WO2023179095A1 - Procédé et appareil de segmentation d'image, équipement terminal et support d'enregistrement - Google Patents

Procédé et appareil de segmentation d'image, équipement terminal et support d'enregistrement Download PDF

Info

Publication number
WO2023179095A1
WO2023179095A1 PCT/CN2022/137369 CN2022137369W WO2023179095A1 WO 2023179095 A1 WO2023179095 A1 WO 2023179095A1 CN 2022137369 W CN2022137369 W CN 2022137369W WO 2023179095 A1 WO2023179095 A1 WO 2023179095A1
Authority
WO
WIPO (PCT)
Prior art keywords
module
convolution
decoder
image
image segmentation
Prior art date
Application number
PCT/CN2022/137369
Other languages
English (en)
Chinese (zh)
Inventor
郭菲
马世强
唐继军
郗文辉
Original Assignee
中国科学院深圳理工大学(筹)
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 中国科学院深圳理工大学(筹) filed Critical 中国科学院深圳理工大学(筹)
Publication of WO2023179095A1 publication Critical patent/WO2023179095A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30016—Brain
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30096—Tumor; Lesion

Definitions

  • the present application relates to the field of image processing technology, and in particular, to an image segmentation method, device, terminal equipment and storage medium.
  • Image segmentation technology is a technology that divides images into several specific areas to extract specific types of targets.
  • Image segmentation technology has been widely used in remote sensing, medicine and other fields. Taking the medical field as an example, image segmentation technology can extract the regions where organs or lesions are located in medical images to assist medical experts in making corresponding medical diagnoses.
  • Embodiments of the present application provide an image segmentation method, device, terminal equipment and storage medium, which can solve the problem of inaccurate image segmentation results of existing image segmentation methods.
  • embodiments of the present application provide an image segmentation method.
  • the method includes: acquiring an image to be processed; inputting the image to be processed into a trained image segmentation model for processing, and outputting the segmented image;
  • the image segmentation model is an encoder-decoder encoder structure, and a group integration module is provided between the corresponding encoder and decoder.
  • the group integration module is used to extract m compensation information of different scales from the first input feature of the encoder, and combine m different scales of compensation information.
  • the compensation information is input to the decoder, m ⁇ 2.
  • the group integration module includes a first convolutional layer and m-1 semantic compensation modules; the first convolutional layer is used to extract first-scale compensation information from the first input feature of the encoder; m-1 semantic compensation modules are used to extract m-1 pieces of compensation information at different scales from the first scale of compensation information, where the i-th semantic compensation module among the m-1 semantic compensation modules The output is the input of the i+1th semantic compensation module, 1 ⁇ i ⁇ m-1.
  • the semantic compensation module includes a second convolutional layer, a plurality of third convolutional layers and atrous convolutional layers connected in sequence.
  • the decoder includes m group convolution modules
  • Input m compensation information of different scales into the decoder including:
  • Each group of m group convolution modules decodes a set of second input features and one scale of compensation information to obtain m groups of decoded features
  • the output features of the decoder are obtained according to m sets of decoding features.
  • the group convolution module includes an upsampling module and a first convolution module
  • a set of second input features and a scale of compensation information are decoded through each of the m group convolution modules, including:
  • a set of second input features are upsampled through the upsampling module in each group convolution module, and the output of the upsampling module is spliced with a scale of compensation information as the first convolution in the group convolution module. module input.
  • the encoder includes a second convolution module, a downsampling module, and a compression and excitation module that are connected in sequence.
  • the second convolution module includes a batch normalization layer, an excitation layer, and a convolution layer that are connected in sequence. .
  • the image to be processed is a medical image.
  • an image segmentation device including: an acquisition unit to acquire an image to be processed; a segmentation unit to input the image to be processed into a trained image segmentation model for processing, and output the segmented image, image
  • the segmentation model is an encoder-decoder structure, and there is a group integration module between the corresponding encoder and decoder.
  • the group integration module is used to extract m compensation information of different scales from the input features of the encoder, and m compensation information of different scales are input to the decoder, m ⁇ 2.
  • embodiments of the present application provide a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor.
  • the processor executes the computer program, any of the above-mentioned aspects in the first aspect are implemented. one method.
  • embodiments of the present application provide a computer-readable storage medium.
  • the computer-readable storage medium stores a computer program.
  • the computer program is executed by a processor, the method in any one of the above-mentioned first aspects is implemented.
  • embodiments of the present application provide a computer program product, which when the computer program product is run on a terminal device, causes the terminal device to execute any of the methods of the first aspect.
  • a group integration module is set between the corresponding encoder and decoder, and the group integration module extracts from the first input feature of the encoder, etc. Compensation information of multiple different scales is input to the corresponding decoder for processing.
  • the group integration module can increase the receptive field of the network layer and reduce coding through compensation information of multiple different scales.
  • the difference between the low-level edge detail information in the decoder and the high-level semantic feature information in the corresponding decoder avoids the dilution of high-level semantic feature information due to the direct fusion of low-level edge detail information and high-level semantic feature information, thus improving the model treatment Segmentation accuracy of processed images.
  • Figure 1 is a schematic structural diagram of an image segmentation model provided by an embodiment of the present application.
  • Figure 2 is another structural schematic diagram of an image segmentation model provided by an embodiment of the present application.
  • Figure 3 is a schematic structural diagram of an encoder provided by an embodiment of the present application.
  • Figure 4 is a schematic structural diagram of a compression and excitation module provided by an embodiment of the present application.
  • Figure 5 is a schematic structural diagram of a group integration module provided by an embodiment of the present application.
  • Figure 6 is a schematic structural diagram of a semantic compensation module provided by an embodiment of the present application.
  • Figure 7 is a schematic structural diagram of a group convolution module in a decoder provided by an embodiment of the present application.
  • Figure 8 is a schematic structural diagram of an image segmentation device provided by an embodiment of the present application.
  • Figure 9 is a schematic structural diagram of a terminal device provided by an embodiment of the present application.
  • this application provides an image segmentation method. After obtaining the image to be processed, the image to be processed is input into the image segmentation model based on the encoder-decoder structure provided by this application. During processing, the segmented image of the image to be processed is obtained.
  • a group integration module is provided between the corresponding encoder and decoder, and the group integration module extracts many other features from the first input feature of the encoder. Compensation information of different scales is input to the corresponding decoder, increasing the receptive field of the network layer, and reducing low-level edge detail information in the encoder and high-level semantic feature information in the corresponding decoder through compensation information of multiple different scales. The difference between them solves the problem in the existing U-Net network model that directly splices low-level edge detail information and high-level semantic feature information, causing the high-level semantic feature information to be diluted, thereby improving the model's segmentation accuracy of the image to be processed.
  • FIG. 1 is a schematic structural diagram of an image segmentation model provided by an embodiment of the present application.
  • the image segmentation model includes 2 convolutional layers, 3 encoders, and 3 decoders corresponding to the 3 encoders. Specifically, it includes: convolutional layer 1, encoder 1, encoder 2, and encoder connected in sequence. 3. Decoder 3, decoder 2, decoder 1 and convolutional layer 2.
  • convolution layer 1 includes 32 convolution kernels.
  • the input image to be processed is processed through convolution layer 1, and the number of channels of the feature map is increased, resulting in a channel number of 32 and a size of 176 ⁇ 144 feature map.
  • the convolution layer 2 includes a convolution kernel, and the feature map with a channel number of 32 and a size of 176 ⁇ 144 output by the decoder 1 is processed through the convolution layer 2 to obtain a segmented image of a size of 176 ⁇ 144.
  • each encoder includes a second convolution module, a downsampling module, and a compression and excitation (Squeeze and Excitation, SE) module connected in sequence.
  • the second convolution module is used to increase the number of channels of the feature map
  • the downsampling module is used to reduce the scale of the feature map
  • the compression and excitation module is used to recalibrate the feature map of each channel.
  • the second convolution module includes a batch normalization (Batch Normalization, BN) layer, an excitation layer and a convolution layer connected in sequence.
  • the excitation layer may be a ReLU (rectified linear unit) excitation layer.
  • the compression and excitation module includes the convolution operation F tr ( ⁇ , ⁇ ), the squeeze operation F sq ( ⁇ ), the excitation operation F ex ( ⁇ , w), and the weighting operation F scal ( ⁇ ).
  • the input feature map with the number of channels c_1 is converted into a feature map with the number of channels c_2 through a convolution operation.
  • the feature map with the number of channels c_2 is compressed through the Squeeze operation, and the feature map of each channel is converted into a real number, thereby converting the feature map with the number of feature channels c_2 into a feature vector with the dimension c_2.
  • the weight of the feature map of each channel is evaluated through the Excitation operation to obtain the weight w corresponding to each channel.
  • the weight w is weighted with the feature map with the number of feature channels c_2 through the weighting operation, so that in the channel dimension Recalibrate the original features.
  • a group integration module is provided between the corresponding encoder and decoder.
  • the group integration module is used to extract compensation information of m different scales from the input features of the encoder, and combine the m different scales
  • the compensation information is input into the decoder, m ⁇ 2. That is, as shown in Figure 1 and Figure 2, the group integration module 1 is used to extract m compensation information of different scales from the input features of the encoder 1 and input it into the decoder 1, and the group integration module 2 is used to extract the compensation information from the encoder 2 Extract m compensation information of different scales from the input features and input it into the decoder 2.
  • the group integration module 3 is used to extract m compensation information of different scales from the input features of the encoder 3 and input it into the decoder 3.
  • the group integration module includes the first convolutional layer and m-1 semantic compensation modules (Semantic Compensation Module, SCM).
  • the first convolution layer is used to extract the compensation information of the first scale from the input features of the encoder, and the first convolution layer is a convolution kernel with a size of 1 ⁇ 1.
  • m-1 semantic compensation modules are used to extract m-1 compensation information of different scales from the compensation information of the first scale, where the output of the i-th semantic compensation module among the m-1 semantic compensation modules is the i-th +1 input to the semantic compensation module, 1 ⁇ i ⁇ m-1.
  • each semantic compensation module includes one second convolutional layer, at least two third convolutional layers and an atrous (Atrous) convolutional layer connected in sequence.
  • the size of the convolution kernel of the second convolution layer is 1 ⁇ 1
  • the size of the convolution kernel of each third convolution layer is 3 ⁇ 3.
  • Atrous convolutional layers can increase the receptive field in order to eliminate the semantic gap between the low-level edge information of the encoder and the high-level semantic information of the decoder.
  • the m-1 semantic compensation modules in the group integration module include SCM_1 to SCM_m-1.
  • the first convolutional layer in the group integration module is used to extract the compensation information y 1 of the first scale from the input features of the encoder.
  • SCM_1 is used to process the compensation information y 1 of the first scale to obtain y 2 .
  • SCM_3 is used to process the compensation information y 2 of the second scale to obtain the compensation information y 3 of the third scale, and so on, until SCM_m-1 processes the compensation information y m-1 of the m-1th scale to obtain the compensation information ym-1 of the third scale. Compensation information y m of m scales.
  • each decoder includes m group convolutional modules.
  • Each group convolution module includes an upsampling module and a first convolution module.
  • the second input features input to the decoder are divided into m groups, and the m groups of second input features correspond to the m group convolution modules one-to-one.
  • Each group of m group convolution modules decodes a set of second input features and one scale of compensation information to obtain m groups of decoding features. According to the m groups of decoding features, the output features of the decoder can be obtained.
  • the second input feature of the decoder is the output of the network structure located in the layer above the decoder.
  • the network structure located in the layer above the decoder can be the last decoder in the image segmentation model, or it can be the network structure located in the layer above the decoder. Another decoder one layer above the decoder.
  • a set of second input features are upsampled through the upsampling module in the group convolution module, and the output of the upsampling module is spliced with a scale of compensation information yi
  • yi As the input of the first convolution module in the group convolution module, 1 ⁇ i ⁇ m.
  • the outputs of m group convolution modules are concatenated and used as the output of the decoder.
  • the upsampling module includes a convolution layer with a convolution kernel of 1 ⁇ 1 and a deconvolution layer.
  • the first convolution module includes a batch normalization (Batch Normalization, BN) layer, an excitation layer and a convolution layer connected in sequence.
  • the excitation layer may be ReLU (rectified linear unit).
  • the first input feature of encoder 3 is a feature map with a channel number of 128 and a size of 44 ⁇ 36
  • the decoder includes four group convolution modules, namely group convolution module 1 to group convolution module 4.
  • the first input feature of the encoder 3 is input to the group integration module 3 for processing, and four compensation information of different scales are obtained, namely y 1 , y 2 , y 3 and y 4 .
  • the output features obtained are the second input features of the decoder 3, and the second input features include Feature map with channel number 256 and size 22 ⁇ 18.
  • the upsampling module in the group convolution module 1 is used to decode the features of the latent space from the first set of second input features, and combine the output of the upsampling module with the compensation of the first scale
  • the information y 1 is spliced on the channel and input into the first convolution module in the group convolution module 1. It is processed by the batch normalization layer, excitation layer and convolution layer in the first convolution module in turn and then the channel is output.
  • the upsampling module in the group convolution module 2 is used to process the second set of second input features, and the output of the upsampling module and the second scale compensation information y 2 are After splicing on the channels, it is input into the first convolution module in group convolution module 2.
  • the output features of the four group convolution modules in decoder 3 are spliced on the channel to obtain a feature map with a channel number of 128 and a size of 44 ⁇ 36, which is used as the second input feature of decoder 2.
  • the decoder in this application uses group convolution to replace the convolution block in the traditional decoder.
  • the m group convolution modules in each decoder are sub-networks with the same structure. Each sub-network is equivalent to a classifier.
  • the use of group convolution can not only effectively reduce the number of parameters, but also according to the parameters of multiple classifiers. Prediction diversity improves the segmentation accuracy of the model.
  • the image segmentation model provided by this application can be applied to the field of medical image segmentation to realize the extraction of lesions, organs and other regions, such as brain tissue segmentation, brain tumor segmentation, lung nodule segmentation and other tasks. It can also be applied to any task that requires segmentation or extraction of target areas in the image to be processed.
  • an initial model for image segmentation is constructed.
  • a corresponding training set is collected.
  • the training set includes multiple image sample pairs. Each image sample pair includes a brain tumor image sample and a segmented image sample corresponding to the brain tumor image sample.
  • the training set can be used to iteratively train the initial image segmentation model to minimize the loss function, thereby obtaining the trained image segmentation model.
  • the method of image segmentation using the above-trained image segmentation model and the above-mentioned method of training the initial image segmentation model may be executed by the same terminal device, or may be executed by different terminal devices.
  • the terminal device may not be limited to various smart phones, portable notebooks, tablets, smart wearable devices, computers, robots, etc.
  • the image segmentation model provided by this application is applied to brain tumor segmentation.
  • the effect of the number of group convolution modules of the decoder in the image segmentation model GEU-Net provided by this application on the brain tumor segmentation results is first verified through experiments, and the results are compared with the traditional small U-Net based on multiple different indicators. The segmentation results of the network are compared.
  • the evaluation indicators include Dice similarity coefficient, Sensitivity, Specificity, Hausdorff 95 and Parameter.
  • Dice similarity coefficient is used to measure the similarity between the network segmentation results and the standard segmentation results.
  • Sensitivity is used to measure the model's ability to identify positive examples.
  • Specificity is used to measure the model's ability to identify negative examples.
  • Hausdorff 95 is used to measure shape similarity. The larger the values of Dice similarity coefficient, Sensitivity and Specificity, the better the performance of the model. The smaller the Parameter value, the smaller the number of parameters in the model.
  • the number of group convolution modules in the decoder has a greater impact on the brain tumor segmentation results.
  • the GEU-Net model provided by this application has the best performance on the indicator Dice similarity coefficient.
  • the appropriate number of group convolution modules can be designed according to different segmentation tasks.
  • the segmentation results of the image segmentation model of this application are compared with the traditional U-Net model, small U-Net model, medium U-Net model, DeepLabV3+ model, PSP-Net (Pyramid Scene Parsing Network, The brain tumor segmentation results of the Pyramid Scene Parsing Network) model and the Attention U-Net model are compared.
  • the experimental results are shown in Table 2. It can be seen from Table 2 that the GEU-Net model provided by this application has strong advantages over other models in terms of indicators Dice similarity coefficient, Sensitivity and number of parameters.
  • the image segmentation model provided by the embodiment of the present application sets a group integration module between the corresponding encoder and decoder, and uses the group integration module to extract multiple different scales from the first input feature of the encoder.
  • the compensation information is input to the corresponding decoder, increasing the receptive field of the network layer, and reducing the difference between the low-level edge detail information in the encoder and the high-level semantic feature information in the corresponding decoder through compensation information at multiple different scales.
  • multiple group convolution modules are used to replace the traditional single convolution block in the decoder, making the decoder similar to integrating multiple classifiers. It can not only greatly reduce the number of parameters, but also integrate multiple classifiers.
  • Output combining improves the segmentation accuracy of the model.
  • the image segmentation model provided by the embodiments of the present application solves the problem in the existing U-Net network model that the low-level edge detail information and the high-level semantic feature information are directly spliced, resulting in the dilution of the high-level semantic feature information, thereby improving the processing efficiency of the model.
  • the segmentation accuracy of the image solves to a certain extent the problem of low accuracy of segmentation results of the traditional U-Net model.
  • FIG. 8 shows a structural diagram of an image segmentation device provided by an embodiment of the present application.
  • the image segmentation device may include:
  • Acquisition unit 801 acquires images to be processed.
  • the segmentation unit 802 is used to input the image to be processed into a trained image segmentation model for processing, and output the segmented image.
  • the image segmentation model has an encoder-decoder structure, and a group integration is provided between the corresponding encoders and decoders. Module, the group integration module is used to extract m compensation information of different scales from the input features of the encoder, and input m compensation information of different scales into the decoder, m ⁇ 2.
  • the group integration module includes a first convolutional layer and m-1 semantic compensation modules; the first convolutional layer is used to extract the first-scale compensation information from the first input feature of the encoder; m-1 The semantic compensation module is used to extract m-1 compensation information of different scales from the first scale of compensation information, where the output of the i-th semantic compensation module among the m-1 semantic compensation modules is the i-th +1 input to the semantic compensation module, 1 ⁇ i ⁇ m-1.
  • the semantic compensation module includes a second convolution layer, a plurality of third convolution layers and atrous convolution layers connected in sequence.
  • the decoder includes m group convolution modules
  • Input m compensation information of different scales into the decoder including:
  • Each group of m group convolution modules decodes a set of second input features and one scale of compensation information to obtain m sets of decoding features; the output features of the decoder are obtained according to the m sets of decoding features.
  • the group convolution module includes an upsampling module and a first convolution module
  • a set of second input features and a scale of compensation information are decoded through each of the m group convolution modules, including:
  • a set of second input features are upsampled through the upsampling module in each group convolution module, and the output of the upsampling module is spliced with a scale of compensation information as the first convolution in the group convolution module. module input.
  • the encoder includes a second convolution module, a downsampling module, and a compression and excitation module that are connected in sequence.
  • the second convolution module includes a batch normalization layer, an excitation layer, and a convolution layer that are connected in sequence.
  • the image to be processed is a medical image.
  • Module completion means dividing the internal structure of the device into different functional units or modules to complete all or part of the functions described above.
  • Each functional unit and module in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.
  • the above-mentioned integrated unit can be hardware-based. It can also be implemented in the form of software functional units.
  • the specific names of each functional unit and module are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present application.
  • For the specific working processes of the units and modules in the above system please refer to the corresponding processes in the foregoing method embodiments, and will not be described again here.
  • the terminal device 900 of this embodiment includes: a processor 901 , a memory 902 , and a computer program 904 stored in the memory 902 and executable on the processor 901 .
  • the computer program 904 can be run by the processor 901 to generate instructions 903.
  • the processor 901 can implement the steps in each of the above image color optimization method embodiments according to the instructions 903.
  • the processor 901 executes the computer program 904, it implements the functions of each module/unit in each of the above device embodiments, such as the functions of unit 801 and unit 802 shown in FIG. 8 .
  • the computer program 904 can be divided into one or more modules/units, and one or more modules/units are stored in the memory 902 and executed by the processor 901 to complete the present application.
  • One or more modules/units may be a series of computer program instruction segments capable of completing specific functions. The instruction segments are used to describe the execution process of the computer program 904 in the terminal device 900 .
  • FIG. 9 is only an example of the terminal device 900 and does not constitute a limitation on the terminal device 900. It may include more or less components than shown in the figure, or some components may be combined, or different components may be used.
  • the terminal device 900 may also include input and output devices, network access devices, buses, etc.
  • the processor 901 can be a central processing unit (Central Processing Unit, CPU), or other general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), or an on-site processor.
  • Programmable gate array Field-Programmable Gate Array, FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
  • a general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.
  • the memory 902 may be an internal storage unit of the terminal device 900, such as a hard disk or memory of the terminal device 900.
  • the memory 902 may also be an external storage device of the terminal device 900, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SD) card, a flash memory card (Flash) equipped on the terminal device 900. Card) etc.
  • the memory 902 may also include both an internal storage unit of the terminal device 900 and an external storage device.
  • the memory 902 is used to store computer programs and other programs and data required by the terminal device 900 .
  • the memory 902 may also be used to temporarily store data that has been output or is to be output.
  • the terminal device provided in this embodiment can execute the above method embodiments.
  • the implementation principles and technical effects are similar and will not be described again here.
  • Embodiments of the present application also provide a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method described in the above method embodiment is implemented.
  • Embodiments of the present application also provide a computer program product.
  • the computer program product is run on a terminal device, the method described in the above method embodiment is implemented when the terminal device executes it.
  • the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.
  • the present application can implement all or part of the processes in the methods of the above embodiments by instructing relevant hardware through a computer program.
  • the computer program can be stored in a computer-readable storage medium.
  • the computer program When executed by a processor, the steps of each of the above method embodiments may be implemented.
  • the computer program includes computer program code, which may be in the form of source code, object code, executable file or some intermediate form.
  • the computer-readable storage medium may at least include: any entity or device capable of carrying computer program code to the camera device/terminal device, recording media, computer memory, read-only memory (Read-Only Memory, ROM), random access Memory (Random Access Memory, RAM), electrical carrier signals, telecommunications signals, and software distribution media.
  • any entity or device capable of carrying computer program code to the camera device/terminal device recording media, computer memory, read-only memory (Read-Only Memory, ROM), random access Memory (Random Access Memory, RAM), electrical carrier signals, telecommunications signals, and software distribution media.
  • ROM read-only memory
  • RAM random access Memory
  • electrical carrier signals telecommunications signals
  • telecommunications signals telecommunications signals
  • software distribution media for example, U disk, mobile hard disk, magnetic disk or CD, etc.
  • first and second are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, features defined as “first” and “second” may explicitly or implicitly include at least one of these features.
  • connection should be understood in a broad sense.
  • it can be a mechanical connection or an electrical connection; it can be a direct connection or a connection through
  • the intermediate medium is indirectly connected, which can be the internal communication of two elements or the interactive relationship between two elements.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Quality & Reliability (AREA)
  • Radiology & Medical Imaging (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Medical Informatics (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)

Abstract

La présente demande se rapporte au domaine technique du traitement des images, et concerne un procédé et un appareil de segmentation d'image, un équipement terminal et un support d'enregistrement. Le procédé de segmentation d'image consiste à : obtenir une image à traiter ; entrer ladite image dans un modèle de segmentation d'image entraîné pour le traitement, et délivrer en sortie une image segmentée, le modèle de segmentation d'image étant d'une structure de codeur-décodeur, un module d'intégration de groupe étant agencé entre un codeur et un décodeur correspondant l'un à l'autre, le module d'intégration de groupe étant utilisé pour extraire m éléments d'informations de compensation de différentes échelles à partir de caractéristiques d'entrée du codeur et entrer les m éléments d'informations de compensation de différentes échelles dans le décodeur, et m étant supérieur ou égal à 2. Selon le procédé et l'appareil de segmentation d'image, l'équipement terminal et le support d'enregistrement fournis par la présente demande, le problème selon lequel des résultats de segmentation d'image des procédés de segmentation d'image existants sont imprécis peut être résolu dans une certaine mesure.
PCT/CN2022/137369 2022-03-24 2022-12-07 Procédé et appareil de segmentation d'image, équipement terminal et support d'enregistrement WO2023179095A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210295676.2A CN114782686A (zh) 2022-03-24 2022-03-24 一种图像分割方法、装置、终端设备及存储介质
CN202210295676.2 2022-03-24

Publications (1)

Publication Number Publication Date
WO2023179095A1 true WO2023179095A1 (fr) 2023-09-28

Family

ID=82424911

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/137369 WO2023179095A1 (fr) 2022-03-24 2022-12-07 Procédé et appareil de segmentation d'image, équipement terminal et support d'enregistrement

Country Status (2)

Country Link
CN (1) CN114782686A (fr)
WO (1) WO2023179095A1 (fr)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117409100A (zh) * 2023-12-15 2024-01-16 山东师范大学 基于卷积神经网络的cbct图像伪影矫正系统及方法
CN117523645A (zh) * 2024-01-08 2024-02-06 深圳市宗匠科技有限公司 一种人脸关键点检测方法、装置、电子设备及存储介质

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114782686A (zh) * 2022-03-24 2022-07-22 中国科学院深圳理工大学(筹) 一种图像分割方法、装置、终端设备及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200110960A1 (en) * 2018-10-05 2020-04-09 Robert Bosch Gmbh Method, artificial neural network, device, computer program and machine-readable memory medium for the semantic segmentation of image data
CN111080660A (zh) * 2019-11-14 2020-04-28 中国科学院深圳先进技术研究院 一种图像分割方法、装置、终端设备及存储介质
CN113284088A (zh) * 2021-04-02 2021-08-20 中国科学院深圳先进技术研究院 一种csm图像分割方法、装置、终端设备及存储介质
CN114782686A (zh) * 2022-03-24 2022-07-22 中国科学院深圳理工大学(筹) 一种图像分割方法、装置、终端设备及存储介质

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200110960A1 (en) * 2018-10-05 2020-04-09 Robert Bosch Gmbh Method, artificial neural network, device, computer program and machine-readable memory medium for the semantic segmentation of image data
CN111080660A (zh) * 2019-11-14 2020-04-28 中国科学院深圳先进技术研究院 一种图像分割方法、装置、终端设备及存储介质
CN113284088A (zh) * 2021-04-02 2021-08-20 中国科学院深圳先进技术研究院 一种csm图像分割方法、装置、终端设备及存储介质
CN114782686A (zh) * 2022-03-24 2022-07-22 中国科学院深圳理工大学(筹) 一种图像分割方法、装置、终端设备及存储介质

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
MA SHIQIANG; LI XUEJIAN; ZHANG ZEHUA; TANG JIJUN; GUO FEI: "GEU-Net: Rethinking the information transmission in the skip connection of U-Net architecture", 2021 IEEE INTERNATIONAL CONFERENCE ON BIOINFORMATICS AND BIOMEDICINE (BIBM), IEEE, 9 December 2021 (2021-12-09), pages 1020 - 1025, XP034066509, DOI: 10.1109/BIBM52615.2021.9669580 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117409100A (zh) * 2023-12-15 2024-01-16 山东师范大学 基于卷积神经网络的cbct图像伪影矫正系统及方法
CN117523645A (zh) * 2024-01-08 2024-02-06 深圳市宗匠科技有限公司 一种人脸关键点检测方法、装置、电子设备及存储介质
CN117523645B (zh) * 2024-01-08 2024-03-22 深圳市宗匠科技有限公司 一种人脸关键点检测方法、装置、电子设备及存储介质

Also Published As

Publication number Publication date
CN114782686A (zh) 2022-07-22

Similar Documents

Publication Publication Date Title
WO2023179095A1 (fr) Procédé et appareil de segmentation d'image, équipement terminal et support d'enregistrement
CN111104962B (zh) 图像的语义分割方法、装置、电子设备及可读存储介质
US20200134797A1 (en) Image style conversion method, apparatus and device
WO2020125495A1 (fr) Procédé, appareil et dispositif de segmentation panoramique
WO2022001623A1 (fr) Procédé et appareil de traitement d'image faisant appel à l'intelligence artificielle, dispositif et support de stockage
WO2020125498A1 (fr) Procédé et appareil de segmentation d'images de résonance magnétique cardiaques, terminal et support de stockage
CN110598714B (zh) 一种软骨图像分割方法、装置、可读存储介质及终端设备
WO2022160980A1 (fr) Procédé et appareil de super-résolution, dispositif terminal et support de stockage
WO2022188886A1 (fr) Procédé et appareil d'apprentissage de modèle de matage d'image, et procédé et appareil de matage d'image
WO2022166258A1 (fr) Procédé et appareil de reconnaissance de comportement, dispositif de terminal et support de stockage lisible par ordinateur
CN110619334B (zh) 基于深度学习的人像分割方法、架构及相关装置
CN111950570B (zh) 目标图像提取方法、神经网络训练方法及装置
CN112396605B (zh) 网络训练方法及装置、图像识别方法和电子设备
WO2021139351A1 (fr) Procédé de segmentation d'image, appareil, support d'enregistrement et dispositif électronique
CN107480621B (zh) 一种基于人脸图像的年龄识别方法
WO2020134674A1 (fr) Procédé d'identification d'empreinte palmaire, appareil, dispositif informatique et support d'informations
WO2022156280A1 (fr) Procédé et appareil de classification d'images pour terminal intégré, et terminal intégré
WO2019109410A1 (fr) Procédé d'apprentissage de modèle de réseau entièrement convolutif pour diviser une région de signal anormal dans une image d'irm
CN112488054A (zh) 一种人脸识别方法、装置、终端设备及存储介质
US20230021551A1 (en) Using training images and scaled training images to train an image segmentation model
CN113139617B (zh) 一种输电线路自主定位方法、装置及终端设备
CN115239655A (zh) 一种甲状腺超声影像肿瘤分割分类方法及装置
CN113971830A (zh) 一种人脸识别方法、装置、存储介质及电子设备
CN113780148A (zh) 交通标志图像识别模型训练方法和交通标志图像识别方法
CN115147434A (zh) 图像处理方法、装置、终端设备及计算机可读存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22933146

Country of ref document: EP

Kind code of ref document: A1