CN115689935A - Data enhancement method of colonoscope image - Google Patents
Data enhancement method of colonoscope image Download PDFInfo
- Publication number
- CN115689935A CN115689935A CN202211404052.6A CN202211404052A CN115689935A CN 115689935 A CN115689935 A CN 115689935A CN 202211404052 A CN202211404052 A CN 202211404052A CN 115689935 A CN115689935 A CN 115689935A
- Authority
- CN
- China
- Prior art keywords
- image
- colonoscope
- intensity value
- area
- target domain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a data enhancement method of a colonoscope image, belonging to the technical field of computer vision and comprising the following steps: acquiring a plurality of original colonoscope images, and fusing gray information and saturation information to acquire a coordinate point set of a reflection region of the original colonoscope images; calculating a coordinate point set of a small reflective area from the reflective area coordinate point set and recovering the coordinate point set to obtain a small reflective area recovery image; converting the source image and the target domain image into an LAB color space; calculating the standard deviation and the mean value of each color channel except for a light reflecting area and a dark color area in the source image and the target area image; migrating the color domain of the source image to a target domain corresponding to the target domain image; restoring the dark area of the transferred image to obtain a color domain transfer enhanced image; and forming a colonoscope enhanced image data set by the original colonoscope image, the small reflection area restored image and the color domain migration enhanced image, wherein the colonoscope enhanced image data set is used for subsequent image analysis.
Description
Technical Field
The invention belongs to the technical field of computer vision, and particularly relates to a data enhancement method of a colonoscope image.
Background
Colonoscopy is an important standard for performing screening of intestinal diseases, and currently, manual examination is mainly relied on by doctors. However, limited to constraints of doctor experience, physical strength, equipment imaging and the like, the manual examination still has a large possibility of missed lesion detection.
With the development of artificial intelligence technology, the analysis performance of the convolutional neural network in the visual field is greatly improved, and the performance of the convolutional neural network in some visual analysis tasks even exceeds the human level. In view of this, convolutional neural networks are widely used for colonoscope lesion intelligent analysis, such as polyp detection, typing, bleeding detection, etc. Although the corresponding algorithms have shown certain analysis performance, such as focus recall rate, the examination results at the same expert level still have large gaps. One of the large constraints of existing artificial intelligence algorithms for clinical colonoscopy analysis is their adaptability to cross-domain scenarios. The cross-domain mainly refers to the problem that images generated when different devices acquire the images have difference. The artificial intelligence algorithm is mainly learned through an artificially constructed image data set, and the artificial intelligence algorithm is easily introduced into local optimization through single data. The existing image data enhancement algorithm is mainly carried out through modes of image rotation, cropping, scaling, color random transformation and the like, and a technical method aiming at the image characteristics of the colonoscope is still lacked.
In summary, in a colonoscope scene, when the images of the colonoscopy lesions are analyzed by artificial intelligence, the stability of the analysis result is poor and the reliability is low because the reference image data is single.
Disclosure of Invention
The embodiment of the invention aims to provide a data enhancement method of a colonoscope image, which can solve the technical problems of single reference image data, poor stability of an analysis result and low reliability in the conventional process of carrying out colonoscope lesion analysis by artificial intelligence.
In order to solve the technical problem, the invention is realized as follows:
the embodiment of the invention provides a data enhancement method of a colonoscope image, which comprises the following steps:
s101: acquiring a plurality of original colonoscope images, and converting the color space of each original colonoscope image into a gray scale space and a hue-saturation-brightness color space respectively to obtain gray scale information and saturation information of the original colonoscope images;
s102: fusing the gray information and the saturation information, and obtaining a reflection region coordinate point set of the original colonoscope image according to a first preset intensity threshold and a second preset intensity threshold;
s103: calculating a coordinate point set of a small light reflecting area from the coordinate point set of the light reflecting area according to a third preset intensity threshold value;
s104: restoring the coordinate point of the small reflecting area to obtain a small reflecting area restored image;
s105: selecting a source image and a target domain image, and extracting a light reflecting region and a dark region by fusing a multi-feature space, wherein the source image is an original colonoscope image or a small light reflecting region recovery image;
s106: converting the source image and the target domain image into an LAB color space to obtain component characteristics of the source image and component characteristics of the target domain image, wherein the resolution of the target domain image is higher than that of the source image and has the same sample as that of the source image, and the component characteristics comprise an L component intensity value, an A component intensity value and a B component intensity value;
s107: calculating the standard deviation and the mean value of each color channel of the image area except the light reflecting area and the dark area in the source image according to the component characteristics of the source image and the component characteristics of the target area image, and calculating the standard deviation and the mean value of each color channel of the image area except the light reflecting area and the dark area in the target area image;
s108: migrating the color domain of the source image to a target domain corresponding to the target domain image according to the standard deviation and the mean of the source image and the standard deviation and the mean of the target domain image to obtain a migrated image;
s109: restoring the dark area of the transferred image to obtain a color domain transfer enhanced image;
s110: and forming a colonoscope enhanced image data set by the original colonoscope image, the small reflecting area recovery image and the color domain migration enhanced image, wherein the colonoscope enhanced data image set is used for subsequent colonoscope image analysis.
In the embodiment of the invention, gray information and saturation information of a plurality of original colonoscope images are fused, a small reflecting area coordinate point set is extracted by presetting an intensity threshold value, then the small reflecting area restored images are recovered, data enhancement is carried out on the original colonoscope images, the data volume of the colonoscope image enhanced data set is increased, in addition, the original colonoscope images and the small reflecting area restored images are selected to carry out multi-feature fusion with a target domain image, a reflecting area and a dark area are extracted and converted into an LAB color space, the original colonoscope images and the small reflecting area restored images are subjected to color domain migration to the target domain image, then the dark area is recovered, a color domain migration enhanced image is obtained, and the data volume of the colonoscope enhanced image data set is further expanded. The obtained colonoscope enhanced image data set provides richer reference image data for artificial intelligence analysis of the colonoscope image, and the stability and reliability of an analysis result are improved.
Drawings
Fig. 1 is a flowchart illustrating a data enhancement method for a colonoscope image according to an embodiment of the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings in combination with embodiments.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The data enhancement method for colonoscope image implemented by the present invention is described in detail by the specific embodiments and the application scenarios thereof with reference to the attached drawings.
Referring to fig. 1, a flowchart of a data enhancement method for a colonoscope image according to an embodiment of the present invention is shown.
The embodiment of the invention provides a data enhancement method of a colonoscope image, which comprises the following steps:
s101: acquiring a plurality of original colonoscope images, and converting the color space of each original colonoscope image into a gray scale space and a hue-saturation-brightness color space respectively to obtain gray scale information and saturation information of the original colonoscope images.
In a possible implementation manner, S101 is specifically:
s1011: acquiring colonoscope video data;
s1012: and (4) frame cutting is carried out on the colonoscope video data to obtain a plurality of original colonoscope images.
S102: and fusing the gray information and the saturation information, and obtaining a reflection region coordinate point set of the original colonoscope image according to the first preset intensity threshold and the second preset intensity threshold.
In a possible implementation manner, S102 specifically includes:
s1021: obtaining a set P of coordinates of a retroreflective region of the original colonoscope image according to the following formula 1:
P＝{x|x∈D，G(x)＞Th 1 &S(x)＜Th 2 formula 1
Where P represents the set of light reflection region coordinate points, D represents the set of original colonoscope image coordinate points, G (x) represents the intensity value of the gray scale space at coordinate x in the original colonoscope image, and S (x) represents the intensity value of the saturation space at coordinate x in the original colonoscope image, where Th 1 Indicating a first predetermined intensity threshold, th 2 Indicating a second predetermined intensity threshold, sign&Representing a logical and operation.
Optionally, the first preset intensity threshold and the second preset intensity threshold are set to 240 and 0.3, respectively.
S103: and calculating a coordinate point set of the small light reflecting area from the coordinate point set of the light reflecting area according to a third preset intensity threshold value.
In a possible implementation, S103 specifically includes:
s1031: and (3) calculating a coordinate point set S of the small reflecting area:
wherein S represents a set of coordinate points of the small reflective area, P sub Representing a connected subset in a coordinate point set of an image in a light reflecting region of an original colonoscope image, wherein the symbol is used for calculating the number of coordinate points contained in the set, and Th 3 Representing a third preset intensity threshold.
Optionally, the third preset intensity threshold is set to 1000.
S104: and restoring the coordinate point of the small reflecting area to obtain a small reflecting area restored image.
In a possible implementation, S104 specifically includes:
s1041: scanning the original colonoscope image from top to bottom and from left to right;
s1042: and under the condition that the scanning pixel point coordinates belong to the small reflecting area coordinate point set, filling the scanning pixel points, and taking the filled scanning pixel points as sub-normal pixel points.
Wherein, S1042 specifically includes:
S1042A: with the scanning pixel point as the center, expanding the search area layer by layer in a shape like a Chinese character 'hui', wherein the first layer search area is an 8-neighborhood of the scanning pixel point, and the second layer search area is an outer expansion circle of the first layer search area;
S1042B: under the condition that K normal/secondary normal pixel points are searched, calculating the color vector mean value of the K normal/secondary normal pixel points, and filling the color vector mean value to the pixel point to be filled which is scanned currently.
Optionally, the preset value of K is 6.
It should be noted that, by filling the scanning pixel points belonging to the coordinate point set range of the small reflection region, the color domain effect of the pixel points in the small reflection region and the surrounding field is close, and the display effect deviation between the restored image of the small reflection region and the real image is smaller.
S105: selecting a source image and a target domain image, and extracting a light reflecting region and a dark region by fusing a multi-feature space, wherein the source image is an original colonoscope image or a small light reflecting region recovery image.
In a possible implementation, S105 specifically includes:
s1051: the method for calculating the coordinate point set of the dark region of the source image and the dark region of the target image comprises the following steps:
P d ＝{x|x∈D，G(x)＜Th 4 equation 3
Wherein, P d Representing the acquired dark region coordinate point set, D is the source image full-map coordinate point set, G (x) is the gray intensity value at the source image coordinate x, wherein Th 4 Is a fourth preset intensity value;
the method for calculating the coordinate point set of the dark region of the target domain image comprises the following steps:
P d ＝{x|x∈D，G(x)＜Th 4 equation 4
Wherein, P d Representing the acquired coordinate point set of the dark region, D is the coordinate point of the whole graph of the target domain imageSet, G (x) is the gray scale intensity value at the target domain image coordinate x, where Th 4 Is the fourth preset intensity value.
It should be noted that the process of extracting the light reflection regions of the source image and the target domain image is the same as the extraction method of the light reflection regions of the original colonoscope image, and the extracted target is the coordinate set of the corresponding source image or target domain image.
S106: converting the source image and the target domain image into an LAB color space to obtain component characteristics of the source image and component characteristics of the target domain image, wherein the resolution of the target domain image is higher than that of the source image and has the same sample as that of the source image, and the component characteristics comprise an L component intensity value, an A component intensity value and a B component intensity value.
It should be noted that, the number of pixels of the target domain image sample is N times of the number of pixels of the source image sample,
s107: and calculating the standard deviation and the mean value of each color channel of the image area except the light reflecting area and the dark area in the source image according to the component characteristics of the source image and the component characteristics of the target area image, and calculating the standard deviation and the mean value of each color channel of the image area except the light reflecting area and the dark area in the target area image.
In a possible implementation, S107 specifically includes:
s1071: selecting an original colonoscope image or a small reflecting area recovery image, wherein the calculation methods of the source image variance and the source image mean value respectively comprise the following steps: standard deviation of each color channel of image area except for light reflecting area and dark color area in source image
Standard deviation of source image:
the sets L (x), A (x) and B (x) respectively represent an L component intensity value, an A component intensity value and a B component intensity value of an LAB color space of a corresponding pixel point set, the sets sigma (L (x)), sigma (A (x)) and sigma (B (x)) respectively represent variances of the calculation sets L (x), A (x) and B (x), the sets sigma (L (x)), sigma (A (x)) and sigma (B (x)) respectively form an L component intensity value, an A component intensity value and a B component intensity value of each color channel variance of a source image, and x is all pixel points except a light reflecting region and a dark color region in the source image;
mean value of source image:
wherein μ (L (x)), μ (a (x)) and μ (B (x)) respectively represent the mean values of the computation sets L (x), a (x), B (x), x represents a pixel point in the source image excluding both the source image light reflection region and the source image dark region, and the sets μ (L (x)), μ (a (x)) and μ (B (x)) respectively constitute an L component intensity value, an a component intensity value and a B component intensity value of the mean value of each color channel of the source image;
s1072: selecting N target domain images, wherein the target domain image variance and the target domain image mean are calculated respectively;
target domain image standard deviation:
wherein, x represents a pixel point in the target domain image for removing the target domain image reflection area and the target domain image dark area, the sets σ (L (x)), σ (A (x)) and σ (B (x)) respectively form an L component intensity value, an A component intensity value and a B component intensity value of each color channel variance of the target domain image, and D i 、P i And P di Respectively representing a full-graph coordinate point set of an ith target domain image, a coordinate point set of a reflection region image of the ith target domain image and a coordinate point set of a dark region image of the ith target domain image;
target domain image mean:
wherein, the sets μ (L (x)), μ (a (x)), and μ (B (x)) constitute an L component intensity value, an a component intensity value, and a B component intensity value, respectively, of the target domain image color channel mean value.
S108: and transferring the color domain of the source image to a target domain corresponding to the target domain image according to the standard deviation and the mean of the source image and the standard deviation and the mean of the target domain image to obtain a transferred image.
In a possible implementation manner, S108 is specifically:
where T (x) represents the LAB color space L, A, and B component intensity values of the migration image at the migration image coordinates x, σ 0 ,、σ 1 And σ 2 Respectively representing the L component intensity value, A component intensity value and B component intensity value, mu, of each color channel variance of the source image 0 、μ 1 And mu 2 Respectively representing the L component intensity value, A component intensity value and B component intensity value, sigma, of the mean value of each color channel of the source image M0 ,、σ M1 And σ M2 An L component intensity value, an A component intensity value and a B component intensity value, mu, respectively representing the variance of each color channel of the target domain image M0 、μ M1 And mu M2 And the L component intensity value, the A component intensity value and the B component intensity value respectively represent the average value of each color channel of the target domain image.
Background pixels in the target domain image and the source image do not belong to any object category, and color domain migration is mainly performed on sample data in the migration process.
It should be noted that, before the color domain of the source image is migrated to the target domain corresponding to the target domain image, the light reflecting region and the dark region are excluded, so that the influence of the imaging region of the special image on the obtained migrated image in the image migration process can be effectively avoided.
Aiming at a colonoscope scene, the imaging characteristic of a colonoscope image is fully considered, a data set is enriched mainly through color domain directional migration, the characteristic that random color change is separated from actual imaging is avoided, and a foundation is laid for obtaining a more stable and reliable colonoscope image intelligent analysis model.
S109: and restoring the dark color area of the transferred image to obtain a color area transfer enhanced image.
It should be noted that, restoring the migration image, that is, filling pixel points in the region belonging to the dark region coordinate point set in the migration image.
Optionally, in the filling process, after the 20 th normal or sub-normal pixel point is searched, the color vector mean value of the searched 20 pixel points is calculated, and the color vector mean value is filled to the scanning pixel point to be filled.
S110: and forming an enhanced image data set by the original colonoscope image, the small reflection area restored image and the color domain migration enhanced image.
Wherein the enhanced image data set is used for subsequent image analysis.
It should be noted that the acquired original colonoscope image, and the small reflection region restored image and the color domain enhanced image acquired based on the original colonoscope image can provide a rich data set for reference in artificial intelligence analysis of the lesions of the colonoscope image.
In the embodiment of the invention, gray information and saturation information of a plurality of colonoscope original colonoscope images are fused, a small reflecting area coordinate point set is extracted by presetting an intensity threshold value, then the small reflecting area coordinate point set is recovered to obtain a small reflecting area recovery image, data enhancement is carried out on the original colonoscope images, the data volume of the colonoscope image enhancement data set is increased, in addition, the original colonoscope images and the small reflecting area recovery image are selected to carry out multi-feature fusion with a target domain image, a reflecting area and a dark area are extracted and converted into an LAB color space, the original colonoscope images and the small reflecting area recovery image are subjected to color domain migration to the target domain image, then the dark area recovery is carried out on the target domain image to obtain a color domain migration enhanced image, and the data volume of the colonoscope enhanced image data set is further expanded. The obtained colonoscope enhanced image data set provides richer reference image data for artificial intelligence analysis of the colonoscope image, and the stability and reliability of the analysis result are improved.
The above description is only an example of the present invention and is not intended to limit the present invention. Various modifications and alterations to this invention will become apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.
Claims (10)
1. A method of data enhancement of a colonoscope image, comprising:
s101: acquiring a plurality of original colonoscope images, and converting the color space of each original colonoscope image into a gray scale space and a hue-saturation-brightness color space respectively to obtain gray scale information and saturation information of the original colonoscope images;
s102: fusing the gray information and the saturation information, and obtaining a reflection region coordinate point set of the original colonoscope image according to a first preset intensity threshold and a second preset intensity threshold;
s103: calculating a coordinate point set of a small light reflecting area from the coordinate point set of the light reflecting area according to a third preset intensity threshold value;
s104: restoring the coordinate point of the small reflecting area to obtain a small reflecting area restored image;
s105: selecting a source image and a target domain image, and extracting a light reflecting region and a dark region by fusing a multi-feature space, wherein the source image is the original colonoscope image or the small light reflecting region recovery image;
s106: converting the source image and the target domain image into an LAB color space to obtain component features of the source image and component features of the target domain image, wherein the resolution of the target domain image is higher than that of the source image and has the same sample as that of the source image, and the component features comprise an L component intensity value, an A component intensity value and a B component intensity value;
s107: calculating the standard deviation and the mean value of each color channel of the image area except the light reflecting area and the dark area in the source image according to the component characteristics of the source image and the component characteristics of the target area image, and calculating the standard deviation and the mean value of each color channel of the image area except the light reflecting area and the dark area in the target area image;
s108: according to the standard deviation and the mean value of the source image and the standard deviation and the mean value of the target domain image, migrating the color domain of the source image to a target domain corresponding to the target domain image to obtain a migrated image;
s109: restoring the dark color area of the transferred image to obtain a color domain transfer enhanced image;
s110: and combining the original colonoscope image, the small reflection region restored image and the color domain migration enhanced image into a colonoscope enhanced image data set, wherein the colonoscope enhanced image data set is used for subsequent colonoscope image analysis.
2. The image data enhancement method according to claim 1, wherein the S101 is specifically:
s1011: acquiring colonoscope video data;
s1012: and performing frame truncation on the colonoscope video data to obtain a plurality of original colonoscope images.
3. The image data enhancement method according to claim 1, wherein the S102 specifically includes:
s1021: obtaining a set of coordinates P of the retroreflective regions of the original colonoscope image according to the following formula 1:
P＝{x|x∈D，G(x)＞Th 1 &S(x)＜Th 2 formula 1
Where P represents the set of light reflection region coordinate points, D represents the set of original colonoscope image coordinate points, G (x) represents the grayscale intensity value of the grayscale space at coordinate x in the original colonoscope image, S (x) represents the intensity value of the saturation space at coordinate x in the original colonoscope image, where Th is 1 Representing said first predetermined intensity threshold, th 2 Representing said second predetermined intensity threshold, sign&Representing a logical and operation.
4. The image data enhancement method according to claim 1, wherein the S103 specifically includes:
s1031: and (3) calculating a coordinate point set S of the small reflecting area:
wherein S represents a set of coordinate points of the small reflective area, P sub Representing a connected subset in a coordinate point set of a reflection region image of an original colonoscope image, wherein the symbol is used for calculating the number of coordinate points contained in the set, and Th 3 Representing a third preset intensity threshold.
5. The method according to claim 1, wherein the S104 specifically includes:
s1041: scanning the original colonoscope image from top to bottom and from left to right;
s1042: and under the condition that the scanning pixel point coordinates belong to the small reflecting area coordinate point set, filling the scanning pixel points, and taking the filled scanning pixel points as sub-normal pixel points.
6. The image data enhancement method according to claim 5, wherein the S1042 specifically includes:
S1042A: with the scanning pixel point as a center, expanding search areas layer by layer in a shape like a Chinese character 'hui', wherein a first layer of search area is an 8-neighborhood of the scanning pixel point, and a second layer of search area is an outer expansion circle of the first layer of search area;
S1042B: under the condition that K normal/sub-normal pixel points are searched, calculating the color vector mean value of the K normal/sub-normal pixel points, and filling the color vector mean value into the pixel point to be filled which is scanned currently.
7. The image data enhancement method of claim 6, wherein the preset value of K is 6.
8. The image data enhancement method according to claim 1, wherein the S105 specifically includes:
s1051: the method for calculating the coordinate point set of the source image dark region and the target region image dark region comprises the following steps:
P d ＝{x|x∈D，G(x)＜Th 4 equation 3
Wherein, P d Representing the acquired dark region coordinate point set, D is the source image full-map coordinate point set, G (x) is the gray intensity value at the source image coordinate x, wherein Th 4 Setting the fourth preset intensity value as the fourth preset intensity value;
the method for calculating the coordinate point set of the dark region of the target domain image comprises the following steps:
P d ＝{x|x∈D，G(x)＜Th 4 equation 4
Wherein, P d Representing the obtained set of dark region coordinate points, D being the set of full-scale coordinate points of the target domain image, G (x) being the gray scale intensity value at the coordinate x of the target domain image, where Th 4 And the fourth preset intensity value is obtained.
9. The image data enhancement method according to claim 8, wherein the S107 specifically includes:
s1071: selecting an original colonoscope image or a small reflecting area recovery image, wherein the calculation methods of the source image variance and the source image mean value respectively comprise the following steps: the standard deviation of each color channel of the image area except the light reflecting area and the dark color area in the source image
The standard deviation of the source image is as follows:
the sets L (x), A (x) and B (x) respectively represent an L component intensity value, an A component intensity value and a B component intensity value of an LAB color space of a corresponding pixel point set, the sets sigma (L (x)), sigma (A (x)) and sigma (B (x)) respectively represent variances of the sets L (x), A (x) and B (x), the sets sigma (L (x)), sigma (A (x)) and sigma (B (x)) respectively form an L component intensity value, an A component intensity value and a B component intensity value of each color channel variance of the source image, and x is all pixel points except a light reflecting region and a dark color region in the source image;
the source image mean value:
wherein μ (L (x)), μ (a (x)) and μ (B (x)) respectively represent mean values of calculation sets L (x), a (x), B (x), x represents a pixel point in the source image from which both the source image light reflection region and the source image dark region are removed, and the sets μ (L (x)), μ (a (x)) and μ (B (x)) respectively constitute an L component intensity value, an a component intensity value, and a B component intensity value of the mean value of each color channel of the source image;
s1072: selecting N target domain images, wherein the calculation methods of the variance of the target domain images and the mean value of the target domain images are respectively;
the standard deviation of the target domain image is as follows:
wherein x represents a pixel point in the target domain image from which a target domain image reflection area and a target domain image dark area are removed, the sets σ (L (x)), σ (a (x)) and σ (B (x)) respectively form an L component intensity value, an a component intensity value and a B component intensity value of each color channel variance of the target domain image, and D i 、P i And P di Respectively representing the ith coordinate point set of the target domain image, the ith coordinate point set of the target domain image reflection region image and the ith target domain image dark color region imageA set of image coordinate points;
the target domain image mean value:
wherein the sets μ (L (x)), μ (A (x)) and μ (B (x)) constitute an L component intensity value, an A component intensity value and a B component intensity value of the target domain image color channel mean value, respectively.
10. The image data enhancement method according to claim 9, wherein S108 specifically is:
wherein T (x) represents the LAB color space L, A and B component intensity values of the migration image at the migration image coordinates x, σ 0 ,、σ 1 And σ 2 Respectively representing the L component intensity value, the A component intensity value and the B component intensity value of each color channel variance of the source image, mu 0 、μ 1 And mu 2 Respectively representing the L component intensity value, the A component intensity value and the B component intensity value, sigma, of the mean value of each color channel of the source image M0 ,、σ M1 And σ M2 An L component intensity value, an A component intensity value and a B component intensity value, mu, respectively representing each color channel variance of the target domain image M0 、μ M1 And mu M2 And respectively representing an L component intensity value, an A component intensity value and a B component intensity value of each color channel mean value of the target domain image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211404052.6A CN115689935A (en) | 2022-11-10 | 2022-11-10 | Data enhancement method of colonoscope image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211404052.6A CN115689935A (en) | 2022-11-10 | 2022-11-10 | Data enhancement method of colonoscope image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115689935A true CN115689935A (en) | 2023-02-03 |
Family
ID=85050878
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211404052.6A Pending CN115689935A (en) | 2022-11-10 | 2022-11-10 | Data enhancement method of colonoscope image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115689935A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116580210A (en) * | 2023-07-05 | 2023-08-11 | 四川弘和数智集团有限公司 | Linear target detection method, device, equipment and medium |
-
2022
- 2022-11-10 CN CN202211404052.6A patent/CN115689935A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116580210A (en) * | 2023-07-05 | 2023-08-11 | 四川弘和数智集团有限公司 | Linear target detection method, device, equipment and medium |
| CN116580210B (en) * | 2023-07-05 | 2023-09-15 | 四川弘和数智集团有限公司 | Linear target detection method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108062525B (en) | Deep learning hand detection method based on hand region prediction | |
| JP7026826B2 (en) | Image processing methods, electronic devices and storage media | |
| CN109583345B (en) | Road recognition method, device, computer device and computer readable storage medium | |
| CN111145209B (en) | Medical image segmentation method, device, equipment and storage medium | |
| WO2021159767A1 (en) | Medical image processing method, image processing method, and device | |
| CN110570352B (en) | Image labeling method, device and system and cell labeling method | |
| TW202014984A (en) | Image processing method, electronic device, and storage medium | |
| Zhou et al. | FSAD-Net: feedback spatial attention dehazing network | |
| CN108830149B (en) | Target bacterium detection method and terminal equipment | |
| WO2023137914A1 (en) | Image processing method and apparatus, electronic device, and storage medium | |
| JP7076681B2 (en) | Image processing methods and equipment, and training methods for neural network models | |
| CN110956681B (en) | Portrait background automatic replacement method combining convolution network and neighborhood similarity | |
| CN113313680B (en) | Colorectal cancer pathological image prognosis auxiliary prediction method and system | |
| CN112581370A (en) | Training and reconstruction method of super-resolution reconstruction model of face image | |
| CN112102929A (en) | Medical image labeling method and device, storage medium and electronic equipment | |
| CN113793301A (en) | Training method of fundus image analysis model based on dense convolution network model | |
| Zhu et al. | Towards automatic wild animal detection in low quality camera-trap images using two-channeled perceiving residual pyramid networks | |
| CN115689935A (en) | Data enhancement method of colonoscope image | |
| CN110473176B (en) | Image processing method and device, fundus image processing method and electronic equipment | |
| CN112070041B (en) | Living body face detection method and device based on CNN deep learning model | |
| CN115272378A (en) | Character image segmentation method based on characteristic contour | |
| CN108596893B (en) | Image processing method and system | |
| CN113012167A (en) | Combined segmentation method for cell nucleus and cytoplasm | |
| CN108776815A (en) | A kind of conspicuousness object detection method based on depth characteristic | |
| CN116612385B (en) | Remote sensing image multiclass information extraction method and system based on depth high-resolution relation graph convolution |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |