61 results about "Feature estimation" patented technology

A detection and classification method of a shape in a medical image is provided. It is based on generating a plurality of 2-D sections through a 3-D volume in the medical image. In general, there are two steps. The first step is feature estimation to generate shape signatures for candidate volumes containing candidate shapes. The feature estimation method computes descriptors of objects or of their images. The second general step involves classification of these shape signatures for diagnosis. A classifier contains, builds and / or trains a database of descriptors for previously seen shapes, and then maps descriptors of novel images to categories corresponding to previously seen shapes or classes of shapes.

The invention discloses an integrated automatic focusing camera device and a definition evaluation method. The reliable evaluation of the focusing definition is guaranteed by taking the high-frequency component of an image signal as a feature estimation value of the focusing definition and by adopting a spatial domain and frequency domain combined algorithm method. The automatic focusing device realizes the calculation of the definition evaluation value and the control over the back and forth movement of an optical focusing lens by controlling a stepper motor, makes an image formed at a most definite focusing position by adopting a spatial domain and frequency domain multi-feature climbing search strategy and ensures quick focusing by adopting a digital signal processor for acceleration. By combining a communication interface and a control protocol, an integrated camera device is realized.

A diverse wireless channel analysis platform based on antenna radiation directional diagram and a method thereof are disclosed; a transmitting antenna and a receiving antenna are respectively fixed at certain positions in the space and transmits and receives data through different radiation directional diagram of the antenna, the diversity of the antenna radiation directional diagram is used as a sampling space of the wireless transmission channel response on the space domain, and simultaneously high precision parameter estimation algorithm is used for obtaining detailed features of the wireless transmission channel in the directions of incident and exit through a manner of array signal processing, and conclusion and model construction are carried out on the feature of the wireless transmission channel by combining the feature estimation results of the channel on the time domain and time delay domain. The invention eliminates the fuzzy judgment problem of time and space of the traditional channel analysis method and improves the preciseness of channel analysis.

The invention provides a method and a device for identifying a scene integrated by a multi-feature vision codebook, which belong to the technologies of image processing and mode identifying. The method comprises the following steps of carrying out multi-feature integration on a local area of a scene image through a local classifier, and obtaining the expression of the multi-feature vision codebook of the local area of the scene image; and carrying out overall integration and classification on the expression of the multi-feature vision codebook according to overall integration parameters and classification parameters which are obtained through pre-training. Compared with a mode of generating the expression of a single-feature vision codebook or the expression of a multi-feature signal vision codebook by using a single-feature estimation probability and then carrying out overall feature integration, the method disclosed by the invention has the advantage that a more accurate automatic scene classification result can be obtained.

An image alignment method includes computationally efficient methods of achieving high-accuracy local motion estimates by using phase correlation. The method also estimates motion reliability that allows a generic robust model fitting algorithm to produce more accurate results while operating much more efficiently. One of three methods are used to determine sub-pel motion estimation with improved accuracy. Each of the sub-pel motion estimation methods uses phase correlation, and are based on fitting computationally efficient 2-D quadratic surfaces to a phase correlation surface. A pre-filter is applied which shapes the phase correlation surface to enable appropriate fitting to the quadratic surface. Bias is also compensated for prior to applying a sub-pel motion estimation method. The method also estimates the reliability of the sub-pel motion estimates determined using phase correlation.

The present invention relates to the technical field of probability statistical analysis and prediction of concrete temperature fields, and discloses a bridge-ballastless track structure extreme temperature prediction method and system, so as to avoid subjectivity of the conventional experience distribution method and the cumbersome derivation process of the first-order second-moment method and to obtain a relatively preciser temperature extremum in the case of a certain amount of data. The method disclosed by the present invention comprises: forming a statistical sample by means of measured bridge-ballastless track structure temperature data; calculating firt to fourth moments of the stastical sample to obtain a statistical feature of the sample; constructing virtual distribution by means of standard regular distribution to indirectly describe a distribution condition of the sample; solving a key parameter and feature estimation of the virtual distribution; constructing an extreme state function considering an extreme temperature, and giving an exceedance probability; and calculating first to fourth moments of the extreme state function by means of a reliability indicator formula, and reversing an extreme temperature value corresponding to the exceedance probability and giving a corresponding reoccurrence period.

The invention relates to an arbitrary linear constraint tracking method for simultaneously estimating a target state and a trajectory parameter. The method comprises: acquiring target position measurement information from an observation radar; carrying out augmentation on a state vector by using a slope and a y-axis intercept parameter component to obtain an augmented state vector and augmented state equation, and constructing a pseudo-measurement description constraint relationship; processing measurement information, carrying out nonlinear filtering by using the processed measurement, the augmented state equation and pseudo-measurement, and updating the state estimation and state estimation covariance to realize target tracking. According to the provided method, the trajectory parametersare augmented to the state vector and the shape feature information in the target motion trajectory is utilized reasonably, so that the information wasting is avoided and the estimation precision isimproved; because of sequential processing of radar measurement and pseudo-measurement, the computational burden is reduced; and the filtering result contains the trajectory parameter estimation result, so that the subsequent information processing is performed smoothly.

The invention discloses a wireless channel feature estimation system, and aims to provide a channel estimation method characterized by low implementation complexity, low resource consumption, high estimation accuracy and low processing delay. The wireless channel feature estimation system is implemented by the technical scheme as follows: a signal sender sends a set of known pseudo-random PN sequences as training sequences; a signal receiver converts the serial data of the pseudo-random sequences into parallel data through a serial-parallel conversion module, and sends the parallel pseudo-random sequences to a correlation operation module; the correlation operation module carries out sliding correlation operation on the received pseudo-random sequences and local pseudo-random sequences toobtain the initial estimated value of channel feature information, and sends the initial value of channel estimation to a de-noising module; and the de-noising module calculates the noise intensity byusing the initial value of channel estimation, sets a reasonable noise threshold according to the noise intensity, and de-noises the initial value of channel estimation by using the noise threshold to obtain a final channel estimation value.

The present invention provides a multi-scale feature estimation and high-order BING feature-based target detection method. The method includes the following steps that: an image feature pyramid is constructed for an original image by using an existing multi-scale histogram of oriented gradient approximation algorithm; high-order BING features are extracted in each histogram of oriented gradient; each window in sliding windows is judged by using a two-stage linear SVM; and whether the windows contain a target is determined. According to the method of the invention, the BING features are expanded to the histograms of oriented gradient from simple gradient features, wherein the histograms of oriented gradient can show more details of the features, high-order statistical information is extracted, and therefore, feature expressions can be richer; and when the feature pyramid is calculated, the multi-scale histogram of oriented gradient approximation algorithm is adopted, and therefore, a more detailed feature pyramid can be obtained under a condition that calculation quantity is not increased, and the position of the detected target is more accurate.

A method for estimating feature values from a digitized image includes calculating a gradient of an image, defining a filter weighting function, convolving the gradient with the filter weighting function, integrating over a subdomain of the convolved gradient to calculate components of a cost matrix Φ, wherein the cost matrix minimizes a distance metric in a frequency domain of the image, solving for the eigenvalues of the cost matrix, and calculating one or more features from the cost matrix eigenvalues. One or more predesigned bandpass filters can be applied to the image, and cost matrix eigenvalues are analyzed to select one or more appropriate bandpass filters to be applied to the image.

The invention discloses a method for predicting the atmospheric pollution condition based on an integrated gate recursion unit neural network GRU, and the method comprises the steps: 1), carrying outthe multi-modal feature extraction of an atmospheric pollutant data set S11 (t) through a local mean decomposition function LMD, and obtaining an atmospheric pollutant feature data set; 2) establishing a gate recursion unit neural network GRU by using the training data set, and training the gate recursion unit neural network GRU by using the training data set; 3) inputting the normalized differenttypes of feature data sets into a gate recursion unit neural network GRU, and outputting a normalized sub-mode prediction value; and 4) performing multi-modal feature estimation value integration onthe normalized sub-mode prediction value by adopting inverse LMD operation to obtain a trained LMD-GRU neural network model. According to the method, the problems that the performance of the model islower than that of a multi-mode feature learning model, the precision is low and the actual prediction effect is not ideal due to the fact that feature learning is not obvious when the LSTM model performs regression prediction on haze are solved.

The invention discloses a real-time adaptive convergence estimation system for time synchronization. The real-time adaptive convergence estimation system comprises a synchronization error estimation unit, a synchronization error feature estimation model and a real-time convergence detection model. The method can be applied to a distributed system or a wireless network with a time synchronization requirement, the method is integrated into a time synchronization algorithm used by an application object, and the synchronization error convergence probability is further calculated by utilizing timemigration estimation obtained by the time synchronization algorithm. The synchronization error convergence probability obtained by the method can be used as a basis for determining the system time synchronization precision and convergence state by other applications or a time synchronization algorithm.

A brain network activity estimation system includes means for obtaining brain wave measurement data and functional magnetic resonance imaging measurement data simultaneously measured from a subject, means for constructing a feature estimation model which receives the brain wave measurement data as input data and determining a parameter which defines the feature estimation model, means for calculating a feature value for each module based on an output value from each module that is calculated when the brain wave measurement data is provided as the input data, means for calculating an image feature value for each brain network based on the functional magnetic resonance imaging measurement data, and means for determining one or more modules which express activities of a specific brain network among a plurality of modules by evaluating correlation between the feature value for each module and the image feature value for each brain network.

Automated nuclei area / number estimation utilizes a two-stage estimation framework-area estimation first followed by number estimation. After determining area information, each local patch's shape features are able to be extracted to define a local voting rule. The resulting voting score determines the strength of each local voting peak. The number of voting peaks is exactly the number of nuclei.

The invention discloses a method for autonomous navigation and positioning underwater and belongs to the field of autonomous navigation of an underwater robot. The method uses a random finite set method to realize modeling of a SLAM problem so that the map feature information, map feature observation Information and clutter can be accurately described. In the aspect of map feature estimation, a new mapping information mixing method improves the shortcomings of the prior information of a PHD filter in the prediction phase and improves map feature number and map feature position estimation precision of the traditional probability hypothesis density-synchronous localization and mapping (PHD SLAM) method. In the aspect of robot pose estimation, an empty map feature strategy is used to improve a computing speed of robot pose estimation.

The invention relates to an image classification and recognition method for depth feature estimation based on subspace decomposition. The method comprises the steps of S1, obtaining an unshielded image and an additional image pair of a training set; S2, extracting a depth feature vector from each input image; S3, calculating the difference between the depth feature vectors of the occluded image and the depth feature vectors of the unoccluded image in the additional image pair to obtain an error vector; S4, forming a first subspace by using the calculated depth feature vectors of all the unshielded images, forming a second subspace by using the error vectors, and forming a cascade dictionary based on the first subspace and the second subspace; S5, calculating a coefficient matrix based on the cascade dictionary; S6, training a classifier based on all column vectors of the first subspace; And S7, performing classification and identification on the to-be-identified image based on the cascade dictionary, the coefficient matrix and the classifier. Compared with the prior art, the method has the advantages of wide application range and the like.

The invention discloses a structured scene vision SLAM (Simultaneous Localization and Mapping) method based on point-line-surface features, which comprises the following steps of: firstly, inputting a color image and a corresponding depth image, extracting the point-line-surface features in the image and carrying out feature matching; then detecting a Manhattan world coordinate system according to a plane normal vector, if the Manhattan world coordinate system exists and appears in a Manhattan world map, calculating a camera attitude and tracking point-line-plane feature estimation displacement, otherwise, tracking the point-line-plane feature estimation pose; performing key frame judgment on the current frame, and if the current frame is a key frame, inserting the key frame into a local map; maintaining map information and performing joint optimization on the current key frame, the adjacent key frames and the three-dimensional features; and finally, loopback detection is carried out, and if a closed-loop frame is detected, loopback is closed and global optimization is carried out. The method is a visual SLAM method with high precision and strong robustness, and solves the problem that the visual SLAM precision is reduced and even the system is invalid only based on point features in a low-texture structured scene.

The invention provides a long frequency band radar target jogging feature estimation method. The technical scheme is that a target narrow band echo obtained by a low frequency band radar is subjected to N point discrete sampling to obtain discrete signals, then complex logarithm operation is performed on the discrete signals, phase modulation signals are obtained by phase ambiguity resolving and are subjected to frequency spectrum estimation, and a target jogging spectrum is obtained by calculating, so the jogging feature of the target is obtained. The long frequency band radar target jogging feature estimation method is suitable for a low frequency radar small-amplitude jogging target, the estimation precision of jogging parameters is high, applicable systems have wide frequency bands and the signal to noise ratio demand is low.

The invention relates to a deep clustering voice separation method based on improvement, wherein the method comprises the steps: step 1, mixing experiment data, and extracting logarithm power spectrumfeatures; step 2, building an improved deep clustering voice separation model, and carrying out model training by using a training set; step 3, obtaining an embedded space vector vi from the mixed voice of the test set through the deep clustering speech separation model trained in the step 2, clustering the embedded space vector vi in an embedded subspace through a mean shift clustering method, taking an obtained result as an ideal binary masking value of a training target, and calculating to obtain feature estimation of the two separated voice signals by utilizing the ideal binary masking value and input speech signal features; and step 4, carrying out waveform reconstruction and recovering a voice signal. According to the method, the current voice separation method based on deep clustering is improved, so that the effect of the method is improved under the condition of low signal-to-noise ratio mixed voice input.

The invention discloses a color constancy method based on cascade fusion feature confidence weighting, which provides stable color features for unmanned driving, underwater object recognition, three-dimensional object reconstruction and other computer vision tasks, and comprises the following steps: (1) shooting images and videos under a natural scene light source, and making a data set applied to the color constancy method; (2) according to the particularity of light source colors, weighting a network structure based on cascade fusion feature confidence; (3) performing two-stage training on the network structure by using the data set; and (4) removing the estimated scene light source from the image or the video to realize color constancy of the image and the video. According to the method, the shallow edge texture features and the deep fine-grained deep features in the image are fused in a cascading mode, the feature estimation light source capable of providing more information for light source estimation in the image is fully utilized, and the problem that the light source estimation accuracy is low when a current color constancy method faces a complex environment is solved; and the accuracy of the color constancy method and the robustness of the method in a complex environment are improved.

The invention discloses a multi-target video tracking system based on fractal feature estimation. The multi-target video tracking system comprises an image local fractal feature estimation module, anobservation model establishment and likelihood function calculation module and a multi-Bernoulli filter tracking module. The local fractal feature estimation module calculates local fractal features of each frame of image and forms a new Hurst index image. The observation model establishing and likelihood function calculating module is used for establishing an observation model and calculating a likelihood function by utilizing local fractal characteristics and a Hurst index. The multi-Bernoulli filter tracking module is used for tracking a plurality of targets in the Hurst index image. According to the method, under the condition that the sensor is static or moves, a plurality of targets can be correctly tracked, and good robustness is achieved for complex tracking such as target enteringand leaving, target part shielding and target number changing.

The invention relates to a satellite signal subtle feature estimation method based on precision progressive CZT. The method comprises the steps of collecting to-be-estimated signal data of certain length; transforming the data into a form applicable to a signal rate and carrier frequency subtle feature estimation, and carrying out DFT coarse estimation; and selecting a T refinement interval, calculating the refinement interval, estimating matched data quantity, calculating CZT, and updating an estimation value; judging whether the estimation value satisfies a preset precision demand or not, outputting a result if the estimation value satisfies the preset precision demand, otherwise adjusting an estimation value error, and estimating CZT with higher precision until the estimation precision satisfies the preset precision demand. According to the method, the difficulty that in the prior art, the estimation precision of the subtle features of radiation sources is low is solved; the spectrum refinement analysis is carried out by employing a multi-precision progressive CZT method; compared with a DFT method, the CZT method has the advantage that the symbol rate and carrier frequency with higher order of magnitude precision are obtained; the precision is enough to distinguish the subtle features of different radiation sources, and the reliable basis is provided for identifying the radiation sources.

Embodiments of the present invention provide a method (200) of providing image data for constructing an image of a region of a target object, comprising detecting, by at least one detector (40), at least a portion of radiation scattered by a target object (30) with the incident radiation (10) or an aperture at a predetermined probe position, determining an offset vector (203) for reducing an error associated with the probe position (201), estimating a wavefront (210) based on a probe function having the offset vector applied to the probe position, and providing image data responsive to the detected radiation.

The invention relates to a method for estimating precession parameters of middle-section targets of ballistic missiles, in particular to a method for estimating the precession parameters of the middle-section targets of the ballistic missiles. The objective of the invention is to solve the problems that an existing precession feature estimation method cannot utilize the micro-motion features of general scattering points such as an empennage and the like, and the existing precession feature estimation method is low in identification capability for true and false targets. The method comprises the following steps: 1, calculating a cosine value of an included angle between a radar A relative target direction and a radar B relative target direction; 2, obtaining a radar Doppler spectrum; 3, obtaining a boundary spectrum, searching the position of a frequency peak value, and taking the mean value of the position of the frequency peak value; 4, calculating the estimation result of the target coning frequency, and recording the phase of the frequency spectrum at the peak value position at the same time; 5, calculating the spin frequency of the target; taking a phase difference average value; 6, substituting all known parameters into the equation set, and solving thetaA, thetaB, thetaC and thetaP. The method is used in the field of signal processing.

The invention discloses a point spot detection and background noise feature estimation method for a point source Hartmann wavefront sensor. When the method is used for point spot detection and background noise estimation on a certain sub aperture, as for each pixel in the sub aperture, the difference between the gray variance in a neighborhood and the gray variance outside the neighborhood is calculated; the position of the pixel corresponding to the maximum difference in all pixels is then obtained, the neighborhood of the pixel serves as the detected point spot area, and the gray mean value and the gray variance in a non neighborhood of the pixel serve as the mean value and the variance for noise estimation. The method has a simple and stable process, and the realization is easy; compared with the existing background noise estimation method, the method of the invention can estimate the statistical characteristics of the background noise more accurately, the point spot area can be detected at the same time, and thus, the point source Hartmann wavefront detection stability and the precision are further enhanced.

The invention belongs to the field of visualization of artificial intelligence command control, and particularly relates to a visual imaging method fusing an SAR radar and an infrared imaging technology. The method comprises the steps that firstly, fusion feature estimation conversion is built through obtained dynamic information and static information; secondly, a fusion feature model is built through a training library; thirdly, fusion features are formed; and fourthly, high-precision detection and recognition are completed through searching in a feature level key region. According to the visual imaging method fusing the SAR radar and the infrared imaging technology, the SAR radar and the characteristics of infrared data are fused, and accurate, reliable and robust features are providedfor the detection of the quality of a high-precision combat unit.