When x = p / 2, the maximum value α, x =
Minimum value α-2A at 0 (center in the spatial frequency domain)
Take. If the standard deviation is σ in such a function, the following equation (4) is obtained.

[0054]

In equation (4), σ is given by σ = (CTR−
(P / 2)) / r (CTR: value of x coordinate of center, r: real number). When p = 0, only equation (3) is applied. In Equations (3) and (4), α,
By using A, p, and r as parameters, filters with different degrees of enhancement can be obtained. FIG. 12 shows an example of the characteristics of a filter created under the above-mentioned conditions.
4, A = 1.5, p = 60, r = 4.
To realize filtering with noise suppression effect,
High frequency components considered to contain many noise components are suppressed, for example, α = 1, A = 0, p = 0,
It is sufficient to set r = 3. In addition, in order to realize filtering having an image enhancement effect, a frequency band that is considered to include a lot of information on a structure pattern of an original image is emphasized,
Suppress high frequency components from the band, for example, α = 3, A
= 1, p = 50, r = 4-7.

In step S8 in FIG. 4, it is determined whether or not all the processing up to step S7 for each of the X, Y, and Z pixels is completed. After the end, the coordinate axis inverse transformation is performed in step S9, and an RGB image as a result of the enhancement processing is obtained. Specifically, as shown in the following equation (5), the inverse matrix M2 of the matrix M1, (x ', y',
z ') It may be applied to data.

[0057]

Returning to the RGB image again, step S in FIG.
In (10), (r ', g',
b ') Data is obtained. Then, the monitor 106 displays the RGB processing result as an image.

In this embodiment, the image processing device 104 can obtain a clear image in which the fine pattern of the mucous membrane surface is emphasized and the noise is suppressed with respect to the RGB original image.

In performing the above processing, noise removal processing (for example, a median filter having a 3 × 3 mask size) may be performed as preprocessing for the original image. The method for automatically setting the characteristic region is not limited to the example shown in the present embodiment, but various methods are conceivable. Further, the method of determining the coordinate transformation matrix in step S16 in FIG. 5 is not limited to the method using the KL transformation, and may be appropriately used as long as the coordinate transformation matrix is adaptively derived for the processing target image. It is possible.
Further, each processing in the present embodiment can be realized as parallel processing.

The matrix in the present embodiment is not obtained one by one according to the image, but may be a matrix derived in advance, for example, due to a difference in variance of the R image. In addition, the processing unit of a series of two-dimensional discrete Fourier transform, filtering, and two-dimensional inverse Fourier transform in the present embodiment can be realized as a mask operation by convolution.

[0062]

As described above, the method for processing an endoscope image according to the present invention has an effect that a fine pattern on the surface of a mucous membrane can be clearly observed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an endoscope apparatus.

FIG. 2 is a block diagram illustrating a configuration of an image processing apparatus.

FIG. 3 is a side view showing the entire endoscope apparatus.

FIG. 4 is a flowchart illustrating the operation of the image processing apparatus.

FIG. 5 is a flowchart for explaining the operation of the image processing apparatus.

FIG. 6 is a flowchart illustrating an example of automatic setting of a characteristic region.

FIG. 7 is a flowchart for explaining the operation of a band-pass filtering process.

FIG. 8 is a conceptual diagram showing an example of a distribution of correlation between values of R and G data on an image.

FIG. 9 is a conceptual diagram showing an example of a distribution of correlation after band-pass filtering with respect to the distribution of FIG.

FIG. 10 is an explanatory diagram of weighting using a cos function.

FIG. 11 is an explanatory diagram of a band-pass filtering process.

FIG. 12 is an explanatory diagram of a band-pass filtering process.

FIG. 13 is an explanatory diagram showing a weighting function used for performing filtering.

FIG. 14 is an explanatory diagram showing an example of automatic setting of a characteristic region.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Electronic endoscope 16 ... Solid-state image sensor 104 ... Image processing device 106 ... Monitor

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-1-176561 (JP, A) JP-A-1-1244071 (JP, A) JP-A-63-26783 (JP, A) JP-A-1- 138877 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06T 1/00 G06T 5/00-5/20 A61B 1/04

Claims (3)

(57) [Claims] 1. An endoscope image processing method for performing predetermined image processing on each signal of an endoscope image divided into a plurality of color signals, wherein each pixel information of each signal of the endoscope image is Frequency
Coordinate transformation based on statistical features of component distribution
A coordinate axis determining step of determining a coordinate axis for the operation, based on the coordinate axis determined in the coordinate axis determining step,
A coordinate transformation procedure for coordinate conversion of each signal, put to each pixel information after the coordinate conversion by the coordinate transformation procedure
A filtering process for performing a desired filtering process on a frequency component to be processed, and a plurality of original color signals for each pixel information after the coordinate transformation filtered by the filtering process.
And an inverse coordinate conversion procedure for performing an inverse coordinate conversion to a coordinate axis to be performed. 2. An endoscope image divided into a plurality of color signals.
An endoscope image processing method that performs predetermined image processing on each signal
Method, in each pixel information of each signal of the endoscope image,
Coordinate transformation matrix based on statistical characteristics of component distribution
And the coordinate transformation obtained in the step of calculating the coordinate transformation matrix
Coordinate conversion procedure for performing coordinate conversion of each of the signals using a matrix
And each pixel information after coordinate conversion by the coordinate conversion procedure.
Filter that performs the desired filtering on the frequency components
And a coordinate transformation filtered by the filtering procedure.
For each pixel information after conversion, the inverse matrix of the coordinate transformation matrix
And an inverse coordinate conversion procedure for performing an inverse coordinate conversion using the method . 3. A filter processing performed in the filter processing procedure.
Is characterized by emphasis and / or suppression.
The method for processing an endoscope image according to claim 1 .