15results about How to "Preserve continuity" patented technology

This invention discloses an anisotropic sparse deconvolution enhancement method and system for ultrasound tumor image segmentation. The method includes: establishing a local convolution model for ultrasound RF images based on the anisotropic characteristics of the ultrasound xz imaging plane; constructing an axis-aligned elliptical symmetric anisotropic point spread function (PSF), whose lateral and axial spread scales are obtained through physical parameter conversion or data-driven methods; constructing an ultrasound sparse reconstruction objective function incorporating the anisotropic PSF, adding non-negativity constraints, introducing a weighted Hessian continuity term and an L1 sparse prior term in the xz plane direction, and optimizing the solution to obtain the sparse reconstruction result; using this result as input, employing an RL iterative deconvolution method based on the anisotropic PSF, and achieving image boundary sharpening and noise suppression through PSF flip kernel ratio iteration; inputting the enhanced image into a tumor segmentation network to complete training and inference, thereby achieving ultrasound tumor image segmentation.

The present application relates to the field of automobile driving technology, and can be applied to autonomous driving vehicles, and specifically to a driving style recognition method and device, the method comprising: obtaining each feature vector value of a target vehicle at each continuous moment, the feature vector being used to represent the instantaneous driving state of the target vehicle, wherein the feature vector comprises a state feature for representing the state between the target vehicle and its related objects; sampling each feature vector value with N continuous values ​​as a sampling unit to obtain multiple samples, wherein N is a preset value; determining the driving style of the target vehicle according to the multiple samples. Based on the technical solution provided by the present application, the driving style recognition result can be made more comprehensive and accurate.

This invention discloses a stable torque transmission system for a photovoltaic tracking bracket, comprising a main shaft, a circular tube synchronous shaft, a torque transmission connecting sleeve, an inner reinforcing member, and a flexible limiting structure. The end of the circular tube synchronous shaft is locally shaped to form a shaped torque transmission section including a shaped torque transmission area and an arc-shaped stabilizing area retaining the original circular tube's outer arc contour. The torque transmission connecting sleeve is fitted onto the outer wall of the shaped torque transmission sections of two adjacent circular tube synchronous shafts to achieve torque transmission between the synchronous shafts through surface-to-surface mating. The inner reinforcing member is disposed inside the shaped torque transmission section to enhance the local strength and anti-instability capability when the end is connected to the reducer. The flexible limiting structure is connected between the main shaft and the circular tube synchronous shaft to limit the radial displacement of the synchronous shaft and allow it to rotate relative to the main shaft. This invention balances the stable torque transmission of the shaped surface with the stability of the circular tube structure, reducing off-center wear in the connection area and improving the long-term operational reliability of the system.

The present invention provides a belt tear detection method and system based on a discrete state optional space model, comprising: collecting multiple images of belt running sites; marking tear targets on each image to obtain an image-label pair, and dividing all image-label pairs into a training set and a verification set in proportion; preprocessing the image data in the training set to expand the training set to obtain an input image I and a label L; using the input image I and the input language guidance text to train a belt tear open domain detection backbone network to obtain a trained belt tear open domain detection backbone network, the belt tear open domain detection backbone network consisting of multiple modules based on a discrete state optional space model M; using the trained belt tear open domain detection backbone network to detect images of the belt running site to be detected, and outputting tear detection results of the images of the belt running site to be detected. The present invention can efficiently detect belt tear.

The invention relates to the technical field of rail traffic, and aims to solve the problems that an existing embedded rail structure mainly depends on deformation of an elastic material to achieve longitudinal expansion and contraction, the expansion and contraction amount is very limited, and the embedded rail structure cannot be applied to a large-span continuous beam bridge and other line environments needing large longitudinal expansion and contraction amount. The invention provides a graded sliding type continuous supporting track suitable for a long-span bridge and an installation method of the graded sliding type continuous supporting track, the graded sliding type continuous supporting track comprises a track groove, an elastic base plate and a track are installed in the track groove, the elastic base plate continuously supports the track, and a gap between the track and the track groove is filled with an elastic material; the rail groove is formed in the sleeper, the limiting components are arranged on the two transverse sides of the rail groove and used for limiting the transverse direction and the vertical direction of the rail groove, and the rail groove can slide relative to the sleeper in the longitudinal direction. Graded sliding is adopted to combine integral sliding of components in the rail groove and deformation sliding of elastic materials in the rail groove, larger longitudinal expansion and contraction amount can be achieved, and the bridge expansion joint can adapt to a bridge with a longer span.

The present invention discloses a speech synthesis method and system based on speech radar and deep learning, the method comprising: obtaining vocal cord vibration echo signals by radar; obtaining fundamental waves by mode decomposition of radar echo signals, and then performing center clipping on the fundamental waves as speech excitation sources; training a resonance peak estimation model by a deep learning method using a video of the lip area of ​​a speaker when speaking; obtaining resonance peak parameters by using a video of the speaker's pronunciation as input; and finally performing speech synthesis based on the radar echo excitation source model and resonance peak parameters. The method of the present invention can obtain a speech excitation source that retains the continuous variation characteristics of the fundamental frequency and a more accurate resonance peak parameter estimation, and can obtain a more natural synthesized speech through a speech synthesizer.

This invention discloses a low-dose PET / MR denoising method based on individualized brain region partitioning. The method includes: acquiring a low-dose PET image and a corresponding registered T1-weighted MR image for a target brain region; and obtaining a denoised reconstructed PET image based on the low-dose PET image and the MR image using a trained encoder-decoder model. The encoder-decoder model includes an attention branch and a visual Mamba branch, and incorporates the adjacency relationships of static brain structures with neuroanatomical significance into the image feature modeling process. The attention branch is used to model the shape similarity of local metabolic patterns and calculates attention based on the square of the Pearson correlation coefficient, while the visual Mamba branch is used to model global long-range dependencies. This invention, by synergistically combining local structure modeling and long-range dependency modeling, enables the model to restore detailed texture while taking into account overall metabolic distribution characteristics, thereby improving image visual quality while maintaining quantitative consistency at the brain region level.

The invention relates to the technical field of oil product detection, in particular to a diesel oil Raman spectrum detection method and device based on a convolutional neural network, computer equipment and a computer program product. The method provided by the invention comprises the following steps: acquiring a Raman spectrum of diesel oil to be detected; the Raman spectrum is input into a pre-trained diesel oil detection model, a detection result output by the model is obtained, and the detection result represents the alcohol substance doping possibility of the to-be-detected diesel oil; wherein the diesel oil detection model comprises a multi-scale convolution branch structure, and the multi-scale convolution branch structure comprises three parallel convolution branches with different convolution kernel sizes. According to the method provided by the invention, through parallel execution of three convolution branches, feature information of various scales in the spectrum can be synchronously extracted, the recognition and extraction capability of the model on weak signal features is remarkably enhanced, and efficient and accurate recognition of alcohol doping behaviors in diesel oil is realized.

The invention relates to the technical field of image processing, in particular to an image processing method, device and equipment. In the embodiment of the invention, a plurality of image cutting proportions are configured in advance, and a corresponding scanning sequence is configured for each image cutting proportion, so that when the target image is processed, the target image cutting proportion and the corresponding scanning sequence can be selected according to the length-width ratio of the target image; therefore, the target feature vector sequence corresponding to the target image is obtained, the continuity of the visual information of the image is reserved, the visual information of the image is prevented from being damaged, and the understanding effect of the state space model is improved.

The application belongs to the technical field of energy storage modeling, and discloses a large disturbance transient simulation modeling method and system for an electrochemical energy storage system, which comprises the following steps: establishing a slow dynamic model, an alternating current side electromagnetic transient model and a direct current side energy storage battery equivalent circuit model, wherein the alternating current side electromagnetic transient model outputs a fast state variable; decomposing the fast state variable into a quasi-steady component and a deviation component, and establishing a linearized dynamic equation of the deviation component; resetting the deviation component when a grid voltage suddenly changes, superimposing the reset deviation component and the quasi-steady component to obtain an initial value of the fast state variable; and solving the slow dynamic model, the linearized dynamic equation of the deviation component and the direct current side energy storage battery equivalent circuit model simultaneously with the initial value of the fast state variable to obtain a transient dynamic response of the energy storage system in the whole process. The application completely retains the fast dynamic details and state continuity in the initial stage of the fault, and the simulation result is closer to the real physical response.

The embodiment of the invention discloses a construction behavior warning method and device, electronic equipment and a computer readable medium. A specific embodiment of the method comprises the following steps: controlling monitoring equipment in a target area to collect a real-time video as a to-be-detected video at a top view angle; analyzing the to-be-detected video to obtain a to-be-detected video frame sequence; performing target detection processing on each to-be-detected video frame in the to-be-detected video frame sequence to generate target detection information; determining a to-be-detected person image sequence set and a to-be-detected device image sequence set; performing personnel behavior recognition processing on each to-be-detected personnel image sequence to generate behavior detection information of each construction personnel; performing equipment behavior recognition processing on each to-be-detected equipment image sequence to generate behavior detection information of each construction equipment; and controlling alarm equipment in the target area to perform alarm operation. According to the embodiment, the missed detection rate or the false detection rate of abnormal behaviors can be reduced, so that behavior warning is performed in time, and the construction safety is improved.

The application discloses a compression method of encoded video data, which comprises the following steps: extracting at least one first enhancement layer video frame from first encoded video data; re-encoding the first enhancement layer video frame to obtain a second enhancement layer video frame; and taking the second enhancement layer video frame and a base layer video frame in the first video data as second compressed video data, wherein the first video data is scalable video coding data. The compression method of the application does not change the structure of the video data before and after compression, retains the real-time and continuity of the video data, is compatible with the frame rate stability requirements of various players, and reduces the capacity of the video data without losing the video data information.

A wet clutch friction plate oil groove structure based on depth gradient optimization relates to the technical field of wet clutches and comprises a friction product body, a first annular oil groove, a second annular oil groove and a third annular oil groove are sequentially formed in the friction plate body in the radial direction, and a plurality of radial oil grooves distributed in the circumferential direction are formed in the friction plate body in the third annular oil groove. A first V-shaped oil groove is formed between every two adjacent radial oil grooves, the first oil groove and the second oil groove, and a second V-shaped oil groove is formed between every two adjacent radial oil grooves, the second oil groove and the third oil groove. The vertex of the first V-shaped oil groove is located on the first oil groove, two end points of the first V-shaped oil groove are communicated with the intersection point, the vertex of the second V-shaped oil groove is located on the third oil groove, two end points of the second V-shaped oil groove are communicated with the intersection point, and the intersection point is the communication point of the radial oil groove and the second oil groove. The deep oil grooves are formed in the outer ring area so that the heat dissipation efficiency can be improved; a shallow oil groove is adopted in the inner ring area to maintain the structural rigidity, and lubricating oil is guided to flow secondarily so as to further improve the heat exchange effect.

The invention discloses a low-illumination urban rail train key part defect image reconstruction and defect detection method and system based on improved SwinIR. The method comprises the following steps: S1, constructing a data set; s2, constructing an improved SwinIR low-illumination urban rail train key part defect image reconstruction model; s3, designing a loss function, and training the improved SwinIR low-illumination urban rail train key part defect image reconstruction model by using the training set and the verification set to obtain a trained optimal model and weight; then testing the trained optimal model by using a test set, and evaluating a test result; s4, inputting the urban rail train key part defect image into the trained improved SwinIR low-illumination urban rail train key part defect image reconstruction model to obtain a final super-resolution reconstruction image; and S5, inputting the super-resolution reconstructed image into the YOLO detection model for defect detection, and outputting a detection result.