34 results about "Edge filter" patented technology

The invention relates to the field of image processing, in particular to a visual image processing method and device, equipment and a program product. The method comprises the following steps: acquiring a binocular image, and generating a first disparity map through a stereo matching network; obtaining a second disparity map through extreme value filtering; generating an edge mask through edge detection; performing edge filtering on the second disparity map according to the edge mask to obtain a third disparity map; and horizontal displacement is obtained according to a parallax value in the third parallax image, parameters are calibrated through a camera, the horizontal displacement is converted into depth information, and a three-dimensional point cloud is generated according to the positions of the pixel points and the depth information. According to the method, a real boundary is extracted through edge filtering, parallax of a boundary area is forcibly corrected by using an edge mask, a transition zone of network prediction is suppressed, burrs or outliers of point clouds after conversion can be effectively reduced, scattered point cloud noise is eliminated, the object contour boundary is clearer, and the perception precision of a visual system is improved.

The invention discloses a graph-enhanced social robot detection method and device based on reinforcement learning, and the method comprises the steps: 1, extracting metadata information and text information of a user in a social network through a multi-layer perceptron and a pre-training language model, carrying out the feature fusion of the information through a multi-head self-attention mechanism, and generating user representation; 2, performing oversampling on minority class nodes in a potential feature space by using a linear interpolation method based on neighborhood perception, and constructing a balanced training sample set; 3, dynamically adjusting an edge retention threshold value based on the feature similarity and a reinforcement learning strategy, and executing an edge filtering operation to eliminate unreliable connection and optimize a graph structure; and 4, inputting the enhanced node features and the purified graph structure into a graph neural network classifier to realize accurate identification of the social robot nodes. According to the method, the problem of misjudgment caused by class imbalance can be effectively relieved, and the detection accuracy and stability are remarkably improved.

The present invention provides a method and system for detecting defects in automotive lamp lenses based on traditional image processing, belonging to the field of industrial intelligent production inspection technology. The method includes the steps of detecting clutter and debris. Clutter detection involves collecting an original image of the lampshade surface and preprocessing it to obtain a ROI image; adaptively processing the ROI image to determine whether a pixel value greater than a set threshold determines whether clutter exists; and debris detection involves extracting horizontal, vertical, and left and right 45° gradient information from the ROI image, performing binary segmentation on the filtered gradient information using an adaptive threshold to obtain the sixth, seventh, eleventh, and twelfth images, respectively; performing a sum operation on these images to form a final binary image to obtain the fourteenth image; performing edge filtering on the fourteenth image using a custom mask; and determining whether debris exists by calculating the area of ​​each highlighted block in the image. This method effectively improves data processing efficiency and detection accuracy.

The invention relates to a testing device for optically checking objects for authenticity, in particular for checking whether objects have a predetermined, machine-readable code-encompassing security marking with known fluorescence properties. The test device includes a lighting unit designed and arranged to illuminate an object to be tested with electromagnetic radiation - preferably light visible to the human eye - in a first, fluorescence-stimulating wavelength range. The test device comprises an area sensor with a plurality of light-sensitive sensor elements and an imaging optic which is designed and arranged to project an image of an object onto the area sensor, wherein the area sensor is designed to output an image signal representing an image projected onto the area sensor. The imaging optics comprise an optical filter in the form of a high-pass or low-pass filter, in particular an edge filter, which is configured to block electromagnetic radiation in the first, fluorescence-exciting wavelength range, in particular to absorb it, and to transmit light in a second wavelength range, wherein the second wavelength range comprises wavelengths corresponding to expected fluorescence and wherein the second wavelength range does not overlap with the first wavelength range. The test device is designed to detect the insertion of an object, to check images of the object represented by image signals from the area sensor, and to reject an object if one or more test criteria are not met, or to accept it if the test criterion(s) are not met. The testing device is further designed to identify a security marking on an object and to test a found security marking with regard to its fluorescence properties.

The utility model discloses a multi-folding-edge filter element which comprises a filter element whole body, the filter element whole body is provided with a plurality of folding assemblies, the plurality of folding assemblies are formed by folding filter paper, the filter paper is circumferentially surrounded and fixed to form the filter element after being folded, each folding assembly comprises a first folding surface which is folded from outside to the center, and the first folding surfaces are symmetrically arranged; two symmetrical first V-shaped surfaces are folded towards the center of the filter element between the outer ends of the first folding surfaces, a second V-shaped surface is folded towards the outer side of the filter element in the middle of the inner end of each first folding surface, the depths of the first V-shaped surfaces and the second V-shaped surfaces are the same, the folding assemblies form the whole filter element, the contact surface of the whole filter element is 1.5 times more than that of a traditional filter element, and the filtering effect is enhanced.

A Signal-Agnostic Apparatus, System, and Method for Doppler Correction. In one embodiment, for the correction of significant Doppler shifts such as those present in signal transmissions involving satellites in low-earth orbit. The frequency locked loop apparatus and system for correcting Doppler frequency offset comprising an amplitude normalizer, a positive, a negative frequency band edge filter having a plurality of negative band edge coefficients, a positive or negative frequency content filter output for calculating a raw Doppler frequency offset, a leaky integrator, a zero-crossing counter, a plurality of locking filters, a plurality of tracking filters, an array, and a numerically controlled oscillator.

The present invention relates to the field of power line carrier communication technology, and specifically provides a power line carrier signal channel estimation method and device based on OFDM. The method includes obtaining initial channel estimation values corresponding to multiple subcarriers based on a target signal of a target channel; performing edge filtering on the initial channel estimation values corresponding to the target subcarriers based on the initial channel estimation values corresponding to the target subcarriers and adjacent initial estimation values to obtain denoised channel estimation values corresponding to the target subcarriers; wherein the target subcarriers are at least part of the multiple subcarriers, the adjacent initial estimation values are initial channel estimation values corresponding to the adjacent subcarriers, and the adjacent subcarriers are subcarriers adjacent to the target subcarriers; and obtaining a channel estimation sequence of the target channel based on the denoised channel estimation values corresponding to the target subcarriers. The present invention effectively improves the accuracy and robustness of the channel estimation results through edge filtering.

The embodiment of the present application discloses an infrared image multi-edge filtering denoising method for suppressing salt and pepper noise and keeping weak edge details. The method comprises the following steps: acquiring original data of an infrared image; correcting the original data of the infrared image to obtain corrected data; calculating a salt and pepper noise suppression parameter; calculating a weak edge detail keeping weight; performing multi-edge filtering on the corrected data by using the salt and pepper noise suppression parameter and the weak edge detail keeping weight to obtain a filtering result; and outputting the filtering result. By implementing the method of the embodiment of the present application, the interference of salt and pepper noise can be effectively suppressed, and the method is superior to bilateral filtering denoising in keeping weak edge details, has a short calculation time, and is strong in real-time performance.

View-invariant edge filtering is provided. In one or more implementations, view-invariant edge filtering may provide anti-aliasing that is effective in reducing or eliminating perceivable aliasing across various different viewing configurations of display content, including foveated rendering configurations and / or viewing configurations in which display content is moveable and rotatable in three virtual dimensions.

An embolic protection device includes a frame, an edge filter screen, and a plurality of supporting rods that are arranged at intervals. Each supporting rod includes a first tail end, a second tail end, and at least one head end that faces a distal end. The first tail end and the second tail end are separately connected to two opposite sides of the frame. The first tail end and the second tail end separately form a first line segment and a second line segment with the head end. The first line segment extends in a direction from the first tail end to the head end and gradually moves away from the frame, and the second line segment extends in a direction from the second tail end to the head end and gradually moves away from the frame.

The utility model is suitable for mould head technical field provides mould head quick screen change self -cleaning device, its mould head body has through installation chamber, built -in coaxial fixed installation carrier, and the carrier side wall opens two groups of parallel limit guide slot, main filter screen, spare filter screen sliding fit among them, all are board -like and with guide rail contact; The mould head body top is equipped with electric push rod, and the movable rod is connected with the receiving plate, and the receiving plate both sides extend fixed block, its abutting face fixed heating type blow head, this device is equipped with main, spare filter screen in installation carrier, and the main filter screen can quickly use the spare filter screen when it is out of order, guarantees the continuous stability of production, avoids single filter screen damage and causes production stagnation, quality fluctuation, and can clean different side edge filter screen flexibly, simplifies the impurity cleaning procedure, improves the cleaning efficiency, reduces the operator's burden, and improves the equipment maintenance efficiency.

An indication of one or more weighting parameters is determined for use in applying upsampling to input pixel values representing an image region to determine a block of one or more upsampled pixel values. A horizontal edge filter determines a first filtered value. A vertical edge filter determines a second filtered value. A horizontal line filter determines a third filtered value. A vertical line filter determines a fourth filtered value. The first, second, third and fourth filtered values are used to determine the indication of one or more weighting parameters, wherein the one or more weighting parameters are indicative of relative horizontal and vertical variation of the input pixel values within the image region. The determined indication of the one or more weighting parameters is output for use in applying upsampling to the input pixel values representing the image region to determine a block of one or more upsampled pixel values.

The invention relates to the technical field of signal filtering, and discloses a filtering circuit and an MCU, and the filtering circuit comprises a rising edge filtering unit, a falling edge filtering unit and a signal output unit. The signal input end of the rising edge filtering unit is electrically connected with the signal input end of the falling edge filtering unit; in actual use, the rising edge filtering unit is used for filtering glitch signals of the rising edge and filtering high-level signals within the first preset time T1, and the falling edge filtering unit is used for filtering glitch signals of the falling edge and filtering low-level signals within the second preset time T2. And the output signal is recovered through the signal output unit, so that the rising edge burr signal and the falling edge burr signal can be filtered at the same time.

The application relates to the technical field of intelligent inspection, in particular to a data center robot inspection scheduling system based on AI, which comprises a multi-modal sensing and collecting module for acquiring environment point clouds and visual images, an edge filtering and preliminary judgment module for generating positioning coordinates and extracting abnormal signals by using a filtering algorithm, a heterogeneous collaborative scheduling module for scheduling heterogeneous robots to perform re-inspection based on the abnormal signals, and a twin mapping and closed loop module for updating a twin model and generating a treatment work order.In the application, a heterogeneous robot collaborative architecture is constructed, the complementary advantages of wheeled and track robots are utilized to realize global coverage, a multi-sensor filtering fusion algorithm is adopted to suppress signal noise, and an edge computing and cloud linkage mechanism is combined to realize millisecond-level preliminary judgment of faults and automatic relay across devices, so that the operation and maintenance response speed and decision accuracy are improved.

The present invention relates to the field of vibration detection technology, and includes a locomotive vibration performance assessment method and system based on locomotive measurement traceability, comprising the following steps: obtaining reference light and measurement light; modulating the measurement light according to a pulse modulation signal using an acousto-optic modulator to obtain a measurement pulse light; calculating a reverse signal of the sound signal to drive a piezoelectric driver to vibrate a detection optical fiber; directing the measurement pulse light into the detection optical fiber, deriving a back-Rayleigh signal generated by the measurement pulse light in the detection optical fiber, and obtaining an anti-ambient sound back-Rayleigh signal; performing guided edge filtering on the anti-ambient sound back-Rayleigh signal to obtain a filtered back-Rayleigh signal; coupling the filtered back-Rayleigh signal with the reference light to obtain an electrical signal; converting the electrical signal into a vibration frequency signal using an IQ demodulator, evaluating the signal using an assessment model, and assessing the vehicle health status based on the vibration frequency signal. The present invention can improve the accuracy of locomotive vibration performance assessment.

The invention provides a cross-domain data feature extraction method and device, equipment and a storage medium, and the method comprises the steps: obtaining multi-modal cross-domain data; representing the cross-domain data of each mode, and carrying out feature extraction on the represented cross-domain data to obtain reference feature information; constructing an initial graph structure based on the cross-domain data and the reference feature information; carrying out edge screening on the initial graph structure based on the similarity between the nodes in the initial graph structure to obtain an information distribution graph; and performing feature extraction on the information distribution map based on a pre-trained map neural network model to obtain fusion feature data. Through the technical scheme of the invention, feature extraction can be quickly and effectively carried out on multi-modal cross-domain data.

The present invention discloses a depth map optimization processing method, intelligent terminal, and storage medium. The method comprises: obtaining an original depth map, performing hole filling, preliminary guided filtering, and preliminary smoothing on the original depth map to obtain a preliminary depth map; performing edge filtering on the preliminary depth map to obtain an intermediate depth map; and performing further smoothing and median filtering on the intermediate depth map to obtain an optimized depth map. By performing two filtering and two smoothing operations on the depth map, the present invention ensures the integrity of depth edges and avoids leaking and false blurring in the image blurring effect.

The application relates to a touch position determination method and device, computer equipment, a touch device, a storage medium and a computer program product. The method comprises the following steps: acquiring touch position information corresponding to a touch operation; determining a current edge filter coefficient corresponding to the touch position information; determining a target filter coefficient of the touch position information according to an overall filter coefficient of a touch area and the current edge filter coefficient; the overall filter coefficient is a filter coefficient corresponding to the touch area; and performing a filter operation on the touch position information by using the target filter coefficient to obtain updated touch position information. The touch operation effect can be improved by using the method.

The invention discloses a homotopy guided robot safe motion planning method and device in an uncertain environment, and relates to the field of robot trajectory planning, and the method comprises the steps: obtaining a generalized Voronoi diagram of a discretization space; the generalized Voronoi diagram comprises a plurality of nodes which are equidistant from the nearest obstacle, and the adjacent nodes are connected with each other; determining a space size attribute and a feature attribute of each node; determining feature nodes based on the space size attribute and the feature attribute of each node; constructing a feature graph based on the feature nodes; performing edge filtering on the feature map to obtain an updated feature map; performing feature map search on the updated feature map to obtain a region of interest; the feature map search comprises a starting position and a target position of a task; and performing non-uniform search based on the region of interest to obtain an optimal path. According to the method, the mobile robot can efficiently search the optimal and belief-feasible path in a complex environment.

The invention discloses a temperature difference compensation method and system for an edge filtering strain sensor, and the method comprises the steps: transmitting a light beam through an ASE light source, filtering the light beam through an intelligent filtering unit, and outputting a linearly and dynamically adjusted light signal; an optical signal is divided into three paths of equal optical signals through an optical fiber coupler, and the three paths of equal optical signals are respectively input into a three-grating sensing array to obtain a three-grating reflection optical signal; the three-grating reflection signals are converted into electric signals through a three-channel photoelectric detection array, and the electric signals of the three channels are synchronously acquired through an acquisition unit; preprocessing the acquired electric signals of the three channels to obtain calibrated signal increments of the three channels; and outputting accurate strain data through dynamic differential operation and multi-dimensional calibration. According to the temperature difference compensation method and system of the edge filtering strain sensor, strain signals and temperature interference are efficiently separated, and the application limitation of a traditional static compensation scheme is broken through.

A system and method are described that dynamically alter hierarchical data structures displayed on a user interface. The system accesses data associated with a plurality of assets, each having a value and binary attributes, and constructs a directed graph comprising vertices and edges based on those attributes. An active core of the graph is determined by analyzing vertex values and selecting a subset of vertices and edges in accordance with user input. The display is updated to reflect the active core, and edge filters and edge score evaluations (ESEs) are applied to refine the selected edges for inclusion. The transitive edge removal policy (TERP) ensures that undesired transitive edges are removed from the structure. The display dynamically updates over time to reflect changes in the underlying data, enabling users to interact with and modify the graph structure based on new insights and changing conditions.

The invention relates to an edge contour enhanced EA-LPNet (LPNet integrating EMA and ACmix) monocular depth estimation method, which comprises the following steps of: 1, constructing a multi-scale network structure by utilizing a Laplacian pyramid, and putting forward and using an improved cavity space pyramid pooling module on an encoder side; step 2, processing edge contour information by using a specific edge filtering operator, and fusing the edge contour information into the network; step 3, utilizing an efficient multi-scale attention mechanism EMA module to splice the edge residual image, the potential feature image and a previous-level depth residual image; 4, replacing a convolution module in a traditional decoder with a self-attention convolution module ACmix, and obtaining potential features in the depth residual image subjected to pre-dimensionality reduction in advance; and step 5, proposing and using an edge information loss function, and designing a specific loss function combination strategy for training. According to the method, the small target identification capability and the performance of the algorithm in the aspect of edge contours are effectively improved, and the network precision is improved to a certain extent.

A filter holding device includes an opening portion and a filter. The opening portion communicates an inside of the filter holding device with an outside of the filter holding device, and has a rim that extends in a communication direction from the inside of the filter holding device toward the outside of the filter holding device. The filter is disposed in the opening portion, and has a first side and a second side. The first side faces the inside of the filter holding device. The second side faces the outside of the filter holding device. The filter has a larger weight density per unit volume at the second side facing the outside of the filter holding device than at the first side facing the inside of the filter holding device in a state in which the filter is disposed in the opening portion.

The present invention discloses a demodulation method for a quasi-distributed FBG sensor based on a novel edge filter, belonging to the technical field of FBG sensor demodulation. The demodulation system includes a pulse laser, an optical fiber coupler, an optical fiber connector, a single-mode optical fiber No. 1, an edge filter, a single-mode optical fiber No. 2, a fiber Bragg grating No. 1, a fiber Bragg grating No. 2, a photodetector, a voltage amplifier, an A / D converter, and a data collector. A novel edge filter is used for demodulation of the fiber Bragg grating sensor. By changing the structural parameters of the edge filter, it can adapt to fiber Bragg gratings with different central wavelengths. The reflected light of the FBG array is converted from a wavelength signal to an energy signal after passing through the edge filter, thereby realizing FBG array sensor demodulation. Ultimately, fast demodulation of the FBG with a wide range, low noise, high sensitivity, high stability, and low cost is achieved.

The present application relates to a method, apparatus, computer device, storage medium, and computer program product for visualizing edge data. The method comprises: performing triangulation processing on the equipment data of an electric power device to obtain a triangulated network of the electric power device; constructing a Voronoi diagram of the electric power device based on the triangles in the triangulated network; performing edge filtering processing on a three-dimensional mesh model obtained based on the Voronoi diagram to obtain a target three-dimensional mesh model of the electric power device; and obtaining a visualization simulation result of the electric power device on the edge based on the coordinate data of points on the edge of the target three-dimensional mesh model. This method can improve the visualization effect of the simulation data of the edges of the target three-dimensional mesh model of the electric power device.

The present invention discloses a phase unwrapping method, apparatus, terminal device, and storage medium based on Poisson correction, relating to the field of interferometric measurement. The method comprises: obtaining an interferometric phase map to be processed at a target position, inputting the map into a winding number prediction model to generate a predicted winding number; calculating a winding number gradient based on the predicted winding number, and performing a lower-edge filtering correction to obtain a corrected winding number gradient; performing a discrete Fourier transform on the horizontal component of the corrected winding number gradient to obtain a first frequency domain transform result; performing a Laplace operator calculation on the corrected winding number gradient and then performing a discrete Fourier transform to obtain a second frequency domain transform result; performing an inverse discrete Fourier transform on the sum of the two frequency domain transform results divided by a preset frequency domain adjustment factor to obtain an error winding number; determining the true phase based on the predicted winding number and the error winding number, and then calculating the actual elevation of the target position. By implementing the present invention, the accuracy of the unwrapped phase can be improved.

The application discloses a low-illumination image denoising method based on phase correlation, which comprises the following steps: extracting image edge information, performing edge filtering on each frame of image to extract effective edge information of the image; performing edge filtering on two adjacent frames of image, performing phase correlation operation on the two frames of image after filtering to obtain displacement of the two frames of image; performing homologous signal constraint and global correlation constraint on the multiple frames of image after registration to construct cross-correlation equation; solving the cross-correlation equation according to cross-correlation maximization optimization equation and gradient maximization equation, and retaining the solution of the current frame to realize iterative update of the denoised image in the cross-correlation equation of the next frame. The application can effectively remove various noises of the low-illumination image under the low-illumination scene while considering the calculation speed and algorithm stability, restores the real scene, makes the scene clear, and enhances the details.

A light source module (100) with integrated wavemeter components (460, 494, 495) for stabilizing the output power and wavelength of a superluminescent diode or other broadband semiconductor light source (121) outputting a broadband output beam. A portion of the source output beam is directed to an optical edge filter (460) with a cross-over wavelength lying within the bandwidth of the output beam. The edge filter (460) divides the light it receives into a short-wavelength component and a long-wavelength component. These two components are then directed onto respective photodetectors (494, 495) that output respective signals to a wavemeter controller. The controller adjusts the drive current and / or temperature of the source to maintain the mean wavelength of the source's output beam at a set constant value according to a control parameter determined from a combination of the photodetector signals such as their ratio or the ratio between their difference and sum.

The invention discloses a hardware implementation method of a time-space domain filtering system combined with image information. The hardware implementation method comprises the following steps: preprocessing an image by adopting downsampling layering and upsampling phase difference; processing the data stream of each layer of original image preprocessing graph in a pipeline form to obtain four-direction edge filtering image video stream data; performing read-write processing on each layer of filtered image video stream data to obtain multi-frame image cache data; performing motion compensation denoising on each layer of filtered image in a time domain; performing kernel convolution on the data video stream after motion compensation denoising through a bilinear interpolation formula in a pipeline form to obtain image data recovered to an original resolution, and fusing video stream data of each layer; and processing the fused image data in an assembly line form to obtain image video stream data after spatial domain processing. According to the method, the detail information in the image is reserved as much as possible while the image noise is suppressed, and the moving target trailing phenomenon in the image scene is effectively suppressed under the condition of the moving scene.

The invention discloses a collaborative defense and attack behavior cognition method for a multi-robot system, and the method comprises the steps: obtaining the suspicious traffic intensity of each robot defense node, and constructing a collaborative filtering architecture; defining an edge / cloud shunting proportion for each defense node, establishing an edge filtering time delay model and an individual filtering cost function, modeling a collaborative filtering process under multi-node resource sharing into a selfish filtering game, and utilizing a load balancing property of edge filtering time delay consistency under balance to obtain a cloud filtering game; designing a distributed iterative algorithm to solve a unique Nash equilibrium shunting strategy in limited iteration; and constructing an optimal model of the attacker budget constraint under the defense balance strategy to solve an optimal attack strategy for performing cognitive prediction and early warning on the attack intensity, the target preference and the attack type. According to the invention, the autonomous decision-making capability of the multi-robot system in a complex confrontation environment can be improved. The method can be widely applied to the technical field of multi-robot cooperative control.