79 results about "Spatial frequency" patented technology

The application provides a multi-region brightness adjustment method of a display device, comprising: obtaining a PWM pulse starting time interval of adjacent partitions from a multi-partition PWM dimming system, and calculating a light-dark alternation window length according to the PWM pulse starting time interval; matching a spatial frequency change trend with a preset time and spatial frequency joint domain to obtain a phase offset range in which visible flicker stripes are located; performing periodic phase sampling on each partition PWM signal in the phase offset range to obtain a periodic phase offset record of each partition; obtaining low gray scale picture data from a current display frame of the display device, aligning the periodic phase offset record of each partition with the low gray scale picture data on a partition-by-partition basis, identifying a partition in which a brightness ripple exceeds a preset threshold, reallocating a PWM phase in the partition, and obtaining an adjusted phase offset.

This application discloses a method and system for synchronous testing of metasurface grating waveguide performance, relating to the field of metasurface grating technology. The method includes polarization state modulation and spatial frequency encoding of multi-wavelength incident beams to obtain a modulated incident light field; coupling this modulated incident light field into the metasurface grating waveguide under test, which then undergoes total internal reflection to form the outgoing light field; acquiring a raw dataset of the entire light field, including Stokes polarization images and spatial spectrum images corresponding to each wavelength of the outgoing beam; calculating the total internal reflection hop count based on the Stokes polarization image; calculating the spatial step size offset based on the spatial spectrum image; and establishing a mapping relationship between the image pixel coordinates of the raw dataset and the internal physical regions of the metasurface grating waveguide under test to obtain local diffraction efficiency distribution data. This application solves the problem that traditional outgoing end intensity testing is difficult to trace to local defects inside the waveguide, achieving synchronous, high-precision quantitative characterization of local diffraction efficiency across multiple wavelengths.

This invention discloses a three-dimensional contour measurement method and apparatus based on multi-projection gratings, relating to the field of three-dimensional topography measurement technology. The method includes: obtaining intrinsic and extrinsic parameters through calibration; projecting a set of speckle patterns; preliminarily analyzing the surface topography gradient of the object to determine a non-uniform spatial frequency sequence; sequentially projecting and acquiring multiple sets of phase-shifting grating patterns, and calculating the wrapping phase map corresponding to each frequency; for the same pixel, based on its reliability assessment in wrapping phase maps at different frequencies, assigning confidence weights to the phase values ​​of each frequency, and calculating and optimizing the phase map and the corresponding fused confidence map; calculating gradient information and combining it with the fused confidence map to perform region division, and executing a region-guided path consistency verification unfolding algorithm to obtain the absolute phase; converting the absolute phase into three-dimensional point cloud coordinates of the object surface. This application aims to solve the problems of complex topography, discontinuities, and noise easily leading to unfolding path errors and order jumps during phase unfolding.

The application discloses a device and method for measuring welding surface temperature, and belongs to the technical field of non-contact temperature measurement, and comprises the following: a short-depth-of-field optical imaging unit for collecting high-pass spatial frequency filtering; the short-depth-of-field optical imaging unit is arranged on one side of a welding surface to be measured; a dichroic light splitting unit for separating incident light into, and wavelengths; the dichroic light splitting unit is located at one end of the short-depth-of-field optical imaging unit away from the welding surface to be measured; and a first photoelectric collection unit, a second photoelectric collection unit and a third photoelectric collection unit for realizing synchronous collection of, and wavelength radiation energy respectively. The application can obtain the welding surface temperature in real time, and improve the stability and accuracy of the infrared light temperature measurement system.

The present application relates to the field of optical imaging technology, and particularly relates to a full-depth-range high-lateral-resolution OCT imaging method, system and storage medium based on interference synthetic aperture. The OCT imaging method comprises sample positioning: moving the point spread function of the sample to a region close to 0 optical path; collecting three-dimensional original data; pre-processing the collected three-dimensional original data to obtain complex signals containing amplitude and phase information; performing three-dimensional Fourier transform on the scattering potential distribution based on a synthetic aperture algorithm to obtain frequency domain signals, correcting the nonlinearity of the spatial frequency by interpolation or frequency domain filtering before interpolation, and then performing inverse Fourier transform to reconstruct the spatial distribution of the scattering potential and obtain three-dimensional image data. The detail information of the defocus region is recovered, high-resolution imaging in the full depth range is ensured, and the inherent contradiction between lateral resolution and focal depth is broken through.

A display system and methods for providing visual stimuli to a user to improve conditions related to amblyopia. The display system comprises a display device configured to provide left and right display elements and a computer unit for providing the visual content to the display device. The computer unit is configured to provide to the left display element left visual content comprising a left base layer and a left stimulation layer and to the right display element right visual content comprising a right base layer and a right stimulation layer. The left stimulation layer and / or the right stimulation layer is / are implemented with one or more mix functions to display a grating image on the display device. The mix functions are configured to exhibit spatial frequency and drift velocity so the grating image of the left stimulation layer and / or the right stimulation layer moves continuously across the display device.

This invention discloses an image semantic segmentation processing method with adaptive context information extraction, belonging to the field of image processing technology. The specific steps of the method are as follows: First, basic features in three domains—spatial, frequency, and semantic—are collected and regularized into an original set. Then, cross-domain mapping is used to generate globally interconnected and fused features. Next, based on regional attributes, the features are decomposed into low-frequency structures and high-frequency detail components, and their proportions are adjusted to extract complete contextual features. Finally, after consistency verification and optimization of the three-domain information, pixel category determination is completed, and a high-precision semantic segmentation result is output. This invention solves the core technical problems of insufficient information coordination, difficulty in balancing global structure and local details, and inability to correct information conflicts between dimensions in traditional methods through a full-chain adaptive information processing mechanism. This improves the accuracy and regional clarity of semantic segmentation, making the segmentation results more consistent with actual scene requirements.

This invention discloses a single-snapshot spatial frequency domain imaging and morphology correction method, system, and device, comprising: acquiring a set of monochromatic fringe images with different center wavelengths based on the same projected image; simultaneously inputting each monochromatic fringe image in the set into two parallel and independent deep learning networks, wherein the first deep learning network is used to extract the DC and AC components of the image; the second deep learning network is used to extract the wrapping phase component and fringe hierarchy information of the image, and perform absolute phase calculation to obtain the three-dimensional morphology before correction; performing height correction on the three-dimensional morphology and calculating the corrected diffuse reflectance image; performing background segmentation on the corrected diffuse reflectance image and spatially aligning it to the same coordinate system; and retrieving the optical characteristic parameters and physiological parameters of the tissue based on the spatially registered multi-wavelength diffuse reflectance data. This invention ultimately achieves real-time, wide-field, high-precision, and highly robust quantitative imaging of tissue physiological parameters.

Provided is an anti-glare film excellent in anti-glare properties and capable of suppressing reflected scattered light. An anti-glare film including an anti-glare layer, the anti-glare film having an uneven surface, wherein for amplitude spectrum of elevation of the uneven surface, when a sum of amplitudes corresponding to spatial frequencies of 0.005 μm−1, 0.010 μm−1, and 0.015 μm−1 is defined as AM1 and an amplitude at a spatial frequency of 0.300 μm−1 is defined as AM2, AM1 is 0.070 μm or more and 0.400 μm or less, AM2 is 0.0050 μm or more, and AM2<AM1.

A spatial image simulation method includes the following steps: constructing a DMD-based optical system simulation model; uniformly sampling the angle and intensity of illumination light within the illumination numerical aperture, and selecting a simulation area; for each illumination angle, simulating the complex amplitude distribution of the micromirror in the open state and the complex amplitude distribution of the micromirror in the closed state; obtaining the interference data of the micromirror at the entrance pupil plane for each illumination angle; obtaining the spatial frequency spectrum at the exit pupil plane for each illumination angle; simulating the corresponding light intensity distribution based on the complex amplitude distribution for each illumination angle, and superimposing the light intensities corresponding to all illumination samples to obtain the final light spot result. Based on the actual working scenario of the DMD, and considering that the tilt angle of the DMD micromirrors has a significant impact on its diffraction, the introduction of various tilt-induced phase differences and tilt factors results in more accurate simulation effects.

Various implementations disclosed herein include devices, systems, and methods that adjust image content to reduce motion sickness during an experience. For example, an example process may include determining a first zone and a second zone of a display, generating images of a three-dimensional (3D) environment, identifying image content of each of the images corresponding to the second zone of the display, and reducing at least one of contrast or spatial frequency of the image content of each of the images corresponding to the second zone of the display.

The utility model relates to a light guide body for scanning system, include: light -receiving surface, microstructure surface and light -emitting surface, the left end surface of light guide body is the light -receiving surface that receives incident light source, the front end surface of light guide body and with light -receiving surface adjacent is microstructure surface, the rear end surface of light guide body and with light -receiving surface adjacent, with microstructure surface symmetry is the light -emitting surface, microstructure surface is equipped with scattering area, microstructure dot area and reflection inhibition area, microstructure dot area surface includes multiple circular or oval dot, the utility model discloses through scattering area and microstructure dot area design random disturbance dense lattice adjustment light -emitting, break spatial frequency resonance to reduce the local light -emitting anomaly caused by dispersion and further realize more uniform light -emitting performance, through microstructure dot area and reflection inhibition area reduce the waveform of tail reflection light causes and further realize more uniform light -emitting performance, the adjustment of microstructure dot shape reduces the local light -emitting anomaly caused by dispersion and further realize more uniform light -emitting performance.

The application belongs to the field of intelligent manufacturing, and discloses a bionic body manufacturing method, device and system supporting remote customization of a user, electronic equipment, a storage medium and a computer program product, solves the problem that the finished product does not meet the expectation due to image transmission distortion in remote customization, and the method comprises the following steps: acquiring customized images (previously obtained by the user by shooting a reference object and a first calibration card by using a first camera) shot from multiple angles; the first calibration card comprises a color calibration card and a skin calibration card; a calibration body is arranged in the color calibration card and / or the skin calibration card, and is used for calibrating an illumination condition and a camera spatial frequency response; for each angle, imaging scene parameters on the user side are generated according to the customized images; cross-scene calibration parameters are determined according to the imaging scene parameters on the user side and the manufacturing side; after the reference object images in the customized images are corrected according to the cross-scene calibration parameters, bionic body 3D modeling is performed according to the reference object images of each angle to manufacture a bionic body.

An ultrasound diagnostic apparatus includes an information processing unit executes processing of acquiring spatial frequency distribution data in a depth direction for Doppler image data, and processing of performing filter processing, which is based on filter characteristics determined in accordance with the spatial frequency distribution data and characteristics of the transmitted ultrasound, on the Doppler image data. The information processing unit searches for a spatial frequency corresponding to a maximal value of a distribution indicated by the spatial frequency distribution data in a search range determined in accordance with a pulse width, and obtains characteristics for suppressing a level of the spatial frequency distribution in a spatial frequency band including the spatial frequency that is searched for.

The present application relates to the technical field of image processing, and particularly relates to an image super-resolution reconstruction method and system based on wavelet transform fusion of space domain and frequency domain, inputting a low-resolution image to be processed into a pre-trained target model, and obtaining a corresponding super-resolution reconstruction image by using the target model, wherein the target model comprises: a convolution layer for extracting shallow features of the low-resolution image, a plurality of space frequency collaborative groups for extracting deep features of the image, a reconstruction output layer for performing super-resolution reconstruction based on the shallow features and the deep features of the image, the plurality of space frequency collaborative groups form a cascaded structure, and each space frequency collaborative group comprises: a residual mixed attention group for spatial context aggregation, and a progressive frequency block for band decomposition and collaborative enhancement of spatial aggregation features based on wavelet transform. The present application decouples and collaboratively enhances the image features in the dual domains by using spatial window attention and wavelet frequency, so as to realize high-quality image super-resolution reconstruction.

This invention proposes a method and system for diagnosing DAS wellbore flow stability based on the square coherence coefficient distribution, belonging to the field of fluid acoustics technology. The method includes: acquiring raw time-series acoustic signals related to wellbore flow using DAS to obtain multi-channel raw time-series DAS data; preprocessing the data and segmenting it according to sliding time windows to obtain multiple time window samples; determining at least one reference channel according to a preset reference channel selection rule; calculating the square coherence coefficient between the reference channel and channels at various spatial locations within the target well section, and constructing a time-space-frequency distribution; extracting high-coherence regions and one or more stability features within the target frequency band; constructing a flow stability index based on the features and performing graded flow stability diagnosis; this invention also discloses a system and storage medium for executing the method, applicable to wellbore or pipeline flow stability diagnosis and abnormal section identification scenarios.

This invention discloses a dynamic scattering medium imaging method based on Mamba channel feature dependency and dual-domain learning, belonging to the field of scattering medium imaging technology. It employs a ScatMamba network with an encoder-decoder architecture to extract initial features from speckle images formed by dynamic scattering media. Multi-scale high-level semantic features are extracted through downsampling and channel filtering enhancement of the Mamba module, and frequency domain global feature modulation is achieved using a spectral modulation module at the encoder bottleneck layer. The decoder fuses multi-scale features through upsampling and skip connections, outputting a clear reconstructed image. End-to-end training is completed using a spatial-frequency dual-domain composite loss function. This invention achieves joint modeling of channel feature dependency and spatial long-range dependency through channel filtering and enhanced Mamba layers, utilizes a spectral modulation layer for adaptive control of high and low frequency components, and strengthens the robustness of the reconstruction mapping through a dual-domain loss function. This enables efficient decoupling of speckle features in dynamic scattering environments, restoring the global structure and fine texture of the target.

A stereoscopic image display of the disclosure includes a display panel and a parallax barrier. The display panel includes a plurality of pixels and displays a plurality of parallax images forming a stereoscopic image. The parallax barrier is disposed facing the display panel and includes a shielding part having a surface on which a pattern of design is formed and a plurality of openings through which light emitted from the display panel passes. In an environment in which the stereoscopic image is observed, a spatial frequency of the pattern of design is set within a visible range determined based on a contrast sensitivity function.

This invention proposes a method for simulating infrared turbulence degradation images based on Unreal Engine. The steps are as follows: constructing a 3D mountain simulation scene based on Unreal Engine; generating a blurred kernel texture image based on a calculated 2D point spread function; rendering the 3D mountain scene using Unreal Engine, convolving the rendered result with the blurred kernel texture image, and obtaining the simulated infrared turbulence degradation image through the convolution result. This invention achieves real-time turbulence blurring of the mountain scene by performing parallel convolution on each frame of the rendered 3D mountain scene with the blurred kernel texture image generated based on the 2D point spread function. The obtained real-time turbulence blurring effect preserves the details of the terrain well. It also considers the influence of the spatial frequency response information of the imaging system itself and the attenuation effect of atmospheric turbulence on image detail transmission, effectively improving simulation accuracy. Furthermore, parallel convolution reduces the complexity of generating the real-time turbulence blurring effect of the mountain scene, exhibiting good real-time performance.

The application discloses an infrared small target false alarm memory distillation detection method and device, and belongs to the technical field of infrared image intelligent perception and light-weight model training. The method acquires an infrared image sample and small target labeling thereof, and constructs a frozen teacher detection model and a light-weight student detection model to be trained; a hard sample weight map is generated according to a teacher-student prediction difference, a target neighborhood low confidence response, a student-only background false alarm, a high-frequency disturbance and a false alarm connected region memory queue; under teacher confidence gating, target region reinforcement learning and background region suppression learning are separately executed, and multi-scale features are subjected to spatial weighting and local high-frequency constraint; combined with background consistency, student-only false alarm punishment and high-frequency false alarm suppression loss, the student model is optimized through three stages of detection ability establishment, spatial frequency migration and false alarm reinforcement suppression. When deployed, only the light-weight student detection model outputting a small target mask, a center point or a target frame after training is retained. The application can concentrate training on difficult-to-detect small targets and background regions prone to false alarms, and reduce false detection under a complex background while maintaining edge real-time performance.

Diagnosis of prostate cancer often involves invasive measures, such as biopsies, in which tissue is removed and graded by a pathologist. Non-invasive measures, such as multi-parametric MRI, can be used to examine lesions and grade them using the PI-RADS scoring scale; however, achievable image resolution is limited by motion. A new paradigm, magnetic resonance histopathology (MRH), is disclosed, which aims to evaluate tissue texture at sub-millimeter resolution through a fast clinical acquisition process. The ability of MRH to identify cancerous tissue in the prostate through computer simulation analysis is demonstrated, reproducing the MRH measurement process on a set of high-resolution, histology slides annotated by pathologists (2, 4, 6, 8, 10, 12, 14). A dataset of spectral intensities (20) at sub-millimeter wavelengths is created, and a deep learning model (22) trained to classify the spectral data is based on normal or tumor tissue. A set of spatial frequencies is identified, which is used to optimize diagnostic capability under the constraint of limited acquisition time. In addition to single-region classification, the disclosed method and architecture integrate spatial context and local information, the inclusion of which improves model performance and denoises inference results. In addition to algorithms for estimating the physical length scale of lesions identified by the model, the trained model is demonstrated to be applied to unlabeled high-resolution 3D MRI data.

This invention provides an adaptive image focus stacking method, system, and pathological slide scanner. It acquires multiple frames of original images from different focal planes within the same field of view, performs perceptual preprocessing on each frame, conducts multi-scale spatial frequency evaluation on each preprocessed frame, performs adaptive raster registration based on the multi-scale weight map, performs intelligent decision fusion on the corrected frames, and performs spatial consistency verification on the fused image, outputting a final full-area sharp focus stacked image. This invention offers advantages such as adaptive processing of high dynamic range images, improved robustness of sharpness evaluation, optimized alignment efficiency, and improved fusion effect, thereby generating high-quality full-area sharp focus stacked images.

The application discloses a multi-domain modeling and semantic embedding enhanced lightweight small target detection method. The multi-domain modeling and semantic embedding enhanced lightweight small target detection method comprises the following steps: acquiring a to-be-detected image; inputting the to-be-detected image into a trained lightweight detection model, so as to determine a backbone feature map of the to-be-detected image through a plurality of high-frequency residual enhancement spatial depth conversion convolution units and a plurality of spatial-frequency context enhancement layer aggregation units; fusing the backbone feature map through a feature pyramid fusion method to obtain a fused feature map; determining a neck feature map based on the fused feature through a plurality of frequency-guided semantic injection units; and detecting the neck feature map through a plurality of detection heads to obtain a detection result of the to-be-detected image, thereby improving the detection precision of small targets.

The application relates to an AI vision-based solid wood board defect intelligent detection technical service platform system, aiming to solve the problems of strong texture interference, low micro-defect recognition rate and poor industrial environment robustness. The system comprises a multi-dimensional perception image acquisition unit, an environment self-adaptive enhancement unit, a double-flow feature decoupling analysis unit, a defect fine classification unit, an edge reasoning execution unit and a cloud collaborative service unit. The texture interference is inhibited through spatial-frequency double-flow decoupling, high-precision defect recognition and full-link closed-loop sorting are realized by combining cloud-edge collaboration and federated learning, and the detection robustness, raw material utilization rate and cross-line collaborative optimization capability are improved.

The application discloses a near-field target positioning method based on atomic norm space frequency modeling. The method establishes a near-field array receiving signal model, then carries out a space frequency structural representation of the near-field signal, then carries out a meshless sparse space frequency representation based on atomic norm minimization, then carries out frequency ridge line extraction and linear fitting, and finally realizes joint estimation of angles and distances and system output. The application fully utilizes the space frequency variation law of the near-field signal, obtains the angle and distance information of the target by extracting the main frequency ridge line and carrying out linear fitting, can realize high-precision and low-complexity joint estimation of angles and distances, has significant advantages in parameter estimation precision, robustness and calculation efficiency, and can reduce the influence of noise and window effect on the estimation precision, thereby providing a high-precision and easy-to-implement technical scheme for near-field target positioning in large-scale MIMO systems and 6G communication.

This invention relates to a highly robust image recognition method based on a frequency-domain cross-domain attention mechanism, belonging to the fields of artificial intelligence and computer vision. The method constructs a full-frequency-domain processing architecture within the network, utilizes discrete cosine transform to achieve dual embedding of space and frequency, introduces a frequency grouping strategy and a cross-domain attention mechanism, and combines adaptive frequency filtering to achieve deep spatial-frequency interaction and noise-resistant extraction of image features. This invention achieves inherent robustness without requiring expensive adversarial training, improves the diversity and anti-interference ability of feature extraction, breaks the trade-off between accuracy and robustness, maintains high recognition accuracy, and effectively replaces the base model of adversarial training methods.

The application discloses a target detection anti-robustness improvement method based on double-domain feature enhancement, and relates to the technical field of artificial intelligence security. The method comprises the following steps: acquiring an original image and extracting spatial domain features; using a robustness score mask to deconstruct the features into a robust subspace and a non-robust subspace; performing bidirectional calibration on the non-robust features through an interference identification mask to suppress noise and enhance useful information; projecting the features into a frequency domain by using a discrete Fourier transform, identifying and fusing high-frequency useful components to realize functional disentanglement and high-frequency recalibration; and fusing the double-domain features through a cross-attention mechanism, generating more robust representations with stronger discrimination, and outputting detection results. Through spatial-frequency double-domain collaborative optimization, the application solves the problems that a traditional model is difficult to distinguish non-robust features and frequency information utilization is insufficient, maintains high precision of clean samples, and significantly improves the safety and reliability of the model against adversarial attacks.

The invention provides a low-noise metasurface holographic 3D display method based on double spatial constraints, and the method comprises the following four steps: 1, carrying out the layering processing of a 3D object, obtaining a target plane light field, calculating the corresponding complex amplitude distribution based on the angular spectrum diffraction principle, and obtaining a target plane light field; by adding double constraints of a step function and bandwidth limitation to the spatial frequency, precise calculation of holographic plane complex amplitude is realized; 2, performing back propagation on the obtained holographic plane complex amplitude and the randomly generated pure phase hologram to obtain a target complex amplitude and a calculation complex amplitude respectively; step 3, generating a loss function by comparing the difference between the amplitude and the phase in the calculated complex amplitude and the target complex amplitude, and continuously iteratively optimizing the loss function through a stochastic gradient algorithm to enable the calculated complex amplitude to approach the target complex amplitude to obtain an optimized pure phase hologram; and 4, designing a metasurface structure according to the optimized pure phase hologram, and when laser irradiates the metasurface structure, reproducing a low-noise reconstructed image of the 3D object.

This invention discloses a detail-context-aware target detection method based on spatial frequency guidance, aiming to solve the problem of insufficient detection accuracy of small targets in remote sensing images due to low resolution and complex backgrounds. First, a designed spatial frequency-guided dual-dilated convolution module is applied to adaptively select the dilated convolution scale, balancing local detail extraction with global semantic extraction and suppressing background interference. Next, a cross-layer fine-grained fusion module is constructed based on this unit to fuse the shallowest layer features to preserve the details of small targets. Then, a multi-scale enhanced decoupled detection head is used, employing multi-kernel parallel convolution to capture multi-scale target edges, improving localization accuracy. Finally, a Gaussian modeling loss function is used to address the sample distribution imbalance problem, stabilize the regression gradient, and achieve high-precision target detection.