81 results about "Fisher information" patented technology

The distance of objects to an optical system is estimated. An optical mask such as a diffractive optical element, continuous phase mask, hologram, amplitude mask, or combination thereof is placed within the optics in front of a sensor array such as a CCD, CID or CMOS device. The optical mask encodes the three-dimensional response of the system. The mask is designed to optimize depth estimation, for example, by maximizing Fisher information. A particular implementation creates a point spread function (“PSF”) that rotates as a function of the object position. The image or images obtained with different PSFs may be digitally processed to recover both a depth map of the scene and other parameters such as image brightness. The digital processing used to recover the depth map of the object may include deconvolution of a PSF from detected images.

The invention discloses a composite phase mask, the phase distribution function of which contains two different functions, namely, a sinusoidal type function and a third power type function. The function expression of the phase distribution function Theta(x, y) is as follows: Theta(x, y) is equal to Alpha question mark (x plus y) plus Beta question mark (sin(Omega question mark x) plus sin(Omega question mark y)), and in the expression, Beta question mark (sin(Omega question mark x) plus sin(Omega question mark y)) is a sinusoidal type function, and Alpha question mark (x plus y) is a third power type function; wherein Alpha is the amplitude of the third power type function, Beta is the amplitude of the sinusoidal type function, Omega expresses the angular frequency of the sinusoidal type function, and x, y is space coordinate with normalized aperture plane. The invention also discloses an imaging system which adopts the composite phase mask compounded by the sinusoidal type function and the third power type function and has smaller numerical value of the Fisher information, and the phase mask is not sensitive to defocusing, thus having better expansion capability of depth of focus.

The invention discloses a small current grounding system single-phase ground fault line selection method based on fisher information. The method includes: S1, a discrete wavelet coefficient is divided into a series of time windows; S2, data points in each time window are packaged into a state; S3, a probability density function pi based on possibility of observing a system state is established in each time window; S4, the fisher information is calculated from the probability density function for each time window; and S5, a line with the minimum fisher information value is selected to be a fault line. By means of the method, under different grounding methods, fault lines and non-fault lines can be distinguished effectively, effects of voltage initial phase angles, ground resistance and fault location are avoided, and fault moments can be located accurately. Under the adverse conditions of short data, small amplitudes and disturbance existence and the like, feature information hidden in a signal sequence can be effectively extracted, simultaneously a complex degree of each signal sequence can be expressed quantitatively, and an effective means for analyzing a non-stable signal sequence is provided.

The invention belongs to the technical field of radar resource utilization, and particularly relates to a radar time resource joint distribution method related to the residence time and the re-access interval. The radar time resource joint distribution method related to the residence time and the re-access interval includes the following steps of firstly, setting the moment k, wherein k equals to 0,1,2...; secondly, establishing a Fisher information matrix of each target at the k moment; thirdly, conducting recurrence to obtain a Fisher information matrix of each target at the k+1 moment; fourthly, calculating the target i, representing the Cramer-Rao bound obtained when radial distance estimation is conducted on the ith target at the k+1 moment as var(si, k+1), setting the error threshold Th1, executing the fifth step if the var(si, k+1), larger than Th1, exists for the corresponding target, and if not, enabling k to be equal to k+1 and repeating the third step, wherein i is larger than or equal to 1 and smaller than or equal to N, and N is the sum of the targets; fifthly, selecting the target irradiated by the radar at the k+1 moment, and obtaining the residence time (please see the specification) of the radar for the j(k)<*>th target at the k+1 moment, wherein the target irradiated by the radar at the k+1 moment is the j(k)<*>th target, and j(k)<*> is larger than or equal to 1 and smaller than or equal to N; sixthly, enabling the value of k to increase by 1 by itself, irradiating the j(k)<*>th target according to the residence time (please see the specification) of the radar for the j(k)<*>th target at the k moment, renewing the Fisher information matrix J[j(k)<*>],k of the j(k)<*>th target at the k moment, and then repeating the third step.

Embodiments include methods, systems, and / or devices that may be used to image, obtain three-dimensional information from a scene, and / or locate multiple small particles and / or objects in three dimensions. A point spread function (PSF) with a predefined three dimensional shape may be implemented to obtain high Fisher information in 3D. The PSF may be generated via a phase mask, an amplitude mask, a hologram, or a diffractive optical element. The small particles may be imaged using the 3D PSF. The images may be used to find the precise location of the object using an estimation algorithm such as maximum likelihood estimation (MLE), expectation maximization, or Bayesian methods, for example. Calibration measurements can be used to improve the theoretical model of the optical system. Fiduciary particles / targets can also be used to compensate for drift and other type of movement of the sample relative to the detector.

The invention discloses an evaluation method for RSS (Received Signal Strength) / AOA (Angle of Arrival) / TDOA (Time Difference of Arrival) positioning performance under an indoor WLAN (Wireless Local Area Network) environment. The evaluation method comprises the following steps: firstly, making different observed quantities to be subject to the respective Gaussian distribution respectively according to a statistical property of signal distribution under the indoor WLAN environment; then using the characteristic of a Fisher information matrix to respectively calculate the error limits of individual RSS, AOA and TDOA positioning methods; secondly, using a multivariate normal distribution function to describe probability distributions of multiple observed quantities, and respectively calculating to obtain the error limits of RSS / AOA, RSS / TDOA, AOA / TDOA and AOA / TDOA / RSS mixed positioning methods; finally, quantitatively analyzing the influence of the positioning method based on different observed quantities to the error limit, and comprehensively evaluating the positioning performance of the RSS / AOA / TDOA. The evaluation method for RSS / AOA / TDOA positioning performance under the indoor WLAN environment improves the positioning precision of the indoor WLAN positioning method.

The invention discloses a target tracking method based on Fisher information matrix and SUKF, and relates to the field of communication technologies, aiming at addressing too many nodes that are required in the control methods of current wireless network nodes and ineffectiveness caused by the need to cluster the entire network when static cluster node is employed to perform selection. According to the invention, the method includes the following steps: establishing a dynamic tracking cluster based on Fisher information matrix method; using the SUKF-based target tracking filter algorithm to conduct filtering and information delivery on the established dynamic tracking cluster. According to the invention, the method herein prevents all observation data from being centralized processed in current centralized filter, lowers energy loss in the process of transmitting data and receiving data, and reduces burden to cluster heads. The task of the filter algorithm is simultaneously and distributively executed by a plurality of sub-nodes. The method herein allocates the work of cluster heads to respective member nodes, so that the method greatly reduces data transmission among nodes, and prevents data collision and interference. Through simulation comparison with the centralized filter algorithm, the method herein reduces network energy consumption at the maximum level and increases estimation precision.

The invention discloses a generation method of an OTH (Over-The-Horizon) radar transmitting waveform, belongs to the technical field of radar communication, and in particular relates to a generation method of the OTH radar transmitting waveform based on a parameter estimation performance of an object detected. The method comprises the steps of: obtaining a possible waveform to be transmitted according to current available frequency range and detection requirements; then, inversing a Fisher information matrix F (theta) based on associated parameter of a radar base station, state of an ionized layer, frequency shift of the ionized layer, sea clutter, noise base and the like; and selecting the waveform with the minimum first and second elements on the diagonal line in a matrix F-1(theta) as the transmitting waveform of the mth transmitting antenna to obtain the transmitting waveform with optimal detection performance. The generation method provided by the invention is suitable for various conditions in which parameters of the radar base station, the state of the ionized layer, the sea clutter, the available frequency range and the like change, and further suitable for MIMO-OTH (Multiple Input Multiple Output-Over-The-Horizon) radars and conventional phased array OTH radars.

The invention relates to a layout optimization method of a planet navigation orbiter based on a prediction track, belonging to the technical field of deep space detection. The layout optimization method provided by the invention has the advantages that a dynamic moving track of a navigation orbiter and a detector in an inlet section in a process that the initial layout of the planet navigation orbiter is optimized is considered, the observability of a Fisher information matrix derivation navigation system is utilized for representing a navigation property, the initial layout of the planet navigation orbiter is optimized based on the integral of the observability to a navigation system, the maximization of the observability of the navigation system is realized, the estimated accuracy of the state when a detector enters is improved, and the optimum of a navigation property is guaranteed.

Provided is a time-to-go online estimation method, which is characterized by establishing models through measured missile-target relative distance and a system identification method and a regression analysis method; estimating miss distance through the established models and the currently known missile-target relative distance to obtain a miss distance estimated value, time corresponding to the miss distance estimated value being intercept time of a predicting missile for a target; calculating a time interval value between missile-target relative distance estimation time and the intercept timeof the predicting missile for the target, and multiplying the time interval value with a sampling period to obtain a time-to-go estimation component of the current time; and obtaining a time-to-go fusion estimation value through a Fisher information fusion method. The time-to-go online estimation method only needs the missile-target relative distance and the sampling period, wherein the missile-target relative distance is obtained through measurement and the sampling period is given by a guiding system, so that the method has better calculation efficiency and feasibility; and besides, the method has higher robustness.

The invention discloses an accelerated life test optimization design method based on a proportional hazards-proportional odds model. The method comprises the following steps: establishing a logarithmic likelihood function based on the proportional hazards-proportional odds model; establishing a Fisher information matrix and an asymptotic variance-covariance matrix based on the proportional hazards-proportional odds model; establishing an objective function and constraint conditions of an optimization problem by adopting an information-based optimization method; and obtaining an accelerated life test optimization design scheme by solving the optimization problem. The invention provides an optimization design method for accelerated life test of constant stress and step stress, the model adopted during the accelerated life test optimization design is a proportional hazards-proportional odds model, the model has wider application range and higher evaluation precision than the proportional hazards model and the proportional odds model; the model is a nonparametric model and has the distribution-free characteristic; and the method avoids the problem of inconsistent optimization results caused by integral interval change in the traditional optimization method.

The invention belongs to the technical field of civil engineering structure health monitoring, and provides a sensor arrangement method for reducing the uncertainty of structural mode recognition. Theinfluences of structure model errors and measurement noises on measurement data are separated; the structural rigidity changes are used as the model errors; and Gaussian noises are used as the measurement noises. By adopting a Monte Carlo method, a large number of possible situations are simulated to obtain a structural mode matrix under the condition of the model errors; a conditional entropy index is proposed for quantifying and calculating the uncertainty of a mode recognition parameter result; and the problem of an uncertain Fisher information array, which cannot be solved by a traditional information entropy method is solved by using the conditional entropy index. A position corresponding to the minimum value of the conditional entropy index is an optimal sensor arrangement position.According to the sensor arrangement method provided by the invention, the influences of the structural mode errors and the measurement noises on the structural mode recognition are fully considered;and a great help is provided for improving the precision of structural mode parameter recognition.

The invention relates to a lower bound establishing method for multi-antenna channel characteristic parameter joint estimation, which relates to the technical field of wireless communication and parameter estimation. A lower bound can be used as an evaluation index when an estimation value and an estimation variance of time-variable channel characteristic parameters such as channel attenuation, time delay and frequency bias are obtained by adopting an estimation method for a multi-antenna channel. The method comprises the steps of initializing a parameter to be estimated and an estimation variance of the parameter; deducing a one-step status transfer probability density function and an observation likelihood function according to a channel characteristic parameter model; and utilizing a recursion way to obtain a Fisher information matrix, and obtaining a Cramer-Rao lower bound after inverting the Fisher information matrix.

The invention discloses a composite phase mask, the phase distribution function of which contains two different functions, namely, a sinusoidal type function and an exponential type function. The function expression of the phase distribution function Theta(x, y) is as follows: Theta(x, y) is equal to Alpha question mark x question mark (Epsilon x Rho(Beta x plus Epsilon x Rho(Beta y)) plus (1 / 2)Alpha(sin(Omega x) plus sin(Omega y)), and in the expression, Omega question mark x question mark (Epsilon x Rho(Beta x plus Epsilon x Rho(Beta y)) is an exponential type function, and (1 / 2)Alpha(sin(Omega x) plus sin(Omega y)) is a sinusoidal type function; wherein Alpha expresses the amplitudes of the exponential type function and the sinusoidal type function, Beta expresses the modulation factor of the exponential type function, Omega expresses the angular frequency of the sinusoidal type function, and x, y is space coordinate with normalized aperture plane. The invention also discloses an imaging system which adopts the composite phase mask compounded by the sinusoidal type function and the exponential type function; and the imaging system has smaller numerical value of the Fisher information, is not sensitive to defocusing, has better inhibiting effect on defocusing amount and has better expansion capability of depth of focus.

The invention belongs to the field of visual tracking, and provides a target tracking algorithm based on color attributes and active feature extraction for solving the problems of low speed, low robustness and inaccurate target tracking of the existing target tracking algorithm. The target tracking algorithm comprises the following steps: firstly, graying an image of a t frame, mapping the image from a RGB color space to a 11-dimensional color attribute space, mapping the image to a two-dimension by using a main component analysis method, combining gray scale information with color information, performing sampling according to a search strategy from thick to thin, mapping extracted high-dimensional multi-scale image features to a low-dimensional subspace via a very sparse random matrix by using the compressed sensing theory, finally constructing a strong classifier based on a criterion function constructed based on a Fisher information matrix, namely an active feature extraction process, classifying the features by using the strong classifier, and selecting the sample having the maximum reaction value of the classifier as the tracking target. By adoption of the method, the shortcomings of the prior art can be overcome, the algorithm obtains balance on illumination invariance and discrimination, the robustness is relatively good, the drift problem can be effectively solved, online tracking can be accomplished in real time, and the target tracking algorithm has important practical significance.

The invention relates to a navigation pulsar selection method based on a Fisher information matrix, and belongs to the technical field of deep space exploration. By using the method, a measurement simplification model for pulsar navigation is constructed; based on a relevant concept of an information theory, the Fisher information matrix in a navigation state is constructed; a navigation pulsar is selected according to a rule that the number of rows and the number of columns of the Fisher information matrix are maximum; the navigation state observability is improved; and the observability and the navigation precision of pulsar navigation are improved.

The invention relates to a short-term load processing method based on multidimensional meteorological factors of Fisher information processing, comprising the following steps: analyzing historical sample data to obtain multidimensional meteorological factor Fisher information window data of a corresponding load; calculating the Fisher information value of the multidimensional meteorological factorFisher information window data on the basis; weighting the Fisher information value to obtain input variables of Fisher information weighting; and inputting the input variables and historical load data to an artificial intelligence prediction model. The comprehensive index is weighted through calculation of multidimensional meteorological factor Fisher information, so as to realize the coupling effect and cumulative effect of multiple meteorological factors on a load and realize accurate prediction of a weather-sensitive load.

The invention provides a multi-AUV cooperative simultaneous guidance method based on information maximization. Aiming at the problem of time constraint and maximization of observation information in multi-AUV cooperative guidance, Considering the second-order nonholonomic characteristics of AUV, a cooperative guidance law based on optimal control is proposed, the performance index function consists of one-step sub-optimal Fisher information matrix (FIM) and the variance of range between AUV and target, which realizes the task of hitting target at the same time under the maximum observation information, and has good guidance performance.

The invention provides an optical remote sensing image target detection method based on Bayesian transfer learning. The method is characterized in that the knowledge contained in a pre-trained target detector on a source data set is represented by using a Fisher information matrix; the Fisher information matrix is utilized to construct a target function of a target detector, and in the training process, the item participates in learning of the target detector on a target data set, so that the learned knowledge is reserved to a certain extent, and the detection precision is improved. Compared with other existing optical remote sensing image target detection algorithms, according to the method, under the premise that no extra to-be-learned parameter is introduced, the optical remote sensing image target detection precision is effectively improved, the human interpretation can be efficiently and accurately assisted, and the manpower is saved.

An optimal input design method and apparatus to achieve rapid broadband nanomechanical measurements of soft materials using the indentation-based method for the investigation of fast evolving phenomenon, such as the crystallization process of polymers, the nanomechanical measurement of live cell during cell movement, and force volume mapping of nonhomogeneous materials, are presented. The indentation-based nanomechanical measurement provides unique quantification of material properties at specified locations. Particularly, an input force profile with discrete spectrum is optimized to maximize the Fisher information matrix of the linear compliance model of the soft material.

The present invention discloses a fisher integration identification terminal. The fisher integration identification terminal comprises a data processing module, a dual-interface CPU module, a seaman identity identification module, a satellite location module, a RFID processing module and a GPRS processing module which are coupled with the data processing module. The dual-interface CPU module is configured to store the original basic information of a fisher; a fisherman card or identity reading module is configured to obtain the fisherman information; a GPS / Beidou location module is configured to obtain fisher position information; the data processing module is configured to process and integrate the fisher integration information and actively transmit the fisher integration information to a data service center through the GPS processing module, or the fisher integration information is sent outwards for a long distance through the RFID processing module, the fisher integration information is obtained through a handset or a shipborne RFID reader, and then the fisher integration information is transmitted to the data service center. The fisher integration identification terminal and the fisher integration identification system are configured to bind the fisher information and the shipman information and upload data in real time based on the RFID, GPRS and GPS technologies so as to realize the common management of the fisher and the fishermen.

The invention provides an adaptive weak measurement method and system for parameter estimation. The adaptive weak measurement method comprises the steps of performing weak measurement to obtain an estimated value of a parameter to be measured; adjusting the pre-selection / post-selection state of the system based on an adaptive algorithm according to the obtained estimated value; repeatedly executing weak measurement and pre-selection / post-selection state adjustment until the pre-selection / post-selection of the system achieves the optimum, namely, the pre-selection / post-selection which enables the measured Fisher information amount to be the maximum. The pre-selection / post-selection state of the weak measurement system is adjustable, the pre-selection / post-selection state of the system is adjusted to the optimum according to the adaptive algorithm when the parameter to be measured changes, and the Fisher information amount obtained by the system is enabled to achieve the maximum, so thatthe accuracy of micro-parameter estimation is greatly improved.

A route sign planning method of deep space detection autonomous navigation comprises the following steps: giving a navigation system information matrix expression based on visual line information according to the definition of a Fisher information matrix; establishing a relationship expression between an error variance matrix and the information matrix by utilizing a Cramer-Rao inequation, carrying out observability analysis based on the characteristic value information of the error variance matrix, and quantitatively assessing the influence of a route sign on the navigation system observability; and finally selecting an appropriate route sign to realize the largest autonomous navigation observability.

The invention discloses an incremental learning method based on a Fisher information matrix. According to the method, the importance of parameters of each layer of the neural network is judged by calculating a Fisher information matrix; by separating interlayer parameters and intra-layer parameters, the calculation efficiency of the Fisher information matrix is greatly improved; by introducing new vehicle incremental data and iteratively training the model, the huge calculation amount of full data training is avoided; corresponding weight coefficients are set for parameters with different importance, so that the priori model and the posteriori model have the maximum similarity; the Fisher information matrix is associated with the classical KL divergence, and more powerful support is provided for manifold hypothesis of the neural network. The test result shows that parameters of all layers of the neural network indeed have different importance, and the distance between the prior model and the posterior model of the neural network can be effectively controlled by introducing the regular term, so that the distance is as small as possible.

The invention provides a three-dimensional reconstruction error boundary evaluation method based on a Fisher information matrix, and relates to the technical field of computer graphics and computer vision. In order to determine the propagation process of the error in the three-dimensional reconstruction process, the upper and lower boundaries of the three-dimensional reconstruction error are givenso as to realize the solution of the information such as the essential matrix in the three-dimensional reconstruction process, determine the propagation process of the error in the three-dimensionalreconstruction process and give the upper and lower boundaries of the three-dimensional reconstruction error. The data analysis process of the calculation model is deduced by using the Fisher information matrix with combination of the lower boundary value mathematical model of three-dimensional reconstruction error propagation, and the lower boundary value calculation expression of three-dimensional error propagation is obtained and finally the eigenvalue matrix is obtained, and the actual deviation of the three-dimensional reconstruction points on x, y, z axes after the secondary projection is determined. The lower boundary after the secondary projection is measured by the value of the deviation. Meanwhile, the influence of some external parameters of the camera is considered and used forcalculating the relative lower boundary value of three-dimensional reconstruction error in calculation of the mathematical model.

The invention discloses a method for estimating gain correlations of MIMO radar channels and belongs to the technical field of channel parameter estimation. The method comprises the steps of setting the false alarm probability and the minimal acceptable detection probability; calculating Fisher information when channel correlations use default values; using the local maximum potential to construct statistics of sampling signal matrixes, and calculating the channel correlation detection probability according to a unilateral detection principle; and changing default values of channel correlations to estimate channel gain correlations again when the channel correlation detection probability is lower than the minimal acceptable detection probability. According to the method, statistics of the local maximum potential are constructed, so that the defect that methods for estimating traditional radar channel gain correlations are complex in calculation can be eliminated.

The invention provides a pulsar selection method based on a Fisher information matrix under a multi-aircraft system. The method comprises the following steps: step I, establishing a kinetic model of a multi-aircraft system to obtain a state equation of the system; step II, establishing an observation model of the multi-aircraft system based on pulsar navigation to obtain an observation equation of the system; step III, establishing a Fisher information matrix of the pulsar navigation under the multi-aircraft system based on the state equation and observation equation; and step IV, selecting a pulsar with a minimum index function as the pulsar observed for the navigation, wherein the index function is a locus of an inverse matrix of the Fisher information matrix of the pulsar navigation under the multi-aircraft system obtained in the step III. By adopting the method, while the pulsar selection index is established, the kinetics information and observation information of the navigation system are also considered; and relative to the situation of only considering the observation information, the information contained in the Fisher information matrix is more complete, so that more effective evaluation can be realized, and the navigation precision is higher.

The invention discloses a remote sensing image target detection method based on random access memory, which is based on SSD network, maximum posteriori distribution, Fisher information, Laplace approximation and other methods. The method comprises the following specific steps: 1, reading image data and preprocessing the data; 2, constructing a convolutional neural network based on a fusion featurepyramid; 3, training the convolutional neural network to obtain static model parameters; 4, proposing a random access memory idea; and 5, detecting a remote sensing image target, and dynamically updating model parameters. According to the invention, a remote sensing image multi-class target detection algorithm under a unified framework is proposed to adapt to detection of the remote sensing imagetarget, and a random access memory thought is proposed by using maximum posteriori distribution to dynamically adjust model parameters in a detection stage, so that the model has good adaptive capacity and generalization capacity to new data. Remote sensing images are input, target category and position information are output, the automation degree is high, the efficiency is greatly improved, andthe cost is reduced.

The invention provides a detection method and a detection system for system dynamics structural mutation. The system comprises a sliding time window definition module 1, a system state probability density distribution calculation module 2, a Fisher information value calculation module 3 and a mutation detection and observation module 4, wherein the sliding time window definition module 1 is used for defining a sliding time window on an observation data sequence set for describing the condition that the system state is changed along with time; the system state probability density distribution calculation module 2 is used for calculating the probability density distribution Pm (Zl) corresponding to a ZI interval in the sliding time window; the Fisher information value calculation module 3 is used for calculating the Fisher information value FI according to the probability density distribution Pm (Zl); and the mutation detection and observation module 4 is used for drawing a change curve of the FI value along with time, and according to the change curve, judging the system dynamics structural mutation situation. The detection system and the detection method disclosed by the invention have wide adaptability and certain antijamming capability, and do not depend on the length and the amplitude of an analyzed signal.