37 results about "Nonlinear interpolation" patented technology

The invention discloses a curve determination method for a centrifugal pump point-by-point integration method, and belongs to the technical field of centrifugal pumps. The curve determination method of the centrifugal pump point-by-point integration method is a parameterized modeling method based on an Euler equation and momentum conservation, the inlet and outlet boundaries of a vur curve are calculated according to design requirements, after a smooth transition curve is constructed in combination with runner geometry and attack angle constraint, the curve is constructed through nonlinear interpolation according to the dynamic correlation streamline arc length, and the curve is determined. The relation between the streamline arc length and the curve is established; according to the method, when the centrifugal pump blade molded line is designed, the curve can be quickly obtained through a parameterization method according to the given design condition and the streamline arc length s.

The invention discloses an industrial multi-source heterogeneous data feature fusion and dynamic modeling method and system, and relates to the technical field of industrial multi-source heterogeneous data processing, and the system comprises a data collection module, an event triggering type local space-time alignment module, a time sequence data set generation module, a feature fusion module and an industrial equipment state model generation module. According to the method, an event-triggered local space-time alignment module is arranged, an event-driven dynamic space-time anchoring mechanism is adopted, and in a preset tolerant time window, an image feature time sequence is constructed through industrial camera superframe sampling to be matched with sensor time sequence data, so that feature dislocation caused by sampling frequency difference is avoided; compared with an existing interpolation method, the non-linear coupling relation between the process parameter data and the sensor time sequence data is obtained, and the problem of feature fusion distortion caused by the time granularity difference is solved by performing non-linear interpolation on the process parameter data through Gaussian process regression and generating a continuous proxy curve synchronous with the sensor time sequence data.

The invention discloses a three-order precision nonlinear interpolation smooth factor design method based on a five-point template, and belongs to the field of computational fluid mechanics, a high-precision format realizes unification of high precision, no oscillation and multi-scale adaptability in complex flow simulation through nonlinear weighting, multi-order combination and smooth factor design, and the method has the advantages of high precision and high precision. The device is an indispensable numerical tool for aerospace. Wherein the smooth factor is a core mechanism for realizing shock wave capture and high resolution in a high-precision format, and weight distribution is dynamically adjusted through smoothness of a quantitative solution, so that balance among shock wave capture, high precision and low dissipation is achieved. Therefore, the invention provides the Z-type third-order precision nonlinear interpolation smooth factor which is based on the five-point template and meets the condition that the optimal precision of the nonlinear weighting framework of the multi-order candidate template is sufficient. A new consistent high-order smooth factor meeting the optimal interpolation precision is designed by using a consistent high-order smooth factor method.

The invention relates to the technical field of vertical shaft guide detection, in particular to an industrial intelligent visual detection and diagnosis system for a vertical shaft guide. The coal dust shadow quantization unit extracts a coal dust region by adopting a double-peak self-adaptive threshold segmentation combined region growing algorithm, and introduces a space weight factor to calculate an average gray value and a coverage area proportion, and the dynamic coupling calibration unit depends on a three-dimensional lookup table calibrated by multiple working conditions and nonlinear interpolation; the optimal deviation compensation coefficient of the jitter and coal dust combination is matched, a dynamic gray segmentation threshold value is generated, and a defect diagnosis decision unit accurately calculates the abrasion depth through partition self-adaptive segmentation, multi-scale morphological filtering and denoising and gradient amplitude and curvature double-constrained sub-pixel edge detection. A hierarchical control instruction is output according to the real-time depth and the historical trend, coupling interference is eliminated, the cage guide abrasion detection accuracy is improved, and a reliable basis is provided for safe operation and maintenance.

The invention discloses a method for determining a parameter lookup table of a neural network activation function, a method and system for calculating the neural network activation function based on the parameter lookup table and a storage medium, and belongs to the technical field of neural network accelerators, and the method comprises the steps: dividing the neural network activation function into a linear region and a nonlinear interpolation region; dividing the nonlinear interpolation region into a plurality of sub interpolation regions; performing linear fitting on the multiple segments of sub-interpolation regions to obtain slopes and intercept of multiple linear functions in one-to-one correspondence with the multiple segments of sub-interpolation regions; and determining a parameter lookup table according to the input variable value range of the neural network activation function corresponding to the multiple segments of sub-interpolation regions and the corresponding relationship between the slopes and intercept of the multiple linear functions, and storing the parameter lookup table in the memory. According to the method, the activation function is divided into the linear region and the nonlinear interpolation region, the activation function of each partition is approximately calculated through the linear function, the parameter lookup table is determined, and the activation function calculation is realized.

The invention discloses a virtual-real interaction method and system for simulated training flight of an unmanned aerial vehicle, and relates to the field of general control or regulation systems.The method comprises the steps that airflow channels are divided in a preset geographic area; performing nonlinear interpolation calculation on the real-time meteorological data according to the position of a preset detection point in each airflow channel to obtain meteorological data of each preset detection point; generating an airflow disturbance characteristic graph; based on the airflow disturbance characteristic pattern, arranging a region conversion buffer zone among the stable airflow region, the weak disturbance region and the strong disturbance region; under the condition that it is determined that the virtual unmanned aerial vehicle enters the area conversion buffer zone according to the real-time position coordinates of the virtual unmanned aerial vehicle in the virtual training environment, the pitch angle, the roll angle and the yaw angle of the virtual unmanned aerial vehicle are adjusted according to a preset time interval and a preset incremental coefficient, and angle variation is obtained; and the angle variation is converted into a force feedback signal, and the force feedback signal is fed back to an operator through a force feedback device. According to the invention, the authenticity and effectiveness of virtual unmanned aerial vehicle training flight are improved.

The invention belongs to the technical field of CT images, and particularly relates to a double-domain CT metal artifact removal method based on deep learning. Comprising the following steps: 1) acquiring CT images of different scanning parts, and processing the CT images to obtain simulated CT images containing metal artifacts; 2) correcting and simulating the CT image containing the metal artifact through a traditional nonlinear interpolation algorithm to obtain an image after nonlinear interpolation correction, and preprocessing the image; 3) constructing a metal artifact removal model based on a deep convolutional neural network, and training the metal artifact removal model; 4) reasoning the preprocessed image by using the trained model to obtain a metal artifact-removed image, and counting the training precision; and 5) carrying out post-processing on the metal artifact-removed image, and carrying out reverse normalization to an original CT magnitude. The method not only performs metal artifact correction in the projection domain, but also performs metal artifact correction in the image domain, so that the metal artifact removal precision is ensured, and the model convergence can be accelerated.

The invention discloses a probability symbol generation method, a probability symbol operation method and equipment, relates to the field of digital signal processing, and solves the problems of high hardware complexity, low nonlinear operation efficiency, insufficient probability calculation precision and large conversion loss of traditional fixed-point calculation. According to the technical scheme, by fusing the advantages of fixed-point and probability calculation, a numerical value is divided into a high-bit fixed-point integer and a low-bit probabilistic part, a multi-bit probability symbol which meets the requirement for expectation unbiasedness and is controllable in variance is constructed, a class fixed-point, class probability and iterative calculation triple system is designed, and the probability of the multi-bit probability symbol is calculated. The multiplexing standard arithmetic unit processes the probability symbol stream to guarantee the linear calculation precision; single-stage multiply-accumulate operation and nonlinear interpolation are realized through a multiplexer, and discrete value selection and weighting are driven by probability; iterative resources are optimized in combination with dynamic filtering conversion. According to the method, a new normal form is provided for a high-energy-efficiency communication chip and a storage and calculation integrated framework, and digital signal processing is promoted to evolve towards the high-precision and low-power-consumption direction.

Embodiments of the present disclosure relate to a nonlinear interpolation unit. The interpolation logic described herein provides a good approximation of bicubic interpolation without performing all of the computations normally required for bicubic interpolation, which is generally smoother than bilinear interpolation. This allows the approximation of the smooth bicubic interpolation to be performed on devices with limited processing resources, such as mobile devices. At each position in a set of predetermined interpolation positions within an array of data points, a set of predetermined weights represents a bicubic interpolation that can be applied to the data points. For a plurality of predetermined interpolation positions surrounding a sample position, the corresponding set of predetermined weights and the data points are used to determine a plurality of surrounding interpolated values that represent the results of performing bicubic interpolation at the surrounding predetermined interpolation positions. Then, a linear interpolation is performed on the surrounding interpolated values to determine an interpolated value at the sample position.

This invention relates to a high-precision digital predistortion phase interpolator with a constant input load, comprising: a multi-phase clock generation circuit, an input buffer circuit, two pre-filter circuits, a phase interpolation core circuit, a CML2CMOS circuit, and a control code decoder circuit. The two pre-filter circuits convert two full-swing square wave signals into two sinusoidal clocks. The phase interpolation core circuit performs nonlinear interpolation on the two sinusoidal clocks with a constant input load based on two weighted control codes, outputting a small-swing signal. This phase interpolator significantly improves interpolation accuracy. Digital predistortion of the interpolation weights can be achieved by adjusting the control code decoder, further enhancing the accuracy of the phase interpolator and enabling high-speed, high-precision output clocks. It is suitable for clock data recovery circuits and phase-locked loops in high-speed SerDes systems.

The application discloses a channel state information compensation method, a communication device and a computer program product. It relates to the technical field of network communication, and the method comprises the following steps: acquiring a target moving speed of relative motion between a first device and a second device, and a plurality of channel estimation results of a channel between the first device and the second device; determining a target nonlinear interpolation coefficient corresponding to the target moving speed based on a corresponding relationship between the moving speed and the nonlinear interpolation coefficient; and compensating channel state information of the channel based on the plurality of channel estimation results and the target nonlinear interpolation coefficient to obtain a channel state information compensation result. The application solves the technical problem that in the related art, the channel considered when performing channel compensation is mostly a channel in a static or slow moving environment, and in the case that the terminal moving speed is relatively high, the channel has strong time variation, the channel reciprocity is poor, and the channel compensation accuracy is low.

The invention discloses a 5G-based underground coal mine unmanned inspection control method, and relates to the technical field of underground coal mine unmanned inspection, and the method comprises the following steps: constructing a multi-dimensional electromagnetic propagation sensing frame on the moving track of an underground coal mine inspection robot, collecting the instantaneous signal intensity, phase deviation and polarization direction, and generating a time sequence electromagnetic feature sequence; a space energy density mapping mechanism is operated based on a time sequence electromagnetic characteristic sequence, a signal distribution state is converted into a space-time continuous energy field, and sequence abnormal points are corrected through nonlinear interpolation to form an energy distribution pattern. According to the method, dynamic description of underground signal propagation and refined display of the risk situation are achieved through multi-dimensional electromagnetic sensing and energy mapping, an inverse phase wave train is generated by introducing a time domain inversion mirror algorithm to actively compensate extremely uneven signals, a propagation channel is uniform and stable, and operation continuity and safety of an inspection robot are ensured.

The invention provides a PCCT device and a control method thereof, which can improve interpolation precision related to substance discrimination and shorten processing time. This PCCT device is provided with: a scanner that rotates, around a subject, an X-ray source that irradiates X-rays to the subject and a photon counting detector that detects X-rays that have passed through the subject for each of a plurality of energy intervals; and an image generation unit that generates a tomographic image using projection data calculated from the output of the photon counting detector, the PCCT device being characterized by being further provided with: a map creation unit that creates a tomographic image using projection data calculated from the output of the photon counting detector; creating a calibration data map on the basis of the number of X-ray photons transmitted through each of a plurality of calibration phantoms having a first base material and a second base material of known material and thickness; a region division unit that divides the calibration data map into a non-linear region and a linear region; and an interpolation unit that performs a non-linear interpolation in the non-linear region and a linear interpolation in the linear region.

The present application relates to the vertical shaft cage detection technical field, specifically, it relates to a kind of industrial intelligent vision detection and diagnosis system for vertical shaft cage, acquisition unit synchronously obtains cage flange surface image and cage car pitch angle, coal dust shadow quantization unit uses bimodal adaptive threshold segmentation combined with region growing algorithm to extract coal dust area, introduce space weight factor to calculate average gray value and coverage area proportion, dynamic coupling calibration unit relies on three-dimensional look-up table and nonlinear interpolation of multi-working condition calibration, match the optimal deviation compensation coefficient of the combination of jitter and coal dust, generate dynamic gray scale segmentation threshold, defect diagnosis decision unit is through zoned adaptive segmentation, multiscale morphological filtering denoising, combined with gradient amplitude and curvature double-constraint subpixel edge detection accurately calculate wear depth, according to real-time depth and historical trend output grading control instruction, eliminate coupling interference, improve cage wear detection accuracy, provide reliable basis for safe operation.

The invention provides a gastrointestinal three-dimensional dynamic imaging analysis method based on conductivity distribution modeling, and the method comprises the steps: applying at least three composite electric signals with different frequencies to a gastrointestinal region, collecting voltage response signals, and constructing a dynamic conductivity parameter matrix based on frequency domain feature separation. And then, spatial discretization processing is carried out by using a nonlinear interpolation algorithm, a three-dimensional conductivity distribution map is generated, dynamic modulation and demodulation processing is carried out on the map, and a frequency band weight coefficient is adjusted. And mapping the modulated data to a dynamic imaging coordinate system through an inverse projection algorithm to generate a three-dimensional dynamic imaging result. According to the space coordinates and the change rate of the conductivity gradient mutation area, the position and the activity state of the gastrointestinal abnormal area are judged, and a self-adaptive feedback mechanism is adopted to update boundary condition parameters of the initial conductivity distribution model. The accuracy and efficiency of gastrointestinal disease diagnosis can be improved.

The present invention discloses a multi-channel magnetic resonance imaging method based on CMA-Net, which includes k-space data preprocessing, GRAPPA reconstruction, construction and training of a CMA-Net network model, data consistency replacement, and image reconstruction. The CMA-Net network model uses a complex multi-scale cascade attention module to improve the global feature extraction capability of the network, embeds a complex attention gate module into the fast connection between the encoding path and the decoding path, and improves the network's feature extraction capability for local region of interest information; utilizes the GRAPPA reconstruction results to construct residual connections, and combines the advantages of linear interpolation and nonlinear interpolation to further improve imaging quality; constructs multi-angle residual connections to effectively alleviate the gradient vanishing problem and improve network performance. Experimental results show that this method can effectively reduce the artifacts of multi-channel undersampled magnetic resonance reconstruction images and reconstruct high-quality magnetic resonance complex images.

The application provides a wide-range high-altitude task environment aircraft high-dimensional temperature prediction method, comprising the following steps: step one: temperature-time first prediction based on regression analysis; step two: temperature-time secondary prediction based on nonlinear fitting and error random representation; step three: temperature-height prediction based on empirical formula and dynamic interpolation; step four: temperature-longitude and latitude prediction based on dynamic interpolation; and step five: high-dimensional temperature fusion prediction. The application comprehensively utilizes mathematical statistics methods such as regression analysis, nonlinear fitting, linear (nonlinear) interpolation and empirical formula to construct a temperature factor prediction model related to four-dimensional variables including time, height, longitude and latitude, so that a reasonable temperature long-term prediction value can be output when any four-dimensional coordinate in the given aircraft task process, thereby providing guidance for the task process.

The present application discloses a method, device, equipment and storage medium for brightness correction of a display panel. The method includes: calculating the corrected brightness value of each grayscale to be corrected within a first grayscale interval according to a nonlinear interpolation algorithm; the corrected brightness value is less than the linear brightness interpolation value obtained according to the binding point grayscale on both sides of the grayscale to be corrected and the linear interpolation algorithm; determining the correction register value corresponding to each grayscale to be corrected according to the first corresponding relationship and the corrected brightness value of each grayscale to be corrected; the first corresponding relationship is the corresponding relationship between the register value and the actual brightness; and burning the correction register value to the storage module of the display panel. According to an embodiment of the present application, the corrected brightness value of each grayscale to be corrected can be recalculated according to the nonlinear interpolation algorithm, and the correction register value corresponding to the corrected brightness value can be burned to the display panel. The display panel displays according to the correction register value, and can reduce the luminous brightness within the first grayscale interval to achieve brightness correction.

The application discloses a curve determination method for point-by-point integration of a centrifugal pump and belongs to the technical field of centrifugal pumps. u The curve determination method for point-by-point integration of the centrifugal pump is a parameterized modeling method based on Euler equation and momentum conservation, the inlet and outlet boundaries of the v r curve are calculated according to design requirements, a smooth transition curve is constructed by combining flow channel geometry and angle of attack constraints, the curve is constructed by dynamic correlation of streamline arc length through nonlinear interpolation, and the relationship between the streamline arc length and the curve is established; the centrifugal pump blade profile can be designed according to the given design conditions and the streamline arc length s through the parameterized method.

The application provides a high-precision hole plate alignment method for a multilayer circuit board, and belongs to the technical field of printed circuit board manufacturing. The method collects measured coordinates of each key layer pad array in the laminated multilayer board through an X-ray perspective system, takes each layer pad array as a distributed deformation sensor, and pairs with the design nominal coordinates to obtain a deviation vector; affine transformation separation and radial basis function nonlinear interpolation are respectively performed on each key layer to reconstruct an anisotropic deformation field covering the whole board area; the range of layers passed through is determined according to the drilling type, and weighted fusion is performed on the deformation field of each layer to calculate the optimal compensation vector of each drilling hole, and a compensated drilling program file is generated. The method upgrades the traditional target alignment and two-step global scaling method to a three-stage compensation process of distributed sampling, nonlinear deformation field reconstruction and hole-by-hole coordinate correction, and controls the hole plate offset to within 25 microns without changing the existing drilling equipment hardware.

The invention relates to a GA-BPNN-based spectral calibration method for a prism dispersion type spectral imager, which is mainly used for solving the problems that in spectral calibration, Gaussian fitting is carried out by using a least square method, so that the requirement on data distribution is high, and a local optimal solution is easily caused; and the technical problems of Runge phenomenon, sensitivity to data noise, local inflexibility, boundary error accumulation and the like when polynomial fitting is used for nonlinear interpolation are solved. According to the GA-BPNN-based spectral calibration method for the prism dispersion type spectral imager, Gaussian fitting is carried out through a genetic algorithm GA to obtain a spectral response function of a selected channel, and nonlinear interpolation is carried out through construction of a BP neural network to determine a spectral response function of a whole channel. The method can adaptively find a global optimal solution of Gaussian fitting and nonlinear interpolation, has strong nonlinear mapping capability and fault-tolerant capability, effectively solves the problem of dispersion nonlinearity of the prism, and improves the quantitative measurement capability of the prism dispersion type spectral imager.

The invention relates to the related technical field of pointing correction, in particular to a method for correcting pointing of a two-dimensional phased array by using a difference-sum ratio. The method comprises the steps that pointing information can be rapidly extracted by adopting an electric scanning mode in combination with a phased array antenna and difference phase information, and pointing precision correction can be rapidly completed by adopting a polynomial interpolation method for accurately extracting the pointing information. The method has the technical effect that the corresponding relation between the beam pointing angle and the beam control angle is accurately positioned by using the difference-sum ratio. The data acquisition efficiency is improved by utilizing an electric scanning mode; and carrying out nonlinear interpolation encryption and correction on the grid by utilizing MATLAB (matrix laboratory).

The application discloses a design method of a three-order precision nonlinear interpolation smoothing factor based on a five-point template, belongs to the field of computational fluid dynamics, and realizes the unification of high precision, non-oscillation and multi-scale adaptability in complex flow simulation through nonlinear weighting, multi-order combination and smoothing factor design, and becomes an indispensable numerical tool for aerospace. The smoothing factor is a core mechanism for realizing shock capturing and high resolution of the high-precision format, and the weight distribution is dynamically adjusted through quantitative solution smoothing, so that a balance is achieved between shock capturing, high precision and low dissipation. Therefore, the application proposes a Z-type three-order precision nonlinear interpolation smoothing factor based on a five-point template, which meets the optimal precision sufficient condition of a multi-order candidate template nonlinear weighting framework. The method uses the method of expanding the template, and uses a consistent high-order smoothing factor method to design a new consistent high-order smoothing factor which meets the optimal interpolation precision.

The invention relates to the technical field of geotechnical mechanics and geophysical exploration, and discloses a three-dimensional electrical resistance tomography method for geotechnical materials and a data acquisition device. The method aims to solve the technical problems that in the prior art, a general linear interpolation method is prone to generating a block mosaic effect and is extremely unfavorable for image analysis, and a nonlinear interpolation method is poor in accuracy, possibly causes calculation difficulty and cannot achieve real-time rapid imaging. The method comprises the following steps: (1) collecting electrode data; (2) screening and summarizing data; (3) carrying out hierarchical plane inversion imaging; and (4) depth three-dimensional interpolation. The spatial electrode arrangement mode is optimized, a numerical inversion imaging system capable of being used for engineering scale measurement is formed, real-time, rapid and accurate monitoring of geotechnical engineering projects is achieved, the detection speed is high, the imaging efficiency is high, the calculated amount is relatively small, the requirement for non-intrusive engineering monitoring is met, and engineering condition damage cannot be caused.

A virtual-reality interaction method and system for simulated training flight of unmanned aerial vehicles (UAVs) relates to the general field of control or regulation systems. In this method, airflow channels are divided in a preset geographical area; real-time meteorological data is nonlinearly interpolated according to the position of preset detection points in each airflow channel to obtain meteorological data for each preset detection point; an airflow disturbance characteristic map is generated; based on the airflow disturbance characteristic map, a regional conversion buffer zone is set between the stable airflow zone, the weak disturbance zone, and the strong disturbance zone; when the virtual UAV is determined to enter the regional conversion buffer zone based on the real-time position coordinates of the virtual UAV in the virtual training environment, the pitch angle, roll angle, and yaw angle of the virtual UAV are adjusted at preset time intervals with a preset incremental coefficient to obtain an angle change; the angle change is converted into a force feedback signal and fed back to the operator through a force feedback device. This application improves the authenticity and effectiveness of virtual UAV training flights.

The present application relates to a high-precision shock wave capture method based on the finite difference method. The method includes: constructing a hyperbolic conservation law equation of the flow field of the aircraft to be simulated, setting a calculation interval; dividing the calculation interval by a semi-uniform grid division method, determining the solution point and the flux point; constructing a nonlinear interpolation format according to the Lagrangian interpolation format, nonlinear weight and linear weight on multiple mutually nested preset sub-templates; obtaining the left upwind value and the right upwind value at the flux point according to the nonlinear interpolation format and the preset sub-template; calculating the numerical flux of the flux point, and using a preset difference format to approximate the derivative of the flux at the solution point; performing numerical simulation by a time-advancing method based on the approximate derivative of the flux, and completing the shock wave capture of the flow field of the aircraft to be simulated. This method can capture shock waves robustly at the boundary of the computational domain, and can also well solve the problem of the numerical solution converging to an erroneous solution or the computational divergence at the interface.

The application discloses a transient power quality disturbance detection method based on digital frequency conversion and nonlinear interpolation, which comprises the following steps: collecting three-phase voltage and three-phase current of a power grid under different disturbances; sequentially performing waveform interception, spectrum segmentation, spectrum shift, sinusoidal interpolation, resampling, empirical mode decomposition and correlation coefficient calculation on single-phase voltage or current; extracting required components and feature values according to correlation coefficients of the components; training an OS-ELM classifier; and detecting transient power quality disturbance and disturbance categories through the OS-ELM classifier.

The application discloses an infrared focal plane array POCC parameter solidification loading method and particularly relates to the field of infrared image sensors; a basic pixel output matrix is collected at a calibration temperature, and initial POCC parameters are calculated; pixel response data is collected at multiple temperature points, and a temperature drift response model is constructed; dynamic prediction parameters are generated based on the initial parameters and the temperature drift model, and a temperature-adaptive POCC solidification matrix Pfused is formed through weighted nonlinear interpolation fusion; automatic loading is performed in the system power-on initialization stage; in the running process, if it is detected that the temperature change exceeds a set threshold, a corresponding subset in the Pfused is called to perform offset correction; the application has the characteristics of temperature drift sensing, autonomous updating and solidification availability, improves the offset correction accuracy and running stability of an infrared imaging system in a complex thermal environment, and is suitable for application scenarios of multiple types of infrared focal plane arrays such as satellite-borne, airborne and vehicle-borne infrared focal plane arrays.

The invention provides a frequency response measurement method and system based on amplitude adaptive adjustment and nonlinear interpolation, and the method comprises the steps: determining the disturbance amplitude of a test frequency point based on each test frequency point of a tested object according to the pre-calculated safety operation threshold information of the test frequency point; performing frequency disturbance on the tested object according to the disturbance amplitude of the test frequency point to obtain the frequency response of the tested object at the test frequency point; based on the frequency response of the tested object at the test frequency point and the safe operation threshold information, correcting the disturbance amplitude of the test frequency point to obtain frequency characteristic data of the tested object at the test frequency point; according to the frequency characteristic data of the tested object at each test frequency point, utilizing a nonlinear interpolation algorithm to output a frequency response characteristic curve of the tested object; by correcting the disturbance amplitude, the modulation error of the disturbance amplitude can be reduced, and by combining the frequency response characteristic curve output by the nonlinear interpolation algorithm, the precision and reliability of frequency response measurement can be improved.

The embodiment of the invention provides an image reconstruction model training method and device, an image reconstruction method and device and electronic equipment, and relates to the technical field of image processing. The image reconstruction model training method comprises the following steps: acquiring a first sample RAW image, a second sample RAW image and a real HDR image acquired from the same scene at the same visual angle; performing image splicing on the first sample RAW image and the second sample RAW image to obtain a predicted spliced image; using an encoder to extract prediction down-sampling features of the prediction spliced image; performing up-sampling processing on the predicted down-sampling feature by using a decoder to obtain a predicted RAW image; performing nonlinear interpolation on the predicted RAW image to obtain a predicted RGB image; and based on the difference between the real HDR image and the predicted RGB image, adjusting model parameters of the image reconstruction model until the image reconstruction model converges, and obtaining a trained image reconstruction model. Based on the RAW image with more complete and more accurate image information, the HDR image with higher quality can be obtained.