21results about How to "Reduce artifacts" patented technology

The invention discloses a rapid magnetic resonance imaging reconstruction and segmentation method and system, a medium and a product, and relates to the technical field of image processing. The method comprises the following steps: decomposing original three-dimensional magnetic resonance data into a plurality of independent two-dimensional magnetic resonance data slices along a slice sequence; image reconstruction is carried out on the two-dimensional magnetic resonance data slices through a pre-trained forward generator in the frequency domain Schrodinger bridge diffusion model so as to generate a two-dimensional magnetic resonance image, and the forward generator integrates a Fourier convolution module and is used for carrying out frequency domain transformation and feature enhancement processing on feature mapping in the image reconstruction process; inputting each two-dimensional magnetic resonance image into a pre-trained segmentation model in sequence, performing segmentation of a target region and a preset dangerous organ, and generating a corresponding two-dimensional segmentation mask; all the two-dimensional segmentation masks are rearranged and stacked according to the slicing sequence, and a three-dimensional segmentation result is generated through three-dimensional fusion operation. By implementing the technical scheme, magnetic resonance imaging reconstruction and segmentation can be quickly and accurately carried out.

This invention relates to the field of image processing technology, providing a three-dimensional scene reconstruction method, electronic device, and storage medium. The method includes: inputting a real image and a corresponding rendered image of the scene to be reconstructed into a motion-static separation module of a three-dimensional reconstruction model; obtaining a dynamic mask, a dynamic 3D point cloud representation, and a static 3D point cloud representation generated by the motion-static separation module based on the feature similarity between the real image and the rendered image, and the semantic segmentation result of the real image; inputting the dynamic mask, the dynamic 3D point cloud representation, and the static 3D point cloud representation into a hybrid rendering module of the three-dimensional reconstruction model; and obtaining a three-dimensional reconstructed scene generated by the hybrid rendering module based on the dynamic mask, the dynamic 3D point cloud representation, and the static 3D point cloud representation. This invention achieves high-precision motion-static separation, obtains a dynamic mask, and performs hybrid rendering of the dynamic 3D point cloud representation and the static 3D point cloud representation using the dynamic mask, significantly reducing artifacts in dynamic scenes.

This invention provides a cathode-unitized X-ray source, a static CT imaging system, and an image enhancement method, comprising: a vacuum tube; a cathode end for emitting electron beams, the cathode end including an exposure control module and multiple electron emission units arranged in a matrix on the same side of the exposure control module, all electron emission units being independent and electrically connected to the exposure control module; an anode target bombarded by the electron beam, the anode target including an anode body and a target surface formed on the anode body, the target surface being located on the emission path of the electron emission units; when the exposure control module receives an external control signal, it establishes a voltage line between the selected electron emission unit and the exposure control module, the electron emission units emitting electrons one by one or row by row toward the target surface, and the target surface reflecting X-rays to a window under the bombardment of electrons. This invention can increase the scanning range, improve the image quality of static CT, and extend the lifespan of the X-ray source.

The application provides a cloud image efficient synthesis method, belonging to the field of image processing, comprising the following steps: calculating the edge distance transform of the mask of the foreground image to generate a multi-level boundary information graph; converting the background image and the foreground image to the Lab color space, and inputting the Lab channel, the mask and the multi-level boundary information graph into a generator network to output a preliminary fusion image in the Lab color space; using a brightness discriminator and a chroma discriminator to respectively judge the local texture consistency of the preliminary fusion image and a real fusion image in the L channel, and the global color harmony of the preliminary fusion image and the real fusion image in the a and b channels; performing the adversarial training of the generator network and the discriminator based on a total loss function until the network model converges, and using the converged generator network to generate a final fusion image. The application can guarantee the high realism and consistency of the fusion image in the details and the overall color tone.

The present disclosure provides a data reconstruction method based on Radon domain sparse representation and related equipment, relating to the technical field of communication. The method comprises: obtaining low-resolution Radon coefficients; inputting the low-resolution Radon coefficients into a neural network of an encoding-decoding structure to obtain first high-resolution Radon coefficients; and increasing the sparsity of the first high-resolution Radon coefficients through an adaptive soft threshold function cascaded at the back end of the neural network to obtain second high-resolution Radon coefficients. The method can reduce the data size, operation time, and operation cost of time-domain inversion of reduced-time variable Radon transform, and improve the stability and resolution of reconstructed data.

A physiological mechanical signal monitoring system and method utilizing motion state to suppress artifacts, belonging to the field of biomedical signal monitoring and processing technology, aims to improve the existing distributed multi-node vibration and physiological signal acquisition systems, which suffer from severe crosstalk between multiple channels, signal aliasing caused by artifacts, and feature distortion, failing to meet the requirements of high-fidelity, multimodal dynamic acquisition. Key technical points: The signal monitoring system includes a multi-point distributed acquisition module, a motion state recognition module, an artifact detection and adaptive filtering module, and a main control and signal fusion module. The main control and signal fusion module is used to fuse and reconstruct the multi-channel physiological mechanical signals after artifact suppression processing, outputting high-fidelity physiological signals. This invention significantly reduces the coupling effect between different channels caused by environmental vibration, body motion impact, and matrix resonance through independent support of multiple spatial nodes and isolation of mechanical paths, significantly improving the spatial discrimination ability and original independence of the signals. This invention achieves multi-point physical isolation at the structural level, and then combines vector superposition to achieve preliminary noise reduction, effectively reducing artifacts caused by non-target actions and environmental disturbances.

This invention relates to a video object removal method based on a text-based video model and learnable negative tokens, comprising: acquiring the original video to be processed and a specified target to be removed; segmenting the target to be removed using a video segmentation model to generate a corresponding video mask sequence; preprocessing the original video and the video mask sequence to obtain masked video latent features and mask latent features; constructing an object removal model based on a text-based video model, wherein the object removal model retains the cross-attention module of the pre-trained text-based video model; setting the text prompt input to empty and introducing learnable negative tokens; inputting the masked video latent features, mask latent features, and negative tokens into the object removal model; the object removal model processes the negative tokens through the cross-attention module, suppressing object generation features within the mask region, and generating filling content based on video background information; and outputting the video after removing the specified target.

This invention relates to the field of bearing processing equipment technology, and provides a belt feeding device for conveying bearing races, including a base and a plate conveyor belt. A bracket is fixedly connected to the top of the base, and rollers are fixedly installed on both the top and bottom of the bracket. The plate conveyor belt is fixedly connected to the outside of the rollers, and a left support plate is fixedly connected to one side of the plate conveyor belt. A limit mechanism is provided inside the left support plate. By setting the limit mechanism, this invention allows the bearing to be conveyed upwards by the plate conveyor belt. When it rotates to the electromagnetic groove, the conductive head is squeezed and pushes the conductive column into the right electromagnetic attraction block. Current is transmitted through the conductive column to the left and right electromagnetic attraction blocks, causing them to open and attract the bearing onto the left and right support plates. This achieves the function of fixing the bearing races during the conveying process of the belt feeding device for conveying bearing races, preventing the bearings from falling off midway due to vibration, and improving the working efficiency of the equipment.

The application provides a fluid bulk modulus parameter extraction method, a fluid identification method, equipment and a medium, wherein the parameter extraction method comprises the following steps: dividing a seismic gather into at least three angle intervals, selecting a typical angle for each interval, and obtaining a corresponding partial stack for each interval; obtaining an elastic impedance curve and a curve of K f , f m and p for each typical angle from well logging data; performing elastic impedance inversion on the elastic impedance curve and the partial stack to obtain elastic impedance data of the typical angle; extracting wellside elastic impedance data from the corresponding elastic impedance data body by well logging; constructing a calculation relationship by using the curve of K f , f m and p and the elastic impedance curve of each typical angle, and bringing the curve of K f , f m and p and the wellside elastic impedance data into the calculation relationship to extract fluid bulk modulus. The technical scheme of the application can improve the resolution of the elastic impedance data body, reduce fluid identification artifacts, and improve the drilling rate of fluid inversion.

The present application relates to a kind of methods for training target segmentation model, comprising: obtaining training set, training set includes multiple frames of continuous original image, each frame original image is annotated with the real label of target;The original image in training set is input into convolutional neural network, obtains the feature map of multiple scales of each frame original image;First feature map of current frame original image and second feature map of previous frame original image are input into gated recurrent unit network, to obtain fusion feature map;Multiple scales of the feature map of current frame original image and fusion feature map are input into feature fusion decoding network, obtain the prediction label of target;Loss between real label and prediction label is calculated based on loss function, and target segmentation model is iteratively trained according to loss, until target segmentation model converges, and the trained target segmentation model is obtained.The target segmentation model obtained can accurately segment target, reduce the interference of complex background and the generation of artifact.

The application provides a kind of child magnetoencephalogram personalized stimulation paradigm optimization method based on reinforcement learning, comprising: step S1, the multi-channel magnetoencephalogram signal of the child to be measured is continuously collected by using optically pumped magnetometer array, and behavior data and physiological data are synchronously collected;Step S2, the magnetoencephalogram signal feature, behavior feature and physiological feature are extracted, and real-time state feature vector is fused and constructed;Step S3, the real-time state feature vector is input into the reinforcement learning model to obtain the prediction Q value of stimulation adjustment action, and the parameters of magnetoencephalogram stimulation paradigm of the child to be measured are dynamically adjusted based on the prediction Q value;Step S4, based on the magnetoencephalogram stimulation paradigm after parameter adjustment, the new multi-channel magnetoencephalogram signal, behavior data and physiological data of the child to be measured are collected, and the comprehensive reward value is calculated to update the reinforcement learning model.The beneficial effect is that the application can intelligently and dynamically adjust the magnetoencephalogram stimulation paradigm, and improve the multi-channel magnetoencephalogram signal quality of the child to be measured.

The invention provides a tunnel harmful gas scanning type monitoring method and equipment, and relates to the technical field of tunnel construction safety monitoring. Comprising the following steps: S1, synchronously scanning by at least two monitoring devices, recording direction angle data and a gas concentration measurement value of each ray, and acquiring tunnel boundary point cloud data; s2, calculating potential intersection points of rays from different devices in the tunnel space, and screening effective intersection points; s3, based on the tunnel boundary point cloud data, constructing a three-dimensional network model of the tunnel surface through a Delaunay triangulation algorithm; s4, in the three-dimensional grid, performing grading treatment according to concentration values from low to high, assigning values to voxels through which effective laser rays penetrate, and reconstructing a three-dimensional concentration field; and S5, generating a tunnel harmful gas concentration distribution three-dimensional model. According to the method, the path integral concentration data is converted into accurate three-dimensional space distribution information, the three-dimensional coordinates of the gas escape source can be accurately identified, accurate positioning of the harmful gas escape source is realized, and the problems of missing detection and misjudgment caused by fuzzy positioning are effectively solved.

This invention discloses a comprehensive ecological revetment system for river channels, relating to the field of ecological revetments. It includes natural earthen slopes along both sides of a river channel, with slope one and slope two constructed on these slopes. A horizontal lower walkway is provided between slope one and slope two, and an upper walkway is provided on the upper side of slope one. An erosion-resistant bank is constructed on the slope two near the river channel. This invention reinforces and protects the riverbank near the water's edge using an erosion-resistant filler composed of revetment bricks, mixed soil, a middle layer of backfill soil, and a top layer of planting soil. Simultaneously, the root system of the planted vegetation extends to protect soil and water, effectively preventing soil erosion. Furthermore, the erosion-resistant bank formed by reinforced steel grids as foundation piles further strengthens the revetment, maximizing protection against erosion. The floating green vegetation layer effectively conceals the clear boundary between the river surface and the riverbank edge.

The present disclosure relates to a knee joint lesion detection method and device, an electronic device and a storage medium, and relates to the technical field of knee joint lesion detection. The method comprises: performing similarity information calculation on a T1 first image and a T2 second image in an acquired knee image, and obtaining a corresponding third image based on the similarity information; synthesizing the T1 first image, the T2 second image and the third image to obtain a to-be-processed image; and performing knee joint lesion detection on the to-be-processed image based on a preset target detection model. The present disclosure can realize knee joint lesion detection.

The application relates to a structure perception-based image enhancement method and device. A low-quality image and a corresponding high-quality clear image are collected to form a data set; a structure perception Transformer network is constructed, including an U-shaped main network and a learnable Sobel convolution branch, the U-shaped main network comprising a plurality of structure guiding cascaded Transformer blocks and a structure guiding dynamic upsampling module; a joint loss function is designed to train the network; and a low-quality image to be processed is input into the network to generate an enhanced image. The learnable Sobel convolution branch extracts robust structure priori, the structure guiding cascaded attention mechanism and the structure guiding gated feedforward network realize deep fusion of global context and structure information, and the structure guiding dynamic upsampling module improves the recovery precision of structure regions. The structure integrity and the detail recovery quality of image enhancement are effectively improved, and clear and natural enhancement results are obtained.

The application discloses a target detection method and system in a sea fog environment, comprising the following steps: acquiring a sea fog image, constructing a gated shared feature extraction network, and generating sea fog features and object features to be detected in the image by using the gated shared feature extraction network; constructing a target detection network, and identifying target features in the object features to be detected by using the target detection network; generating a mask image corresponding to the image containing only the target features; constructing a target-guided defogging network, and guiding the target-guided defogging network to optimize the sea fog around the target features in the image by using the mask image, so as to obtain a target-enhanced feature map; and identifying target position information in the target-enhanced feature map. The sea fog image is processed by the gated shared feature extraction network, the target detection network and the target-guided defogging network, so that the problem that target recognition is difficult due to sea fog interference in the image is effectively solved, and the precision and reliability of target detection in a sea fog environment are improved.

The application discloses an image reconstruction method, device and equipment, relates to the technical field of medical images, and can adaptively adjust a sampling angle according to the total eccentricity of a to-be-detected object, guarantees that enough effective projection data is collected for image reconstruction, and thus the purposes of keeping image resolution, reducing artifacts and improving image quality are achieved. The method comprises the following steps: acquiring the total eccentricity of the to-be-detected object; determining a target sampling visual angle of the to-be-detected object according to a sampling visual angle corresponding to the total eccentricity of the to-be-detected object in a preset curve, wherein the preset curve records a mapping relationship formed by the to-be-detected object in different total eccentricities and sampling visual angles; scanning the to-be-detected object according to the target sampling visual angle, and reconstructing a tomographic image of the to-be-detected object by using the scanned projection data.

The application relates to the technical field of image enhancement, in particular to a high-precision blood vessel imaging enhancement method and system for cardiovascular imaging. The method comprises the following steps: acquiring an ECG curve of an electrocardiosignal before and after injection of a contrast medium of a patient, and acquiring a blood vessel image; determining the compressibility of each wave band according to the changes of the QT interval of each wave band and the heart rate before and after injection of the contrast medium; determining the incremental time length of each wave band by combining a standard time length, an actual time length and the compressibility; standardizing and adjusting the ECG phase of each frame of the blood vessel image according to the incremental time length and the actual time length of each wave band to obtain a standard phase; determining the matched mask of each negative film according to the standard phase of the negative film and the mask; and performing image subtraction on each negative film and the matched mask and performing blood vessel imaging enhancement processing. The application can improve the imaging precision of the blood vessels after digital subtraction, reduce the blood vessel artifact phenomenon, and enhance the imaging precision and reliability of the blood vessels.

The invention discloses an improved BM3D image denoising method based on multi-element fusion block similarity measurement. According to the method, wavelet decomposition is carried out on a noise image, a noise standard deviation is stably estimated on a highest-frequency sub-band, BM3D parameters are initialized, in a basic estimation stage, pixel intensity, local variance and gradient difference are fused to construct block similarity measurement, similar block matching grouping and three-dimensional block group construction are realized, and the method has the advantages of being high in robustness and high in robustness. And in a final estimation stage, a basic estimation image is obtained through three-dimensional collaborative hard threshold filtering and weighted aggregation, in the final estimation stage, secondary matching is guided by the basic estimation image, a three-dimensional block group is constructed again, a structure and texture adjustment factor is introduced to form an adaptive Wiener coefficient for Wiener collaborative filtering, and finally, a target de-noised image is output through weighted aggregation. In the whole process, the phenomena of mismatching and over-smoothing of a traditional BM3D algorithm under the condition of high noise can be reduced, the edge and texture detail keeping capability is enhanced, and the denoising quality of a noise image is improved.

The application discloses a kind of LDCT image reconstruction methods based on depth double-domain joint guide learning, comprising: the projection domain data of low dose CT obtained clinically is first reconstructed and recovered in projection domain network, simultaneously, only the projection domain data of being operated as guide is projected into image domain reconstruction network, realize the reconstruction operation of low dose CT, in addition, double-domain network also adopts a new structure, can better fuse local and global information.The application realizes the high-quality reconstruction of low-dose CT, effectively utilizes the data of original projection domain.The application not only proposes a new domain guide strategy, uses original projection domain data to guide the reconstruction of image domain, but also designs a new network module for fusing global and local information, which enhances the reconstruction effect of traditional double-domain model.

The invention provides a registration and fusion dual-drive misaligned infrared and visible light image fusion method, which comprises the following steps of: firstly, extracting multi-scale common features which are irrelevant to modes and have higher discrimination through a multi-scale feature extraction module based on comparative learning; then, similarity is calculated according to inter-modal multi-scale common features, a dense deformation field is generated, and a registration image is obtained; thirdly, inputting the registered image and the corresponding fixed image into a feature extraction module to obtain image features, and performing feature fusion through a designed collaborative attention fusion module; the fusion module combines window attention, gradient channel attention and feedback characteristics of image fusion on registration, and realizes fine alignment of the characteristics and redundant characteristic suppression of dislocation; and finally, generating a final fusion image from the fusion features through a feature reconstruction module. In conclusion, collaborative optimization of registration and fusion is achieved, and artifacts generated in the fusion process of the unregistered infrared and visible light images are effectively reduced.