358 results about "Pyramid (image processing)" patented technology

The present invention pertains to the technical field of digital image processing. Disclosed are an unsupervised super-pixel segmentation method and system assisted by collaboration between an atrous pyramid and an attention mechanism. The method comprises: combining image RGB channel information with position information of a pixel point; constructing a channel attention module by using an attention mechanism; processing the result of the attention mechanism by using atrous spatial pyramid pooling; constructing a loss function, constructing a clustering loss term, and, by using a spatial smoothing loss term, constructing a reconstruction loss term; updating model parameters by using an Adam optimizer, and using for super-pixel generation an effective depth feature obtained from the last separation; and obtaining the maximum value of a channel dimension by using an argmax function, converting a processing result of the argmax function into a two-dimensional array, and, on the basis of a defining condition, completing adaptive super-pixel segmentation in a CPU. According to the present invention, the complexity is low, the adaptive and generalization capabilities are high, and effective support is provided for improving image processing efficiency and accuracy.

The invention provides an AI-based automatic identity card image cutting and OCR recognition method and system, and relates to the technical field of image processing, and the method comprises the steps: collecting a multi-angle depth image and a visible light image, reconstructing a three-dimensional model of a certificate through a three-dimensional point cloud registration network, and achieving the perspective correction of an image through the combination of attention posture estimation and projection matrix optimization; a double-flow memory neural network is adopted to carry out information area positioning, a feature extraction backbone network constructs a multi-scale feature pyramid, and a dynamic memory updating network carries out feature matching and dictionary updating based on a time sequence attention module and comparative learning; and performing character recognition on the segmented information sub-images through a multi-branch feature enhancement network, improving the recognition precision in combination with a self-attention enhancement unit and a context reasoning unit, and dynamically optimizing network parameters by using a reinforcement learning controller until a preset recognition credibility threshold value is reached.

The invention discloses a strip steel surface defect detection method based on an attention mechanism and an improved UNet network, relates to the field of industrial image processing, and aims to solve the defects of a traditional detection method in defect positioning precision, multi-scale feature fusion and class imbalance. According to the technical scheme, an improved UNet network architecture is adopted, and the method specifically comprises the following steps: in a coding stage, designing a UELAN feature extraction module to improve parameter efficiency; a coordinate attention mechanism (CA) is introduced to enhance the focusing capability on a defect area; and capturing multi-scale high-level semantic features in combination with a cavity space pyramid pooling module (ASPP). In a decoding stage, a convolutional block attention mechanism (CBAM) is integrated, and a channel and spatial feature fusion are dynamically optimized. Besides, a combined loss function is put forward, focus loss and Dice loss are fused, the problem of boundary uncertainty is relieved in combination with a label fuzzification strategy, and category imbalance is coped with through weighted cross entropy. According to the invention, high-precision pixel-level segmentation of strip steel surface defects is realized.

The invention discloses a coal mining subsidence area surface crack intelligent extraction method based on an improved Transform model, and belongs to the technical field of remote sensing image processing. Firstly, an unmanned aerial vehicle carrying a high-resolution optical camera is used for collecting images, the image overlapping rate of 70%-80% is guaranteed, and a training data set is constructed through professional labeling, cutting screening and data enhancement. The encoder of the innovative model is very distinctive, and the adaptive multi-scale patch mapping layer can dynamically adjust the patch size according to the local complexity of the image and efficiently extract features; double-attention fusion is combined with optimization position coding, and long-distance dependency capture is enhanced; and the calculation amount and the overfitting are reduced by the dynamic sparse connection full-connection layer. Residual attention enhancement pyramid pooling and a space-channel attention bottleneck mechanism are adopted, key features are highlighted, and noise is suppressed; and a breakpoint detection and connection rule determination module is utilized to realize complete restoration of the ground fracture. After the data set is used for training a model, deployment is carried out, and through preprocessing, encoding and decoding and post-processing, ground fracture information can be accurately obtained, and a data foundation is built for mining area safety management and geological disaster prevention and control.

The invention relates to the field of medical image processing, in particular to a lung cancer PET-CT fusion segmentation method and system based on multi-modal feature comparative learning, and the method comprises the steps: firstly extracting PET and CT image features, projecting the features to a shared semantic space through a semantic guide type symmetric comparative learning architecture, obtaining key region features through a focus adaptive attention sampling mechanism, and carrying out the segmentation of a target region; optimizing feature representation through a cross-modal feature difference self-calibration mechanism, constructing a multi-scale feature pyramid, fusing features of different scales by using a multi-scale hierarchical contrast learning mechanism, and performing self-supervised learning by combining an anatomical guidance self-supervised contrast learning enhancement module and using a CT anatomical structure, so as to further reinforce the features; a high-precision lung cancer lesion segmentation result is generated through a decoder network, the Dice coefficient is increased from 0.78 to 0.91, and the detection rate of lesions below 10 mm is increased from 65% to 87%. A novel efficient and accurate image processing method is provided for lung cancer diagnosis.

The invention provides a visible light and thermal infrared image target fusion detection method for the visual angle of an unmanned aerial vehicle, and belongs to the technical field of image processing, and the method comprises the steps: carrying out the preprocessing and data enhancement of a data set of visible light and thermal infrared images of a detection region obtained by the unmanned aerial vehicle, and obtaining a plurality of initial images; initial images under different scales are analyzed based on a multi-scale dynamic feature pyramid network, fusion processing is carried out on the initial image under a visible light mode and the initial image under a thermal infrared mode based on an attention fusion mechanism, a candidate area is locked, and then highlighting processing is carried out on a tiny target, a long-distance target and the candidate area; and splitting all the target samples to obtain a training set and a verification set, training a preset network model based on the training set to obtain an initial detection model, and performing model performance evaluation and model optimization on the initial detection model based on the verification set to obtain a target detection model. The problem of target detection challenge of the unmanned aerial vehicle in a complex environment is solved.

The invention relates to the technical field of image analysis, in particular to a QFP chip image pyramid positioning system which comprises a pyramid construction module, a pin initial positioning module, a pin accurate positioning module and an interference area elimination module. According to the method, an original chip image is processed through step-by-step down-sampling and Gaussian filtering, the image processing speed is effectively improved when a multi-layer image pyramid is constructed, a pin connected domain is clearly identified by applying binarization processing and a pin neighbor connection strategy, the pin area identification capability of the system is enhanced, and the identification accuracy of the system is improved. Region recalibration is carried out by using an up-sampling image of an original image data size, accurate pin positioning can be executed while high resolution is maintained, the probability of wrong positioning is remarkably reduced by accurately extracting pin contour information and effectively eliminating non-target connected domains, the chip assembling quality is improved, the chip assembling speed is increased, and the chip assembling efficiency is improved. And remarkable benefits are brought to a high-speed electronic assembly line.

The invention relates to an underwater image processing technology in the field of underwater autonomous perception, in particular to an underwater low-quality image enhancement method based on a Laplacian pyramid and contrast learning. Comprising the following steps: S1, image input and multi-scale decomposition: decomposing an input image into a low-frequency residual layer and a plurality of high-frequency detail layers through Laplacian pyramid decomposition; s2, inputting the low-frequency residual error layer into a global illumination and color correction sub-module to obtain an enhanced low-frequency residual error layer; s3, inputting the enhanced low-frequency residual error layer and the multi-scale high-frequency detail layer into a frequency domain enhancement feature module to obtain multi-scale enhanced high-frequency detail layer output; s4, performing progressive reconstruction on the enhanced low-frequency layer output and the high-frequency enhancement layer output of each scale according to the inverse process of the Laplacian pyramid; and S5, outputting a result and applying. The method can be used for sharpening optical image data in the operation task of the autonomous underwater vehicle, and the image quality is improved.

The invention relates to the technical field of computer image processing, and particularly discloses a double-branch defogging method based on a Laplacian pyramid, and the method comprises the following steps: S1, building a training data set A containing an original clear image and a corresponding foggy image; s2, performing cutting and data enhancement processing on the input image; s3, a defogging network LPSDNet based on the Laplacian pyramid is constructed, the defogging network LPSDNet comprises a multi-feature fusion focusing branch and a Laplacian pyramid supplementary branch, and after feature extraction of the two branches is completed, features learned by the two branches are fused through a learnable fusion module; s4, training the defogging network LPSDNet by adopting the training data set A until a preset loss function is converged; obtaining a trained defogging network LPSDNet in combination with an objective function; and S5, inputting an image to be defogged into the trained defogging network LPSDNet for testing and verification to obtain a defogging result. The global structure information of the image can be fully captured, and the distortion phenomenon is prevented.

The invention provides a three-dimensional map construction method and device, computer equipment and a storage medium, and belongs to the field of image processing.The method comprises the steps that image information is obtained in real time; estimating pose information of the target camera; performing sub-mapping processing on the image information to obtain multiple pieces of sub-map information; performing dense mapping on each piece of sub-map information by adopting Gaussian sputtering, generating Gaussian point cloud corresponding to each piece of sub-map information, and generating a region mask corresponding to each piece of sub-map information in combination with the image information; determining a high color gradient region and a sparse region according to the region mask; constructing a multi-layer Gaussian pyramid of the key frame based on the high color gradient region and the sparse region; and reducing the number of layers of the pyramid through iterative training, completing rendering of the sub-map information, and obtaining a three-dimensional map corresponding to the image information. Therefore, computing resources are saved, the data processing efficiency and the map rendering speed are improved, and the SLAM performance is improved.

The invention belongs to the technical field of image processing, and discloses a frame image and event stream feature fusion-based target detection method, which comprises the following steps of: acquiring RGB (Red, Green, Blue) images and event stream data, and constructing a time-space synchronous multi-modal data pair; constructing a double-flow feature extraction backbone network to extract events and RGB features; based on a cross attention mechanism, multi-head attention is utilized to realize semantic alignment; the fusion weight is adaptively adjusted according to the statistical distribution; a multi-level fusion network is constructed, cross-scale fusion is performed by using an FPN pyramid, and target positioning and classification feature expression are cooperatively enhanced through a spatial semantic aggregation module. According to the target detection method based on frame image and event stream feature fusion, through a multi-modal data collaborative perception and self-adaptive feature optimization mechanism, the detection robustness in a complex scene is remarkably improved, the dynamic target omission ratio and the error recognition rate are effectively reduced, and the target detection efficiency is improved. And an all-weather high-precision environment perception capability is provided for an automatic driving system.

The invention discloses a remote sensing image super-resolution method, system and device based on implicit neural representation, and belongs to the technical field of deep learning and remote sensing image processing. The method comprises the following steps: firstly, acquiring a data set suitable for a remote sensing image, training a remote sensing image super-resolution model based on the data set, then taking a to-be-reconstructed low-resolution remote sensing image and a pre-constructed query array as input, and performing multi-scale feature extraction by a hierarchical feature pyramid extraction module; the double-path implicit attention feature fusion module carries out fusion, and finally the implicit high-resolution reconstruction module carries out the reconstruction of a high-resolution image. According to the method, the feature pyramid, the attention mechanism and the like are combined with implicit neural representation, the features of different levels are fused together, the problems of multi-scale feature fusion and efficient information extraction in the remote sensing image super-resolution process are solved, the remote sensing image super-resolution precision can be improved, and meanwhile the remote sensing image super-resolution efficiency is improved. And the target of continuous scale super-resolution of the remote sensing image is realized.

The invention discloses a visible light and infrared fusion target detection method and system for an all-time unmanned aerial vehicle scene, and belongs to the technical field of unmanned aerial vehicle aviation and intelligent image processing. The method comprises the following steps: analyzing an original visible light image to calculate average brightness and root-mean-square contrast, and quantifying scene illumination conditions; distortion correction and registration are carried out on the original image; a parallel backbone network is adopted to extract visible light and infrared twinborn feature maps, the number of channels allocated to two modal features is dynamically adjusted through 1 * 1 convolution according to illumination condition parameters, and then a dual-light fusion network containing a gated convolution block is utilized to perform weighted fusion; and target prediction is carried out through the progressive feature pyramid network. Through an illumination adaptive fusion strategy and targeted data enhancement, the precision and robustness of target detection in a complex all-day scene are significantly improved, and the method is suitable for real-time target detection tasks of the unmanned aerial vehicle.

The invention discloses a marine litter intelligent identification and classification method based on a multispectral unmanned aerial vehicle remote sensing image, and relates to the technical field of image processing, and the method comprises the steps: collecting and preprocessing a remote sensing image of a preset sea area, constructing a color correction network, and carrying out the color correction of the preprocessed remote sensing image; multiband features and spectral indexes of enhanced garbage detection are obtained, and multispectral features are generated through multi-scale feature pyramid structure fusion; training a conditional generative adversarial network, inputting a remote sensing image and multispectral features, and performing pixel-by-pixel fusion on the enhanced remote sensing image and the remote sensing image after color correction to obtain a multiband fusion feature map; and constructing a lightweight detection network based on the OfficientDet-Lite, taking the multi-band fusion feature map as input, and outputting a bounding box and a garbage category of garbage by adopting a joint detection-classification architecture. According to the method, the problems of color distortion, contour fuzziness, small target missing detection and the like in marine litter identification are solved, and meanwhile, the real-time performance of marine litter identification and classification is ensured.

The invention provides a DEM super-resolution reconstruction method and system fusing gradient feature Transform, and belongs to the technical field of image processing, the reconstruction method is composed of a plurality of residual feature fusion blocks, a multi-scale pyramid segmentation Transform layer in the residual feature fusion blocks fuses multi-head self-attention and pyramid segmentation attention, and the multi-scale pyramid segmentation Transform layer fuses the multi-head self-attention and the pyramid segmentation attention. In order to make the reconstructed DEM closer to reality, a double-filtering convolution block is used, high and low frequency features of the DEM are fully extracted through a parallel structure of two different filtering blocks, and finally super-resolution reconstruction is carried out under double constraints of a gradient domain and a height domain. The method adapts to topographic features of different scales and complexities, and solves the problem that DEM details cannot be accurately recovered by a traditional algorithm.

The invention belongs to the technical field of computer vision and image processing, and discloses a remote sensing small sample target detection method based on double-attention guided transfer learning, and the method comprises the steps: obtaining a remote sensing image data set, and carrying out the preprocessing; taking the preprocessed remote sensing image training set as input, constructing a basic detection model by using a ResNet-101 backbone network, a feature pyramid network and a content awareness upsampling and regional proposal network, and obtaining basic model parameters; basic model parameters are used as input, a DA-FSDET network is trained based on a content awareness strip pyramid and a deformable attention area proposal network, and the trained DA-FSDET network is used to acquire a category detection frame containing small sample categories and confidence. Through cascading and cooperative work of the content awareness stripe pyramid and the deformable attention area proposal network, the detection precision and robustness of the multi-scale target in the remote sensing image are effectively improved.

The invention discloses a hyperspectral image classification method based on super-pixel multi-scale image fusion Transform, and belongs to the technical field of image processing and computer vision. The invention aims to provide a hyperspectral image classification method based on super-pixel multi-scale image fusion Transform, which realizes accurate classification of hyperspectral images through multi-scale image network feature extraction and super-pixel feature global modeling. According to the method, features of a hyperspectral image are divided into a plurality of superpixel areas through superpixel segmentation, a superpixel-based image structure is constructed, and global information fusion is carried out on extracted multi-scale superpixel features through a Transform network self-attention mechanism by applying multi-scale image convolution operation, so that a final classification result is generated. According to the invention, a multi-scale image convolution feature extraction network is developed, context feature extraction is carried out on a superpixel feature pyramid, the spectral feature extraction capability is improved, and the robustness of the model to different scale feature changes is enhanced.

The invention relates to the technical field of photovoltaic module detection, in particular to a TOPCon battery piece defect detection device and a TOPCon battery piece defect detection method. The optical imaging module comprises a 25,000,000-pixel camera, a three-color integral light source array and a multispectral optical filter group, and the light source array is composed of red, green and blue LEDs and supports pulse synchronous triggering; the traditional algorithm processing module is integrated with a Halcon image processing library and is configured with a VarThreshold operator dynamic threshold segmentation unit, an adjustable kernel size corrosion filtering unit and a light and shade separation unit, and an Erosion Rectangle1 operator is used for corrosion filtering; the AI defect detection module is internally provided with an improved YOLOv8 neural network model, and the model structure comprises a Backbone network embedded with a CBAM attention mechanism, a BiFPN network bidirectional feature pyramid and a dynamic loss function optimizer; according to the technical scheme, the limitation problems of low positioning precision, poor detection efficiency, single defect identification dimension and the like of an artificial vision or traditional probe contact detection means can be solved.

The invention provides an urban landscape semantic segmentation system adopting an EGLiteSeg model, and belongs to the field of image processing. Comprising three main parts: an encoder; an polymerizer; and a decoder. The EGLiteSeg framework adopts a lightweight encoder-decoder pipeline to carry out image processing, and an encoder of the EGLiteSeg framework is provided with a five-level DeSTDCNet framework. In the first four stages, depth separable convolution with the stride being 2 is utilized, and the AECA module automatically adjusts the size of a kernel according to the number of input channels so as to enhance the multi-scale feature adaptability. For context aggregation, the framework uses GDSPPM. The module processes features through deep convolution to enhance distinguishability, and then performs multi-scale pyramid pooling to capture hierarchical contextual information. The multi-scale features are connected together and fed into a decoder. The decoder is composed of two key components: a universal attention fusion module (UAFM) and a partition head. The method not only has high semantic segmentation accuracy, but also has superiority in real-time performance, and meets actual requirements.

The invention relates to the field of remote sensing image processing, in particular to a karst landform remote sensing dynamic evolution analysis method and a karst landform remote sensing dynamic evolution analysis system, which can be applied to the fields of karst region resource management, environmental protection, geological disaster early warning and the like. According to the method, karst landform change information is obtained through multi-temporal remote sensing image data, a time-space analysis model composed of self-attention enhancement, a feature pyramid and a recurrent neural network is constructed, a constraint module is designed in combination with karst landform priori knowledge, and accurate prediction is achieved; multi-source data fusion improves the feature capture capability, and the observation advantage is obvious especially under complex meteorological conditions; the multi-module cascade design is used for accurately modeling space-time dynamic change, and the prediction accuracy is improved; priori knowledge constraint is introduced to enhance feature recognition, and misclassification is reduced; on the whole, the method effectively integrates multidisciplinary technologies, the precision and reliability of karst landform dynamic evolution analysis are improved, and a powerful tool is provided for research in related fields.

The invention discloses a heart motion feature extraction method based on optical flow estimation, and relates to medical image processing. Preprocessing the input four-dimensional space-time cardiac magnetic resonance imaging data, namely, scaling pixel values; inputting two frames of images which are continuous in time into a feature encoder and a context encoder for feature extraction, wherein the two frames of images are divided into a reference frame and a moving frame; calculating the correlation between the feature maps through a correlation volume calculation module, constructing a correlation pyramid, and extracting the feature maps to provide matching information for subsequent optical flow estimation; the motion feature iteration enhancement module iteratively and continuously refines an optical flow estimation result through a deformable convolution and global motion aggregation (GMA) module; model parameters are optimized based on a weighted sum of luminosity consistency loss, smoothness loss, and gradient consistency loss. Edge features are adaptively captured through deformable convolution, global and local features are fused by using a GMA module, optical flow prediction errors are effectively reduced, motion estimation quality is improved, and key features of a heart edge region are maintained.

The invention relates to the technical field of remote sensing image processing, in particular to an end-to-end neural network InSAR (Interferometric Synthetic Aperture Radar) phase unwrapping method based on mixed attention, which takes U-Net as a basic framework, extracts key features and improves resolution through down-sampling and up-sampling operations, thereby effectively recovering detail information. Specifically, a convolutional block attention module (CBAM) and two receptive field modules, namely cavity spatial pyramid pooling (ASPP) and a receptive field module (RFB) are combined to construct an RFAUNet network model, and then simulation data sets generated by two methods of digital elevation inversion and random matrix generation are used for training a neural network until a good unwrapping effect is obtained. And finally, carrying out a phase unwrapping experiment on simulation data and real data to verify the effectiveness and robustness of the RFAUNet network model.

The invention relates to the technical field of image processing, in particular to a mining truck edge detection method based on a multi-scale edge membership map. The method comprises the following steps: 1, constructing a multi-scale image pyramid: acquiring a grayscale image as input, performing multi-scale pixel sampling, and performing fuzzy processing by using a Gaussian function to construct a Gaussian image pyramid; 2, gradient calculation: respectively calculating the gradient of the image in the x direction and the y direction by using a Scharr operator; 3, the standard deviation of the local gradient difference is calculated, edge pixels and noise points are further effectively distinguished by calculating the standard deviation of the local gradient difference through the statistical characteristics of the local gradient difference, and therefore the edge detection precision is improved; 4, calculating an edge membership degree; 5, fusing a multi-scale edge membership map; 6, adaptive lagging connection: adopting lagging connection to further process the multi-scale edge membership fusion image so as to improve edge continuity and positioning accuracy; according to the invention, the reliability and safety of automatic loading operation are improved.

The invention provides a multi-scale pyramid weighted fusion underwater image enhancement method based on double prior, and the method comprises the steps: obtaining a degraded underwater image, and carrying out the global and local cooperation body color calibration of the degraded underwater image; decomposing the color correction image into a base layer, a detail layer and a noise layer by adopting a variational decomposition algorithm; performing spectral prior and transmissivity loss constraint on the base layer image to obtain a defogged image; fusing the detail layer image and the noise layer image to obtain a filtered image; performing enhancement processing on the filtered image by adopting a nonlinear mapping and contrast enhancement strategy to obtain an enhanced image; performing multi-level feature integration and reconstruction on the defogged image and the enhanced image by adopting a multi-scale pyramid adaptive weighted fusion method to obtain an underwater image with natural color and high visual definition; according to the method, the traditional image processing and variational optimization thought is combined to effectively correct the color deviation of the degraded underwater image, the image contrast and the detail definition are improved, and the visualization effect of the underwater image is improved.

The invention discloses a multi-modal target detection method, a multi-modal target detection device, a medium and electronic equipment, and relates to the field of image processing. The multi-modal target detection method comprises the following steps: acquiring an infrared image and a visible light image of the same scene; the infrared image and the visible light image are input into a multi-modal target detection model, the multi-modal target detection model comprises a double-flow feature extraction module, a context-guided pyramid network and a prediction module, the double-flow feature extraction module comprises two processing flows, and feature maps of different scales are extracted from the infrared image and the visible light image respectively; the context-guided pyramid network carries out feature fusion on the feature maps from a large scale to a small scale to obtain a fused feature map; the prediction module is used for determining a prediction frame where a target is located in the fusion feature map.

The invention provides a traffic detection method and system based on visible light and infrared image fusion, and belongs to the field of image processing. Comprising the following steps: acquiring an original visible light image and an infrared image in the same scene, and preprocessing; based on adaptive filtering, Laplacian pyramids are constructed for the preprocessed visible light image and infrared image, and multi-scale fine registration is carried out on the two Laplacian pyramids; performing gradient weighted fusion on the high-level images of the two Laplacian pyramids to obtain a high-level fusion image, and performing double-domain average fusion on the low-level images of the two Laplacian pyramids to obtain a low-level fusion image; performing layer-by-layer reconstruction based on the high-layer fusion image and the low-layer fusion image to obtain a reconstructed fusion image; and carrying out traffic detection based on the reconstructed fusion image. According to the method, edge details are reserved through adaptive filtering, the advantages of visible light and infrared images are integrated through different weighted fusion, and the target recognition accuracy and robustness in traffic detection are remarkably improved.

The invention discloses a power line icing grade classification method, electronic equipment and a storage medium, belongs to the technical field of image processing, and solves the problem of how to improve the accuracy of power line icing detection.The method comprises the steps that firstly, an original icing image is obtained, and a standard icing image is preprocessed; a DConvLSTM spatio-temporal feature extraction model is used to extract meteorological spatio-temporal features, Swin Transform and ResNet are combined to extract global and local features of a standard icing image, and fusion of the global and local features of the image is realized through spatial pyramid pooling to obtain multi-scale features. The method comprises the following steps: performing weighted summation on meteorological spatial-temporal features, topographic features and multi-scale features to obtain a spatial-temporal multi-modal feature graph, performing block embedding by using a patch partition module, introducing a related mechanism and a CBAM module for weighted enhancement, and performing lightweight deconvolution, residual connection optimization and bilinear interpolation to obtain a high-quality graph. And the safety of a power line is ensured.

The present invention provides a back segmentation method based on improved Mask RCNN. The method comprises the following steps: collecting adolescent forward-bending bare back human body images to obtain a color image dataset; manually annotating human body bare back regions to obtain annotation files corresponding to original color images; constructing an improved Mask RCNN framework; training an improved Mask RCNN network to obtain a human body bare back segmentation model; and using the back segmentation model to obtain background-free human body bare back. The present invention effectively improves a feature pyramid in a Mask RCNN framework in light of an artificial intelligence technology and an image processing technology, and improves the accuracy of human body back region segmentation on the basis of instance segmentation characteristics of Mask RCNN.

The invention relates to the technical field of image processing and information retrieval, and discloses an image feature extraction and appearance trademark retrieval method based on deep learning, which comprises a feature extraction unit, a cross-modal fusion module, a retrieval strategy module and a dynamic weight distribution unit. The feature extraction unit extracts multi-level semantic information through a multi-branch convolutional network and multi-scale pyramid pooling; the cross-modal fusion module integrates vision, text and color information by using space attention, channel attention and cross attention mechanisms; the retrieval strategy module adopts Hash coding and global self-attention to realize rapid screening and fine matching; the dynamic weight allocation unit suppresses background interference through learnable parameters. The trademark retrieval precision and efficiency can be effectively improved, and the method is suitable for a multi-modal data processing scene.

A method, apparatus, non-transitory computer readable medium, apparatus, and system for image processing include obtaining a text prompt; generating, using a generator of an image generation model, a feature embedding based on the text prompt, wherein the feature embedding includes a first set of channels that encodes a first value of an image characteristic and a second set of channels that encodes a residual between the first value of the image characteristic and a second value of the image characteristic; and generating, using a decoder of the image generation model, a synthetic image corresponding to the second value of the image characteristic based on the feature embedding.