CN115170808A - Image segmentation method and system - Google Patents
Image segmentation method and system Download PDFInfo
- Publication number
- CN115170808A CN115170808A CN202211079177.6A CN202211079177A CN115170808A CN 115170808 A CN115170808 A CN 115170808A CN 202211079177 A CN202211079177 A CN 202211079177A CN 115170808 A CN115170808 A CN 115170808A
- Authority
- CN
- China
- Prior art keywords
- image
- layer
- convolution
- coding
- segmentation
- 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
- 238000003709 image segmentation Methods 0.000 title claims abstract description 43
- 238000000034 method Methods 0.000 title claims abstract description 27
- 230000011218 segmentation Effects 0.000 claims abstract description 31
- 238000000605 extraction Methods 0.000 claims abstract description 23
- 238000010586 diagram Methods 0.000 claims abstract description 16
- 238000012545 processing Methods 0.000 claims abstract description 14
- 230000009466 transformation Effects 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims abstract description 7
- 230000004913 activation Effects 0.000 claims description 12
- 238000013507 mapping Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 9
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
- G06V10/449—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters
- G06V10/451—Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters with interaction between the filter responses, e.g. cortical complex cells
- G06V10/454—Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Abstract
The invention relates to an image segmentation method and an image segmentation system, which comprise the following steps: s1, acquiring an image to be processed, and performing data enhancement on the image; s2, conveying the enhanced image to an end-to-end model for model training to obtain a target segmentation image; the end-to-end model comprises an encoding layer and a decoding layer, wherein the encoding layer comprises an upper convolutional layer and a lower convolutional layer; the upper convolution layer is used for carrying out feature extraction on the enhanced image to obtain a first feature map; the lower convolution layer is used for carrying out size transformation on the image subjected to feature extraction on the upper convolution layer to obtain a second feature map; merging the first characteristic diagram and the second characteristic diagram and transmitting the merged characteristic diagram to a decoding layer; and the decoding layer is used for carrying out segmentation convolution on the combined feature graph to obtain a target segmentation image. The image segmentation method and the image segmentation system improve the final accuracy of image segmentation to a certain extent, and have higher processing efficiency.
Description
Technical Field
The invention relates to the field of image segmentation and data analysis, in particular to an image segmentation method and an image segmentation system.
Background
At present, the known image segmentation is performed by various convolution networks, full convolution networks or attention networks, the segmentation effect cannot meet the requirements, more errors are caused to the hair filaments in the portrait, and meanwhile, due to the uncertainty in the portrait image acquisition process, such as the inconsistency of shooting distance, angle and shooting posture, the problems of partial non-standard portrait angle, lack of contrast and definition of background and portrait and the like often exist in the portrait data set.
The existing image segmentation method can work well under the conditions of large data set, large model and large parameters. However, these image segmentation methods have disadvantages when there is less data and the device for training the model is crude, and therefore, it is necessary to provide a solution to this problem.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an image segmentation method and an image segmentation system.
In order to achieve the object of the present invention, the present invention provides an image segmentation method, comprising the steps of:
s1, acquiring an image to be processed, and performing data enhancement on the image;
s2, conveying the enhanced image to an end-to-end model for model training to obtain a target segmentation image;
the end-to-end model comprises a coding layer and a decoding layer, wherein the coding layer comprises an upper convolution layer and a lower convolution layer;
the upper convolution layer is used for carrying out feature extraction on the enhanced image to obtain a first feature map;
the lower convolution layer is used for carrying out size transformation on the image subjected to feature extraction on the upper convolution layer to obtain a second feature map; merging the first characteristic diagram and the second characteristic diagram and transmitting the merged first characteristic diagram and second characteristic diagram to a decoding layer;
and the decoding layer is used for carrying out segmentation convolution on the combined feature graph to obtain a target segmentation image.
Preferably, the specific steps of obtaining the first characteristic diagram by the upper convolution layer are as follows:
and performing feature extraction on the enhanced image by adopting multiple times of 3x3 convolution operation and relu activation function operation to obtain a first feature map.
Preferably, the image used for the scale transformation of the lower convolution layer is: and (4) images after the first 3x3 convolution operation and relu activation function operation are carried out in the upper convolution layer.
Preferably, the specific steps of performing the scale transformation on the lower convolution layer are as follows:
and carrying out full convolution operation on the image subjected to feature extraction on the upper convolution layer, then conveying the image subjected to full convolution operation to a tandform coding block for multiple times of coding, and then conveying the image subjected to repeated coding to a reshape for size conversion of the image.
Preferably, the specific steps of performing full convolution operation on the image subjected to feature extraction on the upper convolution layer, and then transmitting the image subjected to full convolution operation to the tansformer coding block for multiple times of coding include:
and transmitting the image subjected to feature extraction by the upper convolution layer to a high feature layer for convolution operation, then performing one-dimensional vector mapping on the image subjected to convolution operation by the high feature layer through a Linear projection layer to obtain Embedding, and transmitting the Embedding to a transform coding block for multiple times of coding.
Preferably, the upper convolutional layer is subjected to CNN convolutional coding and downsampling, the lower convolutional layer is subjected to transform coding and downsampling, and the decoding layer is subjected to upsampling on the combined image.
Preferably, the specific step of enhancing the image data in step S1 is:
the image is rotated by taking the coordinate axis as a center, and the rotated image is subjected to horizontal, vertical and mirror image inversion to perform data enhancement.
Preferably, the present invention further provides an image segmentation system, comprising:
an image processing module: acquiring an image to be processed, and performing data enhancement on the image to be processed;
the coding module: the device comprises an upper layer coding module and a lower layer coding module;
an upper layer coding module: the image processing device is used for extracting features of the enhanced image to obtain a first feature map;
a lower layer coding module: carrying out size transformation on the image subjected to feature extraction on the upper convolution layer to obtain a second feature map, merging the first feature map and the second feature map, and conveying the merged first feature map and the merged second feature map to a decoding layer;
a decoding module: and carrying out segmentation convolution on the combined feature graph to obtain a target segmentation image.
Preferably, the size-transformed image in the lower layer coding module is:
and the upper layer coding module performs the image after the first 3x3 convolution operation and the relu activation function operation.
Preferably, the upper layer encoding module performs downsampling on the image, the lower layer encoding module performs downsampling on the image, and the decoding module performs upsampling on the combined image.
The invention has the beneficial effects that: according to the image segmentation method and system provided by the invention, data enhancement is firstly carried out on the image to be processed, and the target segmentation image is obtained by carrying out multi-layer processing on the image, so that the final accuracy of image segmentation is improved to a certain extent, and meanwhile, the processing efficiency is higher.
Drawings
The foregoing and other objects, features and advantages of the invention will be apparent from the following more particular description of preferred embodiments of the invention, as illustrated in the accompanying drawings. Like reference numerals refer to like parts throughout the drawings, and the drawings are not intended to be drawn to scale in actual dimensions, emphasis instead being placed upon illustrating the principles of the invention.
Fig. 1 is a schematic specific flowchart of an encoding layer and a decoding layer in an end-to-end model according to an embodiment of the present invention;
fig. 2 is a schematic flowchart of an image segmentation method according to an embodiment of the present invention.
Detailed Description
The present invention will be better understood and implemented by those skilled in the art by the following detailed description of the embodiments taken in conjunction with the accompanying drawings, which are not intended to limit the scope of the present invention.
Referring to fig. 1-2, an embodiment of the invention provides an image segmentation method, including the following steps:
s1, acquiring an image to be processed, and performing data enhancement on the image;
s2, conveying the enhanced image to an end-to-end model for model training to obtain a target segmentation image;
the end-to-end model comprises a coding layer and a decoding layer, wherein the coding layer comprises an upper convolution layer and a lower convolution layer;
the upper convolution layer is used for carrying out feature extraction on the enhanced image to obtain a first feature map;
the lower convolution layer is used for carrying out size transformation on the image subjected to feature extraction on the upper convolution layer to obtain a second feature map; merging the first characteristic diagram and the second characteristic diagram and transmitting the merged characteristic diagram to a decoding layer;
and the decoding layer is used for carrying out segmentation convolution on the combined feature graph to obtain a target segmentation image.
Referring to fig. 1-2, the image segmentation method provided by the present invention performs a multi-layer processing on an image through upper and lower convolution layers, and combines the processed images of the upper and lower convolution layers and transmits them to a decoding layer for image segmentation, thereby achieving a final segmented image.
The invention has the beneficial effects that: the data of the image to be processed is enhanced, and the target segmentation image is obtained by performing multi-layer processing on the image, so that the final accuracy of image segmentation is improved to a certain extent, and the processing efficiency is higher.
Referring to fig. 1-2, in a preferred embodiment, the specific steps of obtaining the first characteristic diagram by the upper convolution layer are as follows:
and performing feature extraction on the enhanced image by adopting multiple times of 3x3 convolution operation and relu activation function operation to obtain a first feature map.
In general, the enhanced image may be subjected to a triple 3 × 3 convolution operation and a relu activation function operation.
Referring to fig. 1-2, in a preferred embodiment, the images used for scaling the lower convolution layer are: and (4) images after the first 3x3 convolution operation and relu activation function operation are carried out in the upper convolution layer.
The lower convolutional layer performs an image after performing the first 3x3 convolution and relu activation function operations mainly using the upper convolutional layer.
Referring to fig. 1-2, in a further preferred embodiment, the scaling of the lower convolutional layer comprises the following steps:
and carrying out full convolution operation on the image subjected to feature extraction on the upper convolution layer, then conveying the image subjected to full convolution operation to a tandform coding block for multiple times of coding, and then conveying the image subjected to repeated coding to a reshape for size conversion of the image.
Referring to fig. 1-2, in a further preferred embodiment, the specific steps of performing a full convolution operation on the image after feature extraction on the upper convolution layer, and then transferring the image after the full convolution operation to a tandform coding block for multiple coding are as follows:
and transmitting the image subjected to feature extraction by the upper convolution layer to a high feature layer for convolution operation, then performing one-dimensional vector mapping on the image subjected to convolution operation by the high feature layer through a Linear projection layer to obtain Embedding, and transmitting the Embedding to a transform coding block for multiple times of coding.
Referring to fig. 1, in the lower convolutional Layer, an image obtained by performing a first 3x3 convolution operation and a relu activation function operation on the upper convolutional Layer is input into a high feature Layer (the high feature Layer is a 3x3 convolutional Layer), the image obtained by performing the convolution operation on the high feature Layer is subjected to one-dimensional vector mapping through a Linear projection Layer (the Linear projection is used for mapping a feature Layer into a one-dimensional vector), so as to obtain embed (which refers to a one-dimensional characteristic feature of a feature map), and the embed is transmitted to a transform coding block for triple coding, where the transform coding block is a module composed of Layer Norm, MSA and MLP; (the same transform coding block is used for the third-time coding, the first-time output is used as the second-time input, the second-time output is used as the third-time input, and the third-time output is used as the final output). After the transform coding, the Hidden feature is required to be reshaped, the output image of the transform coding block is transformed into an image with the same size as the first feature map size (parameters such as resolution and size) through reshape, and then the first feature map and the second feature map are merged. And sending the combined feature image to a decoding layer, wherein the first layer of the decoding layer firstly performs segmentation convolution (the segmentation convolution refers to deconvolution, and the segmentation convolution enlarges the feature image to a fixed size) to amplify the feature image by 2 times, then performs the second layer of segmentation convolution, enlarges the feature image by 4 times, finally performs the third layer of segmentation convolution, amplifies the feature image by 8 times, and the image after the three times of segmentation convolution is the target segmentation image.
Referring to fig. 1-2, in the preferred embodiment, the upper convolutional layer is encoded and downsampled using CNN convolutional coding, the lower convolutional layer is encoded and downsampled using transform, and the decoding layer upsamples the combined image. Here, the up-sampling refers to enlarging an image, and the down-sampling refers to reducing an image.
According to the image segmentation method provided by the invention, a Transformer is adopted, and the region of interest extracted by the region detection network is used as attention information, so that the network is more focused on regions which are possibly portraits, and the interference of background information in the images is reduced. The core idea is that all the outputs of the encoder are weighted and combined and then input into the decoder at the current position to influence the output of the decoder. By weighting the output of the encoder, more context information of the original data can be utilized while realizing the alignment of the input and the output;
the image segmentation method provided by the invention utilizes the combination of the transformer and the CNN network to carry out segmentation, removes image parts except in the portrait picture, and particularly, the image segmentation only interests in the region of the portrait, the face and the upper part of the body. Dividing a coding layer of an end-to-end model into two parts, wherein the first part is coded and downsampled by using a transform, the second part is downsampled by using CNN convolution, then the two parts of codes are combined and upsampled and then conveyed to a decoding layer, when the image to be processed is processed, parameters of a network model are modified, data enhancement is carried out on original data, the data have diversity, overfitting is avoided, and finally image segmentation is accelerated;
referring to fig. 1-2, in a preferred embodiment, the specific steps of enhancing the image data in step S1 are as follows:
the image is rotated by taking the coordinate axis as a center, and the rotated image is subjected to horizontal, vertical and mirror image inversion to perform data enhancement.
The purpose of data enhancement is to obtain better training results, and the rotation angles taking coordinate axes as the center are 0 degrees, 90 degrees, 180 degrees and 270 degrees.
In the preferred embodiment, noise disturbance is added to the image after horizontal, vertical and mirror image inversion, that is, the whole image is fully paved with random white or black pixel points for adding Gaussian noise, so that high-frequency characteristics are effectively inhibited, the influence of the high-frequency characteristics on the model is weakened, and the learning capability of the model is improved.
The image segmentation method provided by the invention mainly adopts a Deft-TransformarmerCNN neural network model method to segment the portrait image, and is mainly applied to face analysis and scene analysis. Meanwhile, certain preprocessing is carried out on the data before training, including rotation, translation, noise addition and the like, the diversity of the data is kept, and model overfitting is avoided.
Referring to fig. 1-2, in a further preferred embodiment, the present invention further provides an image segmentation system, including:
an image processing module: acquiring an image to be processed, and performing data enhancement on the image to be processed;
the coding module: the device comprises an upper layer coding module and a lower layer coding module;
an upper layer coding module: the image processing device is used for extracting features of the enhanced image to obtain a first feature map;
a lower layer coding module: carrying out size transformation on the image subjected to feature extraction on the upper convolution layer to obtain a second feature map, merging the first feature map and the second feature map, and conveying the merged first feature map and the merged second feature map to a decoding layer;
a decoding module: and carrying out segmentation convolution on the combined feature map to obtain a target segmentation image.
Referring to fig. 1-2, in the preferred embodiment, the size-transformed images in the lower layer coding module are:
and performing the image after the first 3x3 convolution operation and relu activation function operation in the upper layer coding module.
Referring to fig. 1-2, in a preferred embodiment, an upper layer encoding module performs a downsampling operation on an image, a lower layer encoding module performs a downsampling operation on the image, and a decoding module performs an upsampling (segmentation convolution) operation on a combined image.
The invention has the beneficial effects that: the invention provides an image segmentation method and an image segmentation system, which are characterized in that data enhancement is firstly carried out on an image to be processed, and a target segmentation image is obtained by carrying out multi-layer processing on the image, so that the final accuracy of image segmentation is improved to a certain extent, and meanwhile, the processing efficiency is higher.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (10)
1. An image segmentation method, characterized by comprising the steps of:
s1, acquiring an image to be processed, and performing data enhancement on the image;
s2, conveying the enhanced image to an end-to-end model for model training to obtain a target segmentation image;
the end-to-end model comprises an encoding layer and a decoding layer, wherein the encoding layer comprises an upper convolutional layer and a lower convolutional layer;
the upper convolution layer is used for carrying out feature extraction on the enhanced image to obtain a first feature map;
the lower convolution layer is used for carrying out size transformation on the image subjected to feature extraction on the upper convolution layer to obtain a second feature map; merging the first characteristic diagram and the second characteristic diagram and transmitting the merged first characteristic diagram and second characteristic diagram to a decoding layer;
and the decoding layer is used for carrying out segmentation convolution on the combined feature graph to obtain a target segmentation image.
2. The image segmentation method according to claim 1, wherein the step of obtaining the first feature map by the upper convolution layer comprises:
and performing feature extraction on the enhanced image by adopting multiple times of 3x3 convolution operation and relu activation function operation to obtain a first feature map.
3. The image segmentation method according to claim 1, wherein the image used for scaling the lower convolution layer is: and (4) images after the first 3x3 convolution operation and relu activation function operation are carried out in the upper convolution layer.
4. The image segmentation method of claim 1, wherein the down-convolution layer is scaled by:
and carrying out full convolution operation on the image subjected to feature extraction on the upper convolution layer, then conveying the image subjected to full convolution operation to a tandform coding block for multiple times of coding, and then conveying the image subjected to repeated coding to a reshape for size conversion of the image.
5. The image segmentation method according to claim 4, wherein the step of performing full convolution on the image after feature extraction on the upper convolution layer and then transferring the image after full convolution to a tandsformer coding block for multiple times of coding comprises:
and transmitting the image subjected to feature extraction by the upper convolution layer to a high feature layer for convolution operation, then performing one-dimensional vector mapping on the image subjected to convolution operation by the high feature layer through a Linear projection layer to obtain Embedding, and transmitting the Embedding to a transform coding block for multiple times of coding.
6. The image segmentation method of claim 1, wherein the upper convolutional layer employs CNN convolutional coding and downsampling, the lower convolutional layer employs transform for coding and downsampling, and the decoding layer performs upsampling on the combined image.
7. The image segmentation method according to claim 1, wherein the step S1 of enhancing the image data comprises the following steps:
the image is rotated by taking the coordinate axis as a center, and the rotated image is subjected to horizontal, vertical and mirror image inversion to perform data enhancement.
8. An image segmentation system, comprising:
an image processing module: acquiring an image to be processed, and performing data enhancement on the image to be processed;
the coding module: the device comprises an upper layer coding module and a lower layer coding module;
an upper layer coding module: the image processing device is used for extracting features of the enhanced image to obtain a first feature map;
and a lower layer coding module: carrying out size transformation on the image subjected to feature extraction on the upper convolution layer to obtain a second feature map, merging the first feature map and the second feature map, and conveying the merged first feature map and the merged second feature map to a decoding layer;
a decoding module: and carrying out segmentation convolution on the combined feature graph to obtain a target segmentation image.
9. The image segmentation system as set forth in claim 1, wherein the size-transformed image in the lower layer coding module is:
and the upper layer coding module performs the image after the first 3x3 convolution operation and the relu activation function operation.
10. The image segmentation system of claim 1 wherein the upper layer encoding module performs a downsampling operation on the image, the lower layer encoding module performs a downsampling operation on the image, and the decoding module performs an upsampling operation on the combined image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211079177.6A CN115170808A (en) | 2022-09-05 | 2022-09-05 | Image segmentation method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211079177.6A CN115170808A (en) | 2022-09-05 | 2022-09-05 | Image segmentation method and system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115170808A true CN115170808A (en) | 2022-10-11 |
Family
ID=83480570
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211079177.6A Pending CN115170808A (en) | 2022-09-05 | 2022-09-05 | Image segmentation method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115170808A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116205928A (en) * | 2023-05-06 | 2023-06-02 | 南方医科大学珠江医院 | Image segmentation processing method, device and equipment for laparoscopic surgery video and medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114066902A (en) * | 2021-11-22 | 2022-02-18 | 安徽大学 | Medical image segmentation method, system and device based on convolution and transformer fusion |
US20220059200A1 (en) * | 2020-08-21 | 2022-02-24 | Washington University | Deep-learning systems and methods for medical report generation and anomaly detection |
CN114359557A (en) * | 2021-12-10 | 2022-04-15 | 广东电网有限责任公司 | Image processing method, system, equipment and computer medium |
CN114419054A (en) * | 2022-01-19 | 2022-04-29 | 新疆大学 | Retinal blood vessel image segmentation method and device and related equipment |
CN114841320A (en) * | 2022-05-07 | 2022-08-02 | 西安邮电大学 | Organ automatic segmentation method based on laryngoscope medical image |
-
2022
- 2022-09-05 CN CN202211079177.6A patent/CN115170808A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20220059200A1 (en) * | 2020-08-21 | 2022-02-24 | Washington University | Deep-learning systems and methods for medical report generation and anomaly detection |
CN114066902A (en) * | 2021-11-22 | 2022-02-18 | 安徽大学 | Medical image segmentation method, system and device based on convolution and transformer fusion |
CN114359557A (en) * | 2021-12-10 | 2022-04-15 | 广东电网有限责任公司 | Image processing method, system, equipment and computer medium |
CN114419054A (en) * | 2022-01-19 | 2022-04-29 | 新疆大学 | Retinal blood vessel image segmentation method and device and related equipment |
CN114841320A (en) * | 2022-05-07 | 2022-08-02 | 西安邮电大学 | Organ automatic segmentation method based on laryngoscope medical image |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116205928A (en) * | 2023-05-06 | 2023-06-02 | 南方医科大学珠江医院 | Image segmentation processing method, device and equipment for laparoscopic surgery video and medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112149619B (en) | Natural scene character recognition method based on Transformer model | |
Zhao et al. | Simultaneous color-depth super-resolution with conditional generative adversarial networks | |
Kim et al. | Progressive face super-resolution via attention to facial landmark | |
CN110415172B (en) | Super-resolution reconstruction method for face area in mixed resolution code stream | |
WO2021047471A1 (en) | Image steganography method and apparatus, and image extraction method and apparatus, and electronic device | |
CN110580680B (en) | Face super-resolution method and device based on combined learning | |
CN114170174B (en) | CLANet steel rail surface defect detection system and method based on RGB-D image | |
CN109712071A (en) | Unmanned plane image mosaic and localization method based on track constraint | |
Niu et al. | Effective image restoration for semantic segmentation | |
CN115170808A (en) | Image segmentation method and system | |
CN112258436A (en) | Training method and device of image processing model, image processing method and model | |
CN115631107A (en) | Edge-guided single image noise removal | |
CN116957931A (en) | Method for improving image quality of camera image based on nerve radiation field | |
CN115953582A (en) | Image semantic segmentation method and system | |
Akutsu et al. | Ultra low bitrate learned image compression by selective detail decoding | |
CN117274059A (en) | Low-resolution image reconstruction method and system based on image coding-decoding | |
CN116823908A (en) | Monocular image depth estimation method based on multi-scale feature correlation enhancement | |
CN116630369A (en) | Unmanned aerial vehicle target tracking method based on space-time memory network | |
CN115861048A (en) | Image super-resolution method, device, equipment and storage medium | |
Jia et al. | Deep convolutional network based image quality enhancement for low bit rate image compression | |
CN108364258B (en) | Method and system for improving image resolution | |
CN115063685B (en) | Remote sensing image building feature extraction method based on attention network | |
CN115984714B (en) | Cloud detection method based on dual-branch network model | |
CN113674369B (en) | Method for improving G-PCC compression by deep learning sampling | |
Lei et al. | Aparecium: Revealing Secrets from Physical Photographs |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20221011 |
|
RJ01 | Rejection of invention patent application after publication |