79results about How to "Efficient estimation" patented technology

The invention discloses an OFDM (Orthogonal Frequency Division Multiplexing) pilot scheme design and channel estimation method. The method comprises the following steps: pilot symbols are uniformly inserted into data frames of an OFDM system to obtain a pilot pattern, the frequency response of the channel at the pilot position is calculated according to the pilot pattern to obtain the sampling sequence of the channel frequency response information, and modified orthogonal matching tracking channel estimation is carried out according to the sampling sequence of the channel frequency response information and part of discrete Fourier matrices. By adopting the method provided by the invention, the channel estimation algorithm complexity is greatly reduced while the pilot density is greatly lowered, the channel estimation accuracy is improved, and accordingly, the efficient, low-complexity and accurate channel estimation is realized.

The invention provides an ISAR (inverse synthetic aperture radar) imaging orientation calibration method, which comprises the following steps that a radar records echo to obtain echo data using the distance as the line vector and the orientation as the row vector; distance units are selected for forming a new echo matrix; a frequency domain windowing technology is adopted for extracting the scattering point echo; the phase quadratic term factor of the distance unit echo is estimated; the orientation resolution ratio of the ISAR image is calculated on the basis of a least squares estimation method; the orientation calibration on the ISAR images is completed according to the orientation resolution ratio of the ISAR image. The method provided by the invention has the advantages that 2-order DPT conversion is adopted for estimating the phase quadratic term factor of the distance-dimension echo, the defects of real-time application inconvenience and great calculation quantity during the phase quadratic term estimation of the traditional method based on LPFT and image contrast ratio are overcome, and in addition, the rotating angular speed of a target is obtained through the least squares estimation method, so the orientation calibration of the ISAR images is realized.

The invention provides a PSF estimation method based on a hybrid Gaussian model and sparse constraints and relates to the optical image restoration technology field. The method comprises steps that through extracting a surface-like region with relatively good imaging quality from an image, a PSF fitting function is established through utilizing the hybrid Gaussian model, an initial PSF template is acquired through model parameter solution, the template is taken as an initial iteration PSF template of a sparse constraint restoration model, and a final PSF template of an integral image is acquired through iteration. The method is advantaged in that no approximation processing is carried out in a processing process, main operation is carried out in an image space domain, repeated iterative computation is not required, and reliable and high efficiency PSF estimation can be realized.

The invention discloses a method for deblurring an iris image, which comprises the following steps of: a step S1, analyzing the input iris image, judging blur of the iris image belongs to defocus blur or motion blur and initializing according to a judging result to obtain a point spread function; a step S2, selecting the gradient of the input iris image so that the point spread function can be more effectively estimated on a selected gradient map, and further modifying the initialized point spread function on the basis of selected gradient information; and a step S3, on the basis of the modified point spread function, repairing the blurring iris image. According to the method, the adverse effect of the blurring iris image on the identification performance of a system can be effectively reduced, and the method can be widely applied to an iris identification system so as to improve the robustness and the reliability of the system and relax the requirement for a user.

The embodiment of the invention discloses a clustering model training method and device, electronic equipment and a computer storage medium. The method comprises the steps that a global clustering model is acquired based on master nodes; according to any one of at least one slave node in a distributed system, the global clustering model is acquired from the master nodes in the distributed system,clustering estimation is performed based on the global clustering model and calculation data allocated to any corresponding slave node, and a local clustering model corresponding to any slave node isobtained; the distributed system comprises the master nodes and at least one slave node in communicating connection with the master nodes, wherein the master nodes are in communicating connection withall the slave nodes; and the global clustering model is trained based on the obtained local clustering models corresponding to all the slave nodes. Through the clustering model training method in theembodiment, the synchronization rate among the calculation nodes is lowered, and clustering efficiency is improved.

The invention discloses an age estimation method based on deep learning. The method comprises the following steps: firstly, acquiring brain nuclear magnetic resonance images of a plurality of persons,taking the brain nuclear magnetic resonance images as training samples, segmenting the brain nuclear magnetic resonance images as the training samples, extracting and digitalizing the segmented imageslices of each block to obtain image data of the training samples, and preprocessing the image data to obtain a mean value and a variance; establishing a convolutional neural network, obtaining an input value of the convolutional neural network, carrying out training classification on the convolutional neural network to obtain a trained model, and then obtaining a relationship between proportionsof all block areas of the human brain nuclear magnetic resonance image; and finally, taking the brain nuclear magnetic resonance image of the subject as a test sample, substituting the test sample into the model to obtain a feature vector, and sending the feature vector into a support vector machine for training classification to obtain the predicted age of the test sample. According to the method, the age of the subject can be estimated more quickly, efficiently and accurately.

The invention discloses an ultralow-dispersion switching kalman filter (UD-SKF) method for inert measurement unit (IMU) and very-high-frequency radio combined navigation for a mars entry section. The method comprises the following steps of: 1, constructing a dynamic equation of the mars entry section, 2, constructing a measurement equation of the mars entry section, 3, discretizing the dynamic equation (5) and the measurement equation (6), and 4, outputting navigation information by a UD-SKF algorithm. Compared with an existing extended kalman filter (EKF) algorithm, the UD-SKF algorithm adopted by the invention has the advantages that deviation information of the measurement equation is fused into a filter process, so that a filter effect is greatly improved, and the navigation precision is improved; furthermore, by UD, phenomena of larger errors and even dispersion in the filter process are reduced; the stability of the filter process is enhanced; a navigation state of a detector can be efficiently estimated in real time.

The invention discloses a remote sensing image super-resolution reconstruction method based on fuzzy kernel classification and an attention mechanism. Firstly, high-resolution and low-resolution optical remote sensing images corresponding to a certain region are given, and a test sample and a training sample are divided; secondly, performing fuzzy kernel estimation on all low-resolution images inthe data; then, fuzzy kernels of all samples in the training set are used for K-means clustering; classifying the high-resolution image pair and the low-resolution image pair of the test set by usinga clustering model; and then constructing a neural network model based on an attention mechanism, setting absolute value errors of the high-resolution image and the low-resolution image as a loss function, obtaining an optimal model according to a test set reconstruction result, finally reconstructing an input image according to the model, and outputting a final result graph. According to the method, the peak signal-to-noise ratio of the reconstructed image can be improved, the robustness is high, and the definition of image edge details is improved.

The invention discloses a workpiece pose estimation method based on component model expression, and the method comprises the following steps: carrying out the data enhancement of to-be-detected data,and obtaining a processed image; carrying out feature extraction on the image through a convolutional neural network to obtain feature maps of three scales; expressing a network structure of each feature map through a component model to obtain corresponding scores and model response maps of the whole and the components; performing parameter optimization on all the model response graphs to obtain acomprehensive response graph and key points; calculating n 3D points in the space through an EPnP algorithm to be matched with 2D points in the image, and then obtaining the corresponding pose of thecamera. According to the method, information marking only needs to be carried out on the whole workpiece, component information of the workpiece serves as recessive features, effective components areautomatically found through the neural network to carry out component marking, and the method has the advantages of being rapid and efficient in performance and capable of accurately calculating the6D pose of weak-texture objects such as the workpiece in real time.

The invention discloses an intelligent detection method for typical organic pollutants in an urban river water body. The method comprises the following steps of obtaining input independent variables and output variables for soft measurement modeling of the typical organic pollutants in the water body; obtaining the number of optimal convolution kernels for each convolutional layer of a convolutional neural network (CNN) by using an iterative optimization method; extracting main components of training sample data through a PCA (Principal Component Analysis) algorithm, and performing dimensionality reduction optimization on the CNN input data; improving the CNN independent variable input method, and transforming an image classification model into a regression prediction model; constructing aPCA-CNN model by using the principal component-convolutional neural network, and training the PCA-CNN model; and performing soft measurement on the concentrations of the typical organic pollutants inthe water body through the trained PCA-CNN model to obtain an analysis result. According to the intelligent detection method for the typical organic pollutants in the urban river water body disclosedby the invention, the PCA-CNN model is constructed, the concentrations of the typical organic pollutants in the river water body can be effectively predicted; the accuracy degree is high; the operation is simple; and an efficient method is provided for rapid monitoring and water quality assessment of the urban river water body pollution.

The invention discloses an iteration SKF method of multi-source information integrated navigation of the Mars power descent stage, which comprises the following steps: 1, utilizing the kinetics equation of the Mars power descent stage; 2, building the measurement equation of the Mars power descent stage; 3, discretizing the kinetics equation and the measurement equation, and then linearizing the kinetics equation and the measurement equation so as to obtain the novel kinetics equation and measurement equation; 4, outputting navigation information through utilizing the SKF filtering algorithm. Through the four steps, the kinetics equation and the measurement equation are built, and then the influence of errors in the measurement information is eliminated through utilizing the iteration SKF filtering algorithm to ensure the stability of the filtering algorithm, so that the purpose that navigation state of a detector is estimated efficiently and in real time. The method efficiently corrects the influence on the filtering caused by errors in the measurement equation, corrects the filtering errors caused by the truncation error caused by the Taylor series by utilizing the iteration method, improves the navigation precision, and enhances the stability of the filtering process, so that the navigation state of the detector can be estimated efficiently and in real time.

The invention discloses a filtering-based intelligent fault detection method for motors in the industrial field and belongs to the technical field of fault detection. According to the method, a set membership estimation method is utilized to represent a state feasible set by using vectors, so that prior knowledge of model disturbance and noise does not need to be known in advance, and the practicability and accuracy of the fault detection method are improved; in the solving process of an inversion filtering problem, vectors are used for representing interval boxes; the interval boxes belongingto the feasible set are searched for through the Boolean operation of vector sets, so that the problems of large calculated amount, and the exponential-level increase of calculation time along with increase of interval dimensions of a traditional interval filtering algorithm are solved, and a state interval is estimated more efficiently and accurately. Different from a traditional method for realizing fault detection by utilizing the upper and lower boundaries of estimation residual error, the method provided by the invention obtains the estimated interval of a fault, thereby providing guarantee for subsequent fault diagnosis of the motor by estimating the fault range.

The invention provides a lithium battery parameter identification and SOC estimation method based on a suburb wolf optimization algorithm. The lithium battery parameter identification and SOC estimation method comprises the following steps: 1, measuring the current and voltage of a lithium battery through intermittent constant-current discharge; 2, establishing a second-order RC equivalent circuit model of the lithium battery; 3, constructing a coyote optimization algorithm; 4, constructing an extended Kalman filtering algorithm; and 5, determining each parameter in the lithium battery model by using a coyote optimization algorithm, and estimating the SOC of the battery. The method has the beneficial effects that the lithium battery second-order RC model is established, the discrete state space expression of the lithium battery second-order RC model is deduced, and the model parameter identification is carried out by using the coyote optimization algorithm, so that compared with a traditional heuristic algorithm, the method has the advantages of high identification precision, high convergence speed, and small estimation error by using an identification result to carry out SOC estimation, and the accuracy of the coyote optimization algorithm in the aspect of parameter identification is verified.

The invention relates to a method for fast estimating airborne radar clutter space-time spectrum based on improved OMP. For a sparseness unknown condition of a signal to be reconstructed, a designed improved OMP algorithm is used to iteratively estimate linear approximation signal sparseness so as to quickly solve the support set of the sparse signal and its the coefficient of sparse representation so as to achieve high-efficiency estimation of the airborne radar clutter space-time spectrum. The entire design method is small in amount of calculation and easy to implement.

The invention belongs to the technical field of mode recognition and digital image processing, and discloses a face age estimation method and system, electronic equipment and a storage medium, and the method comprises the steps: obtaining a face age image set, and carrying out the preprocessing of the face age image set, and obtaining a preprocessed face age image set; constructing a face age estimation model; constructing a composite loss function containing error compression sorting loss and attribute guidance classification loss based on sorting labels; training a face age estimation model according to the face preprocessing image set to obtain a trained face age estimation model; and testing the trained face age estimation model according to a test data set to obtain a face age estimation result so as to realize face age estimation. According to the method, a high-performance multi-scale attention mechanism residual convolution unit, an attribute guidance module and a composite loss function containing error compression sorting loss are introduced, so that a face age estimation target which is robust, efficient and high in estimation performance index is achieved.

The invention belongs to the technical field of the data de-noising, and discloses a curvelet domain statistics self-adaptive threshold ground penetrating radar data de-noising method and system. Themethod comprises the following steps: importing a massive complex domain threshold function algorithm, analyzing a change rule of the traditional threshold function curvelet conversion de-noising effect along a threshold function control coefficient so as to be used for the subsequent curvelet domain statistics self-adaptive threshold contrast; performing correlation superposition on the curveletconversion coefficient on the scale and direction by utilizing high-order statistics theory, and statistically and self-adaptively determining the distribution scale and rotational direction of an effective signal on the curvelet conversion coefficient through the correlation; determining a noise removing component threshold range, constructing a statistics self-adaptive threshold function curvelet conversion de-noising algorithm. Compared with the prior art, the processing result on the synthetic ground penetrating radar data containing random noise and related noise and the actually-measuredground penetrating radar data has guidance significance on the precise inference interpretation on the complex ground penetrating radar data.

The invention discloses a human body posture estimation method, device and equipment and a storage medium, and relates to the technical field of computers, in particular to the technical field of image processing. According to the specific implementation scheme, a to-be-estimated picture is input into a pre-trained deep learning network, a depth map corresponding to the to-be-estimated picture andkey point thermodynamic diagrams corresponding to all limb parts, which are output by the deep learning network, are obtained, two-dimensional information of the key points of the limb parts is determined according to the key point thermodynamic diagrams, depth information of the key points is determined according to the depth map, and the two-dimensional information and the depth information ofthe key points are output as a human body posture estimation result. The depth map and the key point thermodynamic diagram are output through the deep learning network, the three-dimensional information including the two-dimensional information and the depth information is obtained according to the depth map and the key point thermodynamic diagram, comprehensive understanding of the human body posture is achieved, and the accuracy of human body posture estimation is improved.

The invention provides a method for constructing a training sequence with zero interference window characteristics, which comprises the following steps: firstly, constructing gray sequence pairs (X(m),Y(m)) with length being Lm equal to 2<m>, wherein m is a positive integer more than 1; secondly, constructing a zero-order mother matrix F<(0)> by using the gray sequence pairs, (formula is shown as the lower right picture), wherein *m is a bit reversal sequence of the Y(m), -Xm is an antitone sequence of the X(m); thirdly, obtaining an n-order mother matrix F<(n)> by using a recursion formula: (formula is shown as the lower right picture), wherein n is positive integer, * expresses matrix intersection operation using a line as the unit, the intersection width is 2<m+n-1>; wherein the n-order mother matrix F<(n)> forms a group of spread spectrum sequences with zero-interference window characteristics, each line in the matrix is used as a spread frequency sequence; each spread frequency sequence forms each training sequence by adding a prefix with the length being Z, and the prefix consists of last Z elements of each spread frequency sequence (formula is shown as the lower picture).

The invention discloses a regional vehicle density estimation method based on a dynamic sampling mechanism and an RBF neural network, and the method comprises the steps of preliminarily constructing avehicle density database according to the sampling information of the vehicle density in a target region; comparing the estimation values of the vehicle density for many times, taking the stored sampling data as an input variable of the RBF neural network, defining a group of activation functions, and establishing an estimation model based on the RBF neural network; applying the Kalman filteringalgorithm to an RBF neural network estimation algorithm; estimating the vehicle density of any point in the target area according to the weight coefficient of each neural network and the correlation function of the vehicle density in the target area; and finally, dynamically estimating the vehicle density in the target area by judging whether the estimation result meets the task requirement or not. The method has the advantages of higher estimation efficiency, lower operation load and higher estimation precision, can effectively estimate the time-varying vehicle density in the target area in real time, and has wide application space and practical range.

The invention relates to a method for building a path loss model in a short-distance internet things environment, comprising the following steps of: building an original test data optimizing and acquiring scheme of a short-distance internet things environment path loss model; building the path loss model by a novel path loss model building scheme for evaluating short-distance internet things environment path loss characteristic; and evaluating a path loss model parameter by a novel path loss model parameter evaluating scheme for evaluating the short-distance internet things environment path loss characteristic. According to the method, the short-distance internet things environment path loss characteristic can be effectively evaluated, and the path loss evaluating deviation can be obviously reduced.

The invention discloses a three-dimensional line-of-sight estimation method and device oriented to a resource-constrained scene, and the method comprises the steps: constructing an end-to-end line-of-sight estimation network, carrying out the face detection and line-of-sight estimation at the same time, carrying out the sampling of two data sets at the same time through multi-task learning, and training different branches through different data; performing fusion training on the collected face detection data set and the line-of-sight estimation data set to enable an end-to-end line-of-sight estimation network to adapt to the two different data fields at the same time, and training the network by adopting a multi-task learning mode to obtain a trained model; and compressing and quantifying the trained model, so that the trained model can be deployed on edge equipment, and real-time estimation of three-dimensional implementation is realized. According to the method, an end-to-end method is used, multiple times of feature extraction on the image is avoided, the running speed is increased, and real-time sight line estimation is supported; according to the method, the lightweight model is adopted and is compressed, so that the model can be operated in a resource-limited scene.

The invention discloses a capacity optimization method, a communication method and system of a mobile wireless optical communication system. The capacity optimization method comprises the following steps: establishing a mobile channel impulse response model; calculating a telecommunication noise ratio output by the receiver; calculating bit error rate values of the wireless optical communication system in different candidate modulation formats according to the telecommunication noise ratio output by the receiver; selecting a first modulation format and a second modulation format from differentcandidate modulation formats; determining the number of code elements of a first modulation format and a second modulation format in each data frame; and constructing a time domain hybrid modulationframe according to the number of the code elements in the first modulation format and the second modulation format, modulating data through the time domain hybrid modulation frame, and performing datatransmission. According to the method, the actual state of the mobile wireless optical channel can be effectively adapted, and the signal transmission scheme is adaptively adjusted and optimized according to the mobile state information of the terminal, so that the spectral efficiency and the mobile transmission capacity of the system are dynamically optimal, and the reliable transmission qualityof the link is ensured.

The invention disclsoes an EPR cable service life estimation method based on multiscale space spectrogram information. The method comprises the following steps: acquiring local discharge spectrogram information of the actually running cable and a to-be-detected cable, extracting an aging characteristic factor and estimating service life of the to-be-detected cable according to the aging characteristic factor. The method disclosed by the invention can accurately and efficiently estimate service life of an EPR cable and greatly reduce maintenance workload.