42results about How to "Solve a large amount of computation" patented technology

This invention relates to a method for testing directions of coherent signal sources under non-stable noise background including: collecting data received by channels and storing them in an EMS memory of the system, making adaptive uniform process to data received by the channels to form an array covariance matrix, utilizing a forward space smooth technology to preprocess it smoothly to constitute a forward space smooth covariance matrix and utilizing a forward and backward space smooth technology to smoothly pre-process the matrix to constitute a generalized covariance matrix based on the covariance matrix and a conjugated matrix and carrying out spectrum estimation to the generalized matrix with a sub-space class algorithm to identify a real incident direction of image false peak and signal source and output the real direction of the estimated signal source.

The invention provides a method, a device and a system for positioning a sound source. The method comprises the following steps: acquiring the signal of each microphone in a microphone array, wherein the microphone array is used to collect the sound of a sound source; acquiring controllable power responses of multiple microphone pairs composed of the microphones according to the framing of the signals; acquiring the sum of the controllable power responses of the multiple microphone pairs, and determining the direction angles of the sound source and the microphone array according to the maximum value of the sum of the controllable power responses; and determining the relative position relationship between the microphone array and the sound source according to the direction angles. The problem that the existing controllable power response technology is of low resolution and high computation burden is solved. The real-time performance, stability and accuracy of sound source positioning are improved.

The invention discloses a device and a method for detecting similar texts. The device and the method are used for recognizing the similar texts. The device comprises a Chinese text acquiring unit, a pinyin text acquiring unit, a fingerprint acquiring unit and a detecting unit, wherein the Chinese text acquiring unit is suitable for processing texts so as to acquire Chinese texts; the pinyin text acquiring unit is suitable for transforming Chinese characters in the acquired Chinese texts into pinyin so as to obtain pinyin texts; the fingerprint acquiring unit is suitable for extracting features of the pinyin texts, and forming the feature vector of the pinyin texts by using the extracted features; and the detecting unit is suitable for judging whether the texts to be detected are matched with records in a database or not according to the feature vector. By using the device and the method, the Chinese texts can be obtained by the texts to be detected, then the pinyin texts are obtained, the feature vector of the pinyin texts is formed, whether the texts to be detected are matched with the records in the database or not is judged according to the feature vector, and beneficial effects of reducing operand and accurately recognizing variety of similar texts are achieved.

The invention discloses a non-uniform array design and direction of arrival (DOA) estimation method and mainly aims at solving the problems that arrangement is inflexible and computation complexity isrelatively high in the prior art. A realization process of the method comprises the following steps: according to maximum freedom degree of a nested array, determining head and tail position coefficients of a non-uniform array; calculating position coefficients of all the virtual array element positions meeting the non-uniform array and contained in a differential synthetic matrix; according to the position coefficients, finally obtaining positions of array elements of the non-uniform array; according to received data, calculating a data covariance matrix and vectorizing, so that received data r of a virtual differential synthetic array is obtained; performing redundancy elimination on the r and sorting to obtain received data of a virtual array, and then obtaining a non-singular matrix vof the virtual array; constructing a linear operator, and obtaining a signal subspace by estimating the linear operator; and constructing a selection matrix to obtain a rotation matrix, and finally estimating a direction of arrival by virtue of the rotation matrix. The method disclosed by the invention has the advantages that array configuration is flexible, characteristic decomposition does notneed to be performed on the data covariance matrix and spectral peak search does not need to be performed on the whole airspace angle under the same conditions.

The invention discloses a unite-domain positioning space-time two-dimensional self-adaptive processing method based on data fitting, and relates to radar technology. The method comprises the steps that 1, JDL dimensionality reduction echo data and space-time guide vectors are obtained; 2, a data basis matrix of a distance unit to be detected is constructed; 3, a data basis matrix with targets blocked is obtained; 4, the fitting coefficient of auxiliary echo data is obtained; 5, the minimum fitting error of data of the distance unit to be detected is obtained; 6, unit average constant false alarm rate detection is carried out on the minimum fitting error of the data of the distance unit to be detected. The unite-domain positioning space-time two-dimensional self-adaptive processing method mainly solves the problem that a traditional STAP method is huge in computation burden and strict in sample requirement, reduces the false alarm rate, improves movable target detection probability, and has the good application prospect in measured data processing and STAP performance improvement.

The invention provides a laser three-dimensional imaging radar ground target point cloud data processing method, which is used for carrying out ground target detection and identification on laser three-dimensional imaging radar point cloud data and outputting a suspected interested target, and comprises the following steps: S1, obtaining a point cloud data set of time and space registration; s2, filtering non-ground point data to obtain a ground point data set; s3, extracting, segmenting and summarizing the data of each ground target, and clustering and dividing the data into respective data subsets to obtain a ground target clustering set; and S4, performing interested target identification based on the ground target clustering set, and outputting the suspected interested target. The method is used for solving the problem of ground target detection and recognition in a complex scene, more three-dimensional information of the target is reserved, the operation speed is increased, and the method is more suitable for point cloud data real-time processing and engineering application.

The invention provides an object recognition method and device, wherein, the method comprises the following steps: acquiring a region of interest formed for each scanning depth; wherein the size of the region of interest on the imaging plane is smaller as the scanning depth is deeper; collecting road condition images of the road ahead; a region-wide scan is performed in each region of interest toobtain a recognition result. By using the method, different regions of interest can be formed for different scanning depths, the image range of object recognition can be reduced, the computation amount can be reduced, the object recognition speed can be improved, and the technical problems of large computation amount and long object recognition time caused by only one large region of interest in the prior art can be solved.

The invention provides a data processing method and a storage and calculation integrated chip, the method is applied to the storage and calculation integrated chip, the storage and calculation integrated chip comprises a calculation core, and the calculation core comprises a storage calculation array and a calculation module; the method comprises the steps that the storage and calculation array calculates data of a first neural network which is input into the storage and calculation integrated chip and has a first network attribute; and the calculation module is used for calculating data of asecond neural network which is input into the storage and calculation integrated chip and does not have the first network attribute. By means of the data processing method and device, the problems that in the prior art, in the process of processing the data through a storage and calculation integrated chip, all the data need to be stored in a calculation array for calculation, and consequently thestorage and calculation array is large in calculation amount and high in power consumption are solved.

The invention discloses an image enhancement method, which relates to the artificial neural network field. The method comprises the following steps: acquiring an original image; inputting the originalimage into a preset image enhancement network, and performing step-by-step enhancement on the original image to obtain an enhanced image corresponding to the original image, wherein the image enhancement network comprises at least two levels of sub-networks, the original image is used as input to a first sub-network of the image enhancement network, the first-level sub-network outputs a first-level node image, the node image output by each-level sub-network is used as an input of a next-level sub-network, the last-level sub-network of the image enhancement network outputs a last-level node image, and the last-level node image is used as the enhancement image. The first-level sub-network outputs a first-level node image, and the node image output by each-level sub-network is used as an input of a next-level sub-network. The image enhancement method provided by the invention can effectively improve the convergence of image enhancement operation, reduce the amount of operation required for image enhancement, and improve the operation efficiency and stability of image enhancement.

The invention provides a method for quickly imaging for a double-linear-array three-dimensional imaging synthetic aperture radar on the basis of self-adaption, which aims at the characteristic that an echo wave of an actual double-linear-array three-dimensional imaging synthetic aperture radar only comprises an echo wave signal of a specific curved surface in a three-dimensional space. A self-adaptive curved surface prediction technique is adopted to carry out curved surface prediction and imaging processing aiming at the specific curved surface in the three-dimensional space so as to better solve the problem of large computation workload of the prior imaging method of the double-linear-array three-dimensional imaging synthetic aperture radar. The method can be applied to the fields of the synthetic aperture radar imaging, the earth remote sensing, the geology surveying and drawing, and the like.

The invention provides an MMC capacitor voltage-sharing strategy with low switching frequency and complexity. The MMC capacitor voltage-sharing strategy comprises the steps of S1, rapidly monitoring acapacitor voltage value of each sub-module; S2, determining the number of input and removed sub-modules, and judging the charging and discharging conditions of the input sub-modules in the bridge arm; S3, continuing to use the sorting result of the last period, grouping the sub-modules in the input state and the cut-off state to perform insertion sorting correction for limiting the step number, the step number being determined by the maximum voltage difference, and the larger the difference value is, the larger the step number is; S4, according to the charging and discharging conditions, introducing a retention factor to the capacitor voltage of the corresponding sub-module; S5, merging the two sub-sequences in the input state and the cut-off state in two paths, and updating and recordinga sorting result; S6, generating the trigger pulse. According to the invention, dual optimization of the sorting algorithm operation speed and the device switching frequency is realized, and the problems of large sorting algorithm operation amount and high device switching frequency in the traditional sub-module capacitor voltage-sharing strategy are solved.

The invention provides a high efficiency FIR filter cascade DFT algorithm. Convolution definition is utilized, a calculation formula for R(M+N-1) and I(M+N-1) filtering coefficients of an M-order FIR filter cascade N-point DFT algorithm is derived, after a filtering coefficient after cascade is calculated in an off-line mode, two independent filters are integrated into one filter, calculation efficiency is improved, a problem of large operand caused by filter cascade is effectively solved.

The invention discloses an automatic parking path following processing method and system, a vehicle and a storage medium. The method comprises the steps: 1, planning an arc linear path needed by parking based on parking space information, and obtaining a control point Pi of path planning, i being equal to 0, 1, 2,... N; step 2, performing parametric solution based on a cubic spline interpolation function, and adopting a chord length accumulation method; for the coordinates (xi, yi) of the control point Pi on the plane, solving the accumulated chord length at each control point; step 3, parameterizing x and y into equations of x (s) and y (s) respectively, s being larger than or equal to si and smaller than or equal to si+1, si representing the accumulated chord length of the (i+1)th section, and si+1 representing the accumulated chord length of the (i+2) th section; to be specific, carrying out fitting based on the principle of a cubic spline interpolation function with s as a variableand x and y as functions, and deriving an equation of a chord length accumulation cubic spline curve and first-order derivation and second-order derivation equations of the equation; and step 4, adopting a constraint equation of a fixed boundary for boundary conditions, and solving a first-order derivative of a parameter equation. The problem that the curvature of an arc straight line path is discontinuous is solved.

The embodiment of the invention discloses an image processing method. The method comprises the steps of obtaining a first color parameter value of each pixel of a target image in a target color space;wherein the target color space is different from the RGB color space; obtaining a color mapping table of the target color space; and performing color mapping on the first color parameter value of each pixel based on a color mapping table of the target color space to obtain a second color parameter value of each pixel. The embodiment of the invention further discloses an image processing device, electronic equipment and a storage medium.

The invention provides a matrix operation result determination method and device, a storage medium and an electronic device.The method comprises the steps that a first matrix and a second matrix to be subjected to matrix operation are obtained; splitting the first matrix into at least two first sub-matrixes according to a preset rule, and splitting the second matrix into at least two second sub-matrixes according to the preset rule; performing matrix operation on the at least two first sub-matrixes and the at least two second sub-matrixes to obtain at least two result matrix tables; and summing the at least two result array tables to obtain a matrix operation result of the first matrix and the second matrix. By adopting the technical scheme, the problem that in the prior art, when large matrixes are multiplied, the operation amount of each time of operation is large is solved.

A method and device for evaluating the influence of grid harmonics, wherein the method for evaluating the influence of grid harmonics includes: obtaining at least one random harmonic sample of a harmonic source in the grid; calculating the power grid based on the random harmonic sample Harmonic power flow data; determine the probability density function of the harmonic power flow data; determine the influence of the harmonic source on the power grid according to the probability density function, and solve the problem that the existing harmonic probability evaluation method has a large amount of computation, The problem of low timeliness.