46 results about "Total correlation" patented technology

A plurality of divided exposure images (first to fourth images) are obtained by dividing a normal exposure time. By an image matching method, a motion detection circuit calculates a plurality of cumulative total correlation values between different divided exposure images, and searches the plurality of cumulative total correlation values for the smallest cumulative total correlation value, thereby calculating a motion vector between the different divided exposure images. In addition, based on the level of the smallest cumulative total correlation value, the level of reliability B[i] indicating the accuracy of the motion vector is calculated. An image combining circuit uses the level of reliability B[i] as a mixture proportion. The image combining circuit positions the divided exposure images based on the motion vector, and combines the divided exposure images together one on top of another in mixture proportions (B[i]), so as to form one still image that suffers less from camera shake.

The invention relates to the field of precise engineering survey, in particular to a method for rapidly measuring a foundation pile control net (CPIII) of a rapid transit railway on the basis of movable laser radar measuring technology and high-speed digital imaging technology. control net (CPIII) measurement data is obtained by using a movable high-precision measurement system comprising a laser scanner (1), a GPS/IMU positioning and orientating system (2), a high-speed digital imaging system (3), an industrial computer (4) and a power supply device (5); and united weighted adjustment is carried out on advanced net data, united measured data and auxiliary measurement points extracted from images to obtain high-precision control net (CPIII) control point coordinates. The invention can greatly shorten the measurement time of the control net (CPIII), overcomes the defect of little total correlation of discrete single-point measurement, carries out integral adjustment on all control points, has good closing property and high relative precision, obtains plane and elevation coordinates of all control points, and greatly simplifies the measurement work of the control net (CPIII).

The invention discloses a wind turbine generator system fault diagnosis method and device based on gray correlation. The method comprises the following steps of supposing that m fault types exist, wherein each fault type can be represented by n fault character vectors; determining a character reference vector of each fault type, obtaining m*n dimensional character reference vector spaces of m fault types on the basis of all character reference vectors, and obtaining correlation coefficients of various character reference vectors in to-be-diagnosed vectors and character reference vector space according to a correlation coefficient calculation formula; obtaining total correlation degree of the to-be-diagnosed vectors to different faults of the m fault types by a correlation degree calculation formula, and performing normalized processing on the correlation degree to obtain a confidence value of the to-be-diagnosed vector in different faults; and performing Dempster combination rule fusion on multiple evidences according to a fusion formula to obtain the final diagnosis result. By adopting the method and the device provided by the invention, the confidence degree of the fault mode is greatly enhanced, thus the fault diagnosis can be carried out conveniently and effectively.

Operation of partial correlation or total correlation is carried out between first sync code and receiving signal. Operation of difference detection is carried out for the result of correlation operation. Based on result of partial correlation operation or result of difference detection, determination of time slot synchronization of first step for sub zone searching is carried out in order to raise capability of anti initial frequency error in sub zone searching. Using the said result of difference detection estimates frequency deviation between carrier signal generated by local oscillator and carrier signal in received signal, and controls corresponding frequency in order to reduce frequency deviation. Based on estimated frequency deviation, Operations of correlation for second and third steps in zone searching are selected. Thus, false detection rate in the said second and third steps are reduced, and receptivity of broadcast channel is raised after zone searching.

The present invention discloses a pedestrian detection method for the traffic environment based on the human body tree model, and belongs to the field of road traffic pedestrian detection. The method comprises: selecting a data set with annotation information of the human body joint as a training sample of the model, and expanding the joint into the required human body part; based on the relative position relation between each parent part and child part, using principles of the relative distance of the sample, the mean value of the sample correlation difference and the mean value of the total correlation difference of the sample set, optimizing the initial clustering center of the K-means algorithm to realize the clustering of the various parts of the human body, and obtaining hidden variables of the training samples; using a coordinate reduction method to solve the hidden structure SVM problem, and training, obtaining, detecting and determining the models; in the detection phase, according to the constructed human tree structure, the part state transition equation and the off-line training model, merging the dynamic planning idea to realize the traversal of the pyramid of the test sample, obtaining the whole human body detection result of the image, and using a non-maximal suppression algorithm to obtain the final detection bounding box.

The invention discloses a bill anti-counterfeiting identification method based on bill global characteristics. The method comprises steps that images of a bill under different spectrum irradiation conditions are acquired, the acquired bill images are pre-processed, and bill image parameters are made to be consistent with standard bill image templates; the bill images are transferred to an HSI color space, and bill tone, saturation and brightness component characteristic images are acquired; a tone, saturation and brightness component matrix of a bill global characteristic image and a standard image are detected to acquire tone, saturation and brightness correlation coefficients; a total correlation coefficient of global characteristics of the detected bill image and the standard bill image is calculated, the bill characteristic matching result is determined according to the correlation coefficient, and authenticity of the bill is determined on the basis. The method is advantaged in that intelligent global bill characteristic extraction can be realized, correlation evaluation of the HIS images of the global characteristics of a detected bill and a standard bill is automatically carried out, comparison with a preset threshold is carried out, authenticity of the bill is determined, and intelligent bill anti-counterfeiting identification is realized.

The invention discloses a community discovery method based on a multi-dimensional social network. In the method, multi-level fusion is conducted on a friend relationship network, a comment relationship network, a recommendation forwarding relationship network and an interest similarity network in the social network to obtain the total correlation degree of users; then, with each user regarded as anode and the total correlation degree of the users taken as the transmission probability, the community is divided by the label propagation algorithm, so that social discovery is completed. As users'social behaviors are comprehensively considered and reasonable selection and fusion are conducted, the method has great accuracy and effectiveness.

The invention discloses a direct sequence spread spectrum pseudo code capturing method. The method comprises a pseudo code storage step, a partial correlation operation step and a total correlation step. In the partial correlation operation step, a plurality of groups of partial correlation operation are completed in at least two continuous periods of initiative pseudo sequence, wherein any group of partial correlation operation is completed in the slot time of receiving at least two continuous code elements of the initiative pseudo sequence, and furthermore, after the previous period finishes, the initiative pseudo code sequence participated in the groups of partial correlation operation abandons a first code element, and then the next period of partial correlation operation is started. Due to the method, and according to the direct sequence spread spectrum pseudo code capturing method, the high-speed and long-period pseudo sequence can be effectively captured with lower cost and difficulty. The invention also discloses a direct sequence spread spectrum pseudo code capturing device.

The invention discloses a weak password detection method suitable for being executed in a server.The method includes the steps that a user name and a to-be-detected password input by a user are acquired; the correlation length pair and the total correlation length of the to-be-detected password are calculated; a first ratio of the correlation length pair and the total length of the to-be-detected password is calculated, and a second ratio of the total correlation length and the total length of the to-be-detected password is calculated; if the first ratio is not smaller than a first threshold value, or the second ratio is not smaller than a second threshold value, it is confirmed that the to-be-detected password is a weak password.The invention further discloses a weak password detection device and the server.

Methods and systems are provided for the deconvolution of the NMR spectrum of a mixture into individual components and spin systems by combining covariance total correlation spectroscopy (TOCSY) spectra with covariance NMR. The method may include obtaining a 2D TOCSY spectra of a chemical mixture and then performing a series of analytical steps to identify the individual components of the mixture.

The invention discloses a nuclear magnetic resonance detection method of a glycerol hydrogenation reaction mixture. The method comprises the steps of dissolving a glycerol hydrogenation reaction mixture sample in a deuterated reagent, putting the sample into a nuclear magnetic resonance spectrometer, detecting a component of the mixture sample by utilizing BPPLED pulse of a diffusion ordered spectroscopy technology, performing preliminary attribution, and then further confirming the component of the mixture sample by utilizing SELMLGP pulse of a one-dimensional selective excitation total correlation spectroscopy (TOCSY) technology. An analysis result obtained by the method is visual and clear; the detection sensitivity is higher; the glycerol hydrogenation reaction mixture sample can be well detected; and the method is also applicable to qualitative detection of a component of low carbon alcohol.

The invention relates to a neural network model training apparatus and method. The apparatus comprises an iteration calculation used for performing iterative calculation on a weight of a path in a neural network model; and a determining and output unit used for, when conditions for stopping iteration are satisfied, stopping the iteration and outputting a weight of the path of the iteration this time as a final weight. The iteration calculation unit comprises a weight calculation unit used for calculating the weight of the path of the iteration this time; a correlation function calculation unit used for randomly selecting one sample from a sample set for training the neural network model, according to the weight of the path of the iteration this time, calculating a correlation function of the selected sample, maintaining correlation functions of other samples besides the selected sample in the sample set unchanged, wherein the correlation function of the sample is a function associated with a loss function of the sample; and a total correlation function calculation unit used for calculating a total correlation function of the iteration this time according to the correlation functions of the selected sample and other samples.

The invention discloses an integrated hydrological station network evaluation model based on the combination of Copula function and information entropy theory, comprises the following steps: selectingpotential Archimedean family Copula by using semi-parametric evaluation method, calculating mutual information of two variables and total correlation (TC) of multiple variables; Based on the organiccombination of Copula function theory and information entropy principle, Copula function theory is use to determine that mutual information estimation value under the multi-dimensional variable, MACEsite optimization criterion (minimum value of copula entry, MACE) is proposed, which makes up for the shortcoming of single entropy estimation of mutual information. The invention also adopts three posterior evaluation indexes to have a final evaluation on the model effect: (1) Nash-Sutcliffe efficiency coefficient; (2) negative Copula entropy after averaging; (3) joint information entropy.

The invention discloses a total correlation feature selection method and device based on a Shapley value and hypothesis testing. The method is suitable for feature sets with supervised tasks. A feature selection model for solving a total correlation problem is designed, the model firstly uses a Shapley attribution algorithm to calculate the local importance of features, then uses random features to construct an adaptive threshold, and then uses the importance and the threshold to evaluate the correlation of the features. In the aspect of a selection strategy, dual hypothesis testing is designed, irrelevant features are quickly eliminated by utilizing local hypothesis testing, and then the risk of deleting the relevant features by mistake is reduced by utilizing global hypothesis testing. And finally, all the features related to the problem domain are obtained, so that the interpretability of the feature set is improved, and the prediction reliability is enhanced.

The invention discloses an Alzheimer's disease classification method, an Alzheimer's disease classification system and an Alzheimer's disease classification device based on Gaussian process classification. The Alzheimer's disease classification method comprises the steps of: acquiring magnetic resonance imaging data of Alzheimer's diseases; extracting key features used for Alzheimer's disease classification from the acquired magnetic resonance imaging data by adopting a key feature extraction algorithm based on a total correlation coefficient; and classifying data to be classified by adoptinga Gaussian process classifier according to the extracted key features to obtain a classification result of the Alzheimer's diseases. The Alzheimer's disease classification system comprises a data acquisition module, a feature extraction module, and a classification module. The Alzheimer's disease classification device comprises a memory and a processor. The Alzheimer's disease classification method, the Alzheimer's disease classification system and the Alzheimer's disease classification device improve the feature extraction efficiency of the Alzheimer's diseases by adopting the key feature extraction algorithm based on the total correlation coefficient, ensure the classification performance of the Alzheimer's diseases by means of the Gaussian process classifier, are easy to implement, havebetter nonlinear processing performance, and can be widely applied to the field of computer aided diagnosis.

The present invention provides a structure stability determining method for a polymeric CA (ComplexingAgent). The method comprises: acquiring a molecular configuration, a molecular weight and multiple different preset temperatures of a target CA; obtaining an expression of an intra-molecular correlation function of the target CA by means of calculation; by use of PY (Percus-Yevick) approximation, establishing a closed equation that contains a direct correlation function and a total correlation function of the target CA; establishing a PRISM integral equation that contains the direct correlation function, the total correlation function and the intra-molecular function of the target CA; calculating the closed equation and the PRISM integral equation, so as to obtain an expression of the direct correlation function and an expression of the total correlation function, which correspond to the preset temperatures; carrying out calculation to obtain an X- light scattering intensity, which corresponds to a preset temperature, of the target CA; and according to X-light scattering intensities corresponding to the multiple different preset temperatures, determining structure stability of the target CA.

A signal processing method is adapted for dealing with a plurality of vector matrixes to detect the image of a predetermined range, and the vector matrix data are generated by reflecting a plurality of ultrasonic beams in the predetermined range. The signal processing method of the present invention is that summing all vector matrix data in a predetermined time interval so as to generate a total correlation matrix. In addition, obtaining a correlation matrix through the total vector matrix multiplied by a transposed total vector matrix, and obtaining a weight value according to inversion correlation matrix. Then, a weighting operation is performed for the vector matrix data in the predetermined time interval according to the weight value, so as to obtain a weighting operation result for performing an image synthesis procedure.

System for building a geological model of oil or other mineral deposit. The system determines functionals represented by the sum of coefficients of correlation for a set of well log curves. A technical result is the improvement of the accuracy of evaluating parameters that are used to build geological model of the location of oil or other deposits. The system makes it possible, given set of markers, chosen as initial solution and to calculate their depths at each well to ensure the best total correlation. For any marker in the set, functional is determined, represented by the sum of correlation coefficients for the set of well log methods for pairs of wells located within specified distance from one another. Partial derivatives are evaluated for this functional, and the vector thus obtained is smoothed and used to search for a larger value of the functional within certain segment along this vector.

The invention discloses a station network optimization method based on high-dimensional Copula entropy and Kriging. The method comprises the following steps: (1) constructing a hydrological C-Vine Copula tree structure; (2) estimating a C-Vine Copula parameter by adopting a maximum likelihood estimation method; (3) obtaining high-dimensional mutual information through a function relationship between the multivariable mutual information and the C-Vine Copula density; and (4) optimizing the dynamic rainfall station network through a standardized MiK-MiT-MaJ index and a sliding window method. According to the method, a high-dimensional dependency structure among multiple stations is obtained by adopting C-Vine Copula, and the total amount and the total correlation amount of station network objective function information are optimized; the optimal rainfall station network estimation error and the optimal rainfall information are realized by using the Kriging standard error value; multi-objective optimization is simplified into single-objective optimization, optimization efficiency is improved, and rainfall sequence time-varying characteristics are considered to cause dynamic characteristics of a station network optimization result.

The invention discloses a large-scale MIMO channel model modeling method based on random coupling. The method comprises the following steps: obtaining unparameterized unilateral correlation matrixes RMS and RBS of an MS side and a BS side by using a total correlation matrix; setting the feature bases of the RMS and the RBS as a BS-side guiding matrix which does not change along with a spatial covariance matrix, and setting Nakagami distribution on link fading between the feature bases of the BS side and the MS side; decomposing unparameterized RMS features to obtain a feature base of an MS side, and generating an independent identically distributed matrix O through Nakagami distribution; and constructing a channel impulse response matrix. According to the method, the steering vector is used, the coupling between the steering matrix of the BS side and the feature base of the MS side is used for representing the correlation matrix, the Nakagami distribution is used for describing the fading of the coupling between the steering vector and the feature base, the complexity is low, the accuracy is high, and the method is suitable for a random CIRM model of a large-scale MIMO channel.