86results about How to "The overall calculation is small" patented technology

The invention relates to a mobile phone casing defect detecting method based on depth learning. The method comprises the steps that (1) the image of a mobile phone casing to be detected is acquired and pre-processed; and (2) the pre-processed image is input into a pre-trained defect detection model for defect detecting to acquire the position of a defect on the mobile phone casing, and the confidence of the position as the defect is provided. The defect detection model is a depth network based on depth learning, and comprises a feature extraction network and a classifier and regression device network, wherein the feature extraction network and the classifier and regression device network are in successive cascade. The feature extraction network carries out feature extraction on the pre-processed image to acquire a feature image. The classifier and regression device network classifies and regresses the feature image to acquire the defect position and the confidence of the mobile phone casing. Compared with the prior art, the method provided by the invention has the advantages of high detection precision and accurate and reliable detection result.

The invention relates to a mobile phone information based section speed calculation method. The method is used for calculating the average speed information of a vehicle at each section through the analysis and the treatment of mobile phone information and the combination of traffic road position and direction. The key points of calculating the section average speed lies in massive data processing and precise positioning of a mobile phone user in a movement process. The problems are solved through two-stage three-time map matching in the invention. The map matching at the first stage comprises the following step of: matching Cell with sections, and the map matching at the second stage comprises the following steps of: establishing a subordination relationship of mobile phone data points and the sections through matching user data points with the sections two times, finally calculating the traffic speed of each user on the sections by using the Cell position in which continuous data points are located and the time, removing part of abnormal speeds in the traffic speeds, summing the traffic speeds and averaging the sum to obtain the average speed of each road. In the invention, the large amount of calculation caused by directly matching each data point with the sections is avoided, and the calculation efficiency is improved. The accuracy of the mobile phone data and the sections is improved through a proximity principle and a direction consistency principle.

The invention discloses a current based open-circuit fault detection method for a three-phase inverter. The method comprises the steps of acquiring motor three-phase current in through a motor driving control system; performing weight processing for the cycle average value of the three-phase current and the cycle average value of an absolute value; normalizing to obtain a detection variable dn; comparing the detection variable dn with the preset threshold ki to obtain a fault indicator Dn of the three-phase inverter; positioning an inverter switching tube suffering from open-circuit fault from a fault positioning table according to the fault indicator Dn. The method is simple and easy to be carried out, and is suitably integrated to the control system.

An image processing device includes an image segmentation unit, a stitching unit and a rendering unit. The image segmentation unit divides a plurality of input images into a plurality of first images and a plurality of second images based on color information and depth information. The plurality of input images are sequentially captured. The stitching unit generates a first panoramic image by synthesizing the plurality of first images, and generates a second panoramic image by synthesizing the plurality of second images. The rendering unit generates an output panoramic image based on the first panoramic image and the second panoramic image.

The present invention provides a method of secure communication for a wireless mobile terminal. In the invention, the wireless mobile communication user realizes secure information exchange of terminal to terminal based on the guarantee reliability identity authentication without bearing deployment, maintenance and complex operation of public key infrastructure, without the participation of telecom operator, without replacing or impacting telecom signaling exchange protocol. The mobile communication user uses mobile terminal identity mark as identity, introducing a credible thirst party, named for key generation center, the main function is to help user to generate key according to the identity of user. The user acquires a key corresponding to the identity from the credible third party by mobile phone message or data service. The key guarantees the encryption key exchange and safety parameter between users to negotiate and resist active attack, session key between mobile users is independent to the KGC random generation, realizing the safety communication of terminal to terminal.

The invention discloses a synthetic steering vector based beam forming meter-wave radar height measurement method, and mainly aims to solve the problems of large signal processing operation and computational complexity of the prior art during height measurement. The method includes extracting a target signal from a radar echo and determining a target elevation angle searching range; building a radar wave reflection model and computing ground arc length between a reflection point and a radar center and ground arc length between the reflection point and a target by a searched elevation angle; computing equivalent heights of the target and the radar center and then computing a reflection angle and wave-path difference between a reflection wave and a direct wave; computing a synthetic steering vector; and after data beam forming of an echo, selecting an angle corresponding to a maximum normalization power value as a target elevation angle and computing the target height. Beam forming and the synthetic steering vector are combined, the target reflection model is simplified, elevation angle estimation precision of a meter-wave radar is guaranteed, computation speed of meter-wave radar height measurement is increased, a computational process for height measurement is simplified, and real-time signal processing of the meter-wave radar is easier.

Embodiments of the present invention provide an image splicing method, a computer device, and a display device, and relate to the technical field of display, which is used to solve the problem that the image splicing is performed by using a differential algorithm, resulting in a slow splicing speed. The image splicing method includes steps of: acquiring a first depth image of a current frame by using a first camera; acquiring a second depth image of the current frame by using a second camera; acquiring a depth average of the first depth image or the second depth image, and acquiring a point cloud space coordinate transformation matrix matched with the depth average; converting the first depth image into a first camera coordinate system, and performing imaging in the first camera coordinatesystem to form a first three-dimensional point cloud image; mapping the first three-dimensional point cloud image to in the second camera coordinate system to form a second three-dimensional point cloud image; inverting the second three-dimensional point cloud image into a third depth image located in the second camera coordinate system; and splicing the second depth image with the third depth image.

The invention discloses a method and device for generating a video interpolation frame, relating to video processing. The method comprises the following steps: detecting whether the brightness of the current frame is changed abruptly relative to the previous frame by calculating the difference of the average brightness value of the current frame and the average brightness value of the previous frame image; if so, selecting to copy the current frame or the previous frame as an interpolation output frame, otherwise, searching the motion vector of the interpolation frame corresponding to each subblock through bilateral movement of block matching; and then filtering the motion vector, and generating a movement compensation interpolation frame serving as the interpolation output frame according to the motion vector. According to the invention, large arithmetic amount caused by a great deal of errors during the movement searching of the two adjacent frames with relatively larger brightness difference is avoided, and a false searching result is avoided, therefore, the video with suddenly changed brightness can be effectively processed.

The invention relates to a protection method suitable for multimedia data files in a portable system, comprising the following steps of: performing safety processing on multimedia files at non-portable computer system terminals (for example, computer devices with high performance, including desktop computers, servers, working stations, high-performance embedded computers and the like); downloading the processed result in a portable system terminal; and reducing the multimedia files by the portable system terminal. The protection method has the beneficial effects that: a plurality of types and formats of multimedia files in a portable system can be efficiently protected, the safety degree is adjustable, the computation workload is low, and the cracking difficulty is high.

A power transformer winding deformation reverse acquisition method based on topological optimization includes the steps of building an equivalent circuit model of a transformer winding and acquiring afrequency response function of the equivalent circuit model of the transformer winding, calculating flexibility of the frequency response function of the equivalent circuit model to element parameters, correcting parameter of a simulated equivalent circuit model according to the frequency response function obtained through test, and performing reverse acquisition on deformation status of the transformer winding by a topological optimization method based on a flexible deformation unit. High calculated amount generated in the process of directly utilizing electromagnetic field finite elements to calculate the frequency response function is avoided and correction efficiency is improved; besides, the topological optimization method is introduced to perform visual reduction on the deformed winding, and thus the winding deformation acquisition precision is improved.

The invention discloses a data processing method and device, an electronic device and a computer storage medium. Where data storage structures are connected with hash pointers, Each time the stored data is recursively reflected to the root node of the prefix tree structure, take advantage of that method to store data, which provides convenience for subsequent data reading and data checking, The invention can quickly find the data value of the stored data, and can avoid the huge calculation amount caused by the Meckel tree of the entire database data which must be recalculated in the data checking process after the data changes, and only the hash value of the root node in the final prefix tree structure is needed to be compared during the checking.

The invention provides a social network partitioning method and a system based on cloud computing, and the method comprises the steps: working out maximum groups in a social network; merging the maximum groups according to the proportion of common points among the maximum groups in corresponding maximum groups; and after merging the maximum groups and obtaining a community as a new point, connecting adjacent points by edges, thus obtaining a new social network. The social network is partitioned according to the maximum groups, the new social network is obtained by taking the merged maximum groups as the new point, and the new social network can reflect analysis needs accurately and improve the utilization ratio of the social network; and simultaneously, the maximum groups are worked out in the manner of the cloud computing, which can improve the social network partitioning efficiency.

The invention relates to a method and a system for screening an N-2 fault. An active power limit value for each branch in the power grid, active power variations before and after breaking of each remaining branch for a target branch and an active power value before breaking of each remaining branch for the target branch are acquired firstly; according to the above three values, an overload contribution rate for the target branch by breaking of each remaining branch is acquired respectively; the remaining branches are grouped according to the overload contribution rates, two branches whose overload contribution rate sum is no smaller than an overload contribution rate standard value are screened according to the grouped condition, and an N-2 fault is obtained. By adopting the scheme of the invention, each possible combination of two branches does not need to be screened, grouped screening is carried out according to the grouped condition, the huge amount of computation which is difficult to calculate caused by combination explosion can be avoided, and the computation amount for power grid risk analysis is greatly reduced.

The invention relates to a multi-objective optimization method based on fusion of Dynaform and an intelligent algorithm, and belongs to the technical field of stamping forming process optimization, and the method comprises the following steps: establishing a computer aided design model, performing stamping forming numerical simulation by utilizing finite element analysis software Dynaform, and mapping a stamping forming physical process; adopting a Taguchi orthogonal test method to carry out stamping forming simulation test arrangement, carrying out variance analysis on test results, and comprehensively evaluating the influence degree and the influence rule of the stamping speed, the friction coefficient, the blank holder force, the plate thickness and the die gap on the stamping forming quality in the stamping process; training a radial basis function (RBF) neural network by utilizing simulation experiment data, obtaining a Pareto optimal solution set by combining a non-dominated sorting genetic algorithm (NSGAII), screening out an optimal process through evaluation of a TOPSIS (Tracking Optimal Solution Sorting Method), and verifying the effectiveness of the method. According tothe method, the defects in the prior art are overcome, the die testing cost is reduced, cost is reduced, efficiency is improved, and a theoretical basis is provided for online regulation and control of stamping forming process parameters.

A sequence-pair creating apparatus includes a block placement storing unit that stores information of size of a block b_i in a block set B and information of block placement, creates a sequence-pair (P, M), serving as a pair of a sequence P and a sequence M of the block b_i, and further includes a binary relation setting unit that sets, in accordance with the information of block placement and information of size, a binary relation serving as an order relation that indicates a relative configuration between the blocks of a block pair of two blocks and that is derived from a configuration constraint between the blocks extracted from the information of block placement and information of size or designated by an external input, and a total order relation setting unit that sets a series of ranks of the sequences P and M for all the blocks on the basis of the information of block placement and information of size so as to satisfy all binary relations set by the binary relation setting unit.

The invention discloses a color image completion method based on tensor block cyclic expansion, and belongs to the technical field of image processing. The method comprises the steps of firstly, inputting a to-be-completed image, carrying out n-nearest neighbor initialization on missing pixels, and obtaining a target image; then initializing model parameters, estimating a block cyclic expansion rank of the target image and setting a weight coefficient; and then inputtin the target image into an image completion model in a tensor form and carrying out convex optimization solving on the model through iteration by adopting an alternating direction multiplier method, wherein the image completion model is a low-rank matrix factorization model based on tensor block cyclic expansion. and finally,carrying out data format conversion on the tensor obtained by iteration, so that the tensor is output in the format of the image to be completed. According to the method, when tensor block cyclic expansion is carried out, the connection between image slices is increased, so that the loss of image structure information caused by expansion operation is reduced to a certain extent; the peak signal-to-noise ratio of the complemented image is significantly improved, and texture and detail information are richer.

Quantizing means generates a quantized density and a quantization error component for a target pixel in a plurality of pixels. A subtracter calculates, for the target pixel, a quantization error correction component by subtracting the density of the target pixel contained in preliminary ejection data from the quantization error component output from the quantizing means. An allocator allocates the quantization error correction component calculated by the subtracter to a plurality of the pixels around the target pixel. The quantizing means outputs the quantized density and the quantization error component of the target pixel by performing an addition using the density of the target pixel contained in the image data, the density of the target pixel contained in the preliminary ejection data, and the accumulated diffused error component assigned to the target pixel by the allocator when other pixels are quantized.

The invention relates to a nonlinear dynamic time-history analysis method for a complex building structure, which comprises the following steps of: 1, performing spatial finite element discretizationon the complex building structure, and establishing a motion equation set of a discrete system; 2, selecting parameters and determining global invariants; 3, calculating step by step, and calculatingthe displacement, speed and acceleration of each time step at the end moment. Two key improvements are made on the nonlinear iteration process of the Newmark method, so that the calculation workload is greatly reduced and the calculation efficiency is improved on the premise of ensuring the calculation precision.

The present invention discloses a method for measuring an actual temperature of a flame by using all information of a radiation spectrum and a measurement system thereof. The method includes: conducting more theoretical data processing by using energy level structure correction, wherein all information of the radiation spectrum can be used; and by way of a keyboard input manner or a data transmission input manner, acquiring an energy level structure correction parameter, and finally acquiring a more accurate actual temperature value of a measured flame. The method effectively overcomes a defect that the true temperature of the flame can be obtained by only conducting radiance correction through data processing with great calculations when adpted multi-spectral temperature measurement method. In the existing multi-spectral temperature measurement method at present, only information of several monochromatic radiation capacities in the radiation spectrum can be used; and in the method, information of all the monochromatic radiation capacities, thousands of monochromatic radiation capacities in general, in the radiation spectrum can be used.

According to the OCT image reconstruction method based on the non-uniform discrete Fourier transform, the discrete Fourier transform method of non-uniform time sampling points and non-uniform frequency sampling points is used, so that the loss of high-frequency signals in the discrete Fourier transform process is reduced, and the phenomenon of sensitivity decline at a relatively large depth is improved; conversion and interpolation of a lambda space and a k space in a traditional reconstruction method are not needed, and the operation process and the calculation amount of reconstruction depth information are simplified; frequency domain expansion can be realized by resampling the frequency domain, so that huge calculation amount caused by zero padding of previous and later data is avoided; a non-uniform time sampling point method is selected to eliminate the influence of the linear array CCD on the quality of a reconstructed image when the linear array CCD carries out nonlinear sampling on an original signal; and half of the frequency domain space is selected, and the interference of the complex conjugate image on the imaging quality is eliminated similar to Hilbert transform.