125results about How to "Small time complexity" patented technology

The invention provides a joint measurement method based on a laser radar and a binocular visible light camera to obtain accurate and dense three-dimensional information simply and efficiently. The joint measurement method comprises assuming that the laser radar is a camera device with a fixed internal reference; directly projecting three-dimensional point cloud data into a two-dimensional image, calculating rotation and translation relationships between the laser radar device and the binocular camera by using an image processing method and the matching between the two-dimensional images; creatively introducing an idea for obtaining a matrix norm and a matrix trace to solve a rotation matrix; finally fusing laser radar and binocular stereo vision point cloud data. The method not only can obtain an accurate position and attitude information, but also can reconstruct the special texture and feature information of a target surface, which has a high application value for spacecraft dockingand hostile satellite capture in the military field and workpiece measurement and unmanned driving in the civilian field.

The invention belongs to an infrared dim-small target detection method based on shear wave conversion in the infrared image processing technology. The method comprises the following steps: processing an original infrared image by employing nonsubsampled Laplace pyramid conversion and a Shearlet filter in succession to obtain high frequency information graphs of various directions under different scales, inhibiting background and noise interference information, enhancing target information and extracting a dim-small target. According to the invention, the nonsubsampled Laplace pyramid conversion and a Shearlet filter are employed to process the original infrared image, through same scale and different scales fusion processing of the obtained the high frequency information graphs, the interference information is inhibited, the target information is enhanced, and the high frequency information graphs are subjected to segmentation to obtain a clear dim-small target graph; thereby the method has the characteristics of a short detection processing flow, small data processing amount, short processing time, capability of effectively raising performance of detecting the infrared dim-small target and obviously distinguishing a target and a complex background in the image, a good effect and the like.

The invention relates to an image quality assessment method, in particular to a non-reference image quality assessment method based on information entropy characters, and belongs to the field of image analyzing. The method comprises the first step of carrying out Contourlet conversion on a distorted image to obtain N*M conversion sub-bands, the second step of dividing each conversion sub-band and the unconverted original distorted image, the third step of calculating null domain information entropy and frequency domain information entropy on each block coefficient matrix, and the fourth step of screening the blocking characters and calculating a mean value to obtain the quality character value of each conversion sub-band. The method of a support vector machine and the method of non-reference image quality assessment are utilized for testing on a test set, and quality prediction and assessment are carried out through quality character vectors corresponding to a disaggregated model, an evaluation model and the test set all of which are obtained through training. The non-reference image quality assessment method has the advantages of being high in subjective consistency, small in time complexity and good in university, can be embedded into application systems related to image quality, and has very high application value.

The invention discloses an SAR (Synthetic Aperture Radar) image change detection method based on neighborhood similarity and mask enhancement, relating to the field of remote sensing image processing. The method mainly solves the problem that a difference map construction of SAR image change detection is seriously influenced by SAR image speckle noise and comprises the following steps of: (1) geometrically correcting and registering two SAR images I1 and I2 in the same scene at different times; (2) constructing a difference image map DI of the images I1 and I2 by using a neighborhood similarity operator; (3) performing mask enhancement on the difference image map DI to obtain a new difference image map NDI; and (4) clustering the new difference image map NDI to obtain a change detection result map CDI by using a K mean value clustering method determined by an initial point. The method has the characteristics of high operation efficiency, low algorithm complexity, low time complexity and good experimental effect, and can be used for the change detection of two time-phase SAR images.

The invention provides a facial image layer decomposition method based on an improved guide filter, and the method comprises the steps: carries out the improved guide filtering of a facial image brightness channel, obtaining a human face structure information layer, further carrying out differential operation, and obtaining a texture detail layer. According to the invention, the method carries out the regularization factor adaptive adjustment optimization of a guide filter and Gaussian blurring and corrosion processing of a parameter matrix image, and obtains different filtering effects of a facial image in different regions. The method is used for a digital makeup system, obtains an excellent layer decomposition effect, and remarkably reduces the time complexity of layer decomposition.

The invention provides a non-reference image quality evaluation method based on gradient relevance, and belongs to computer image analysis. The method comprises the following steps: firstly, determining three sub-characters, namely, an image gradient amplitude character, an image gradient direction change character and an image gradient amplitude change character, which are easily influenced by distortion in an image gradient; performing M*M image blocking on the three characters, and determining a statistic variance of the three characters in each image block, wherein the average number of the statistic variances of all the image blocks in the whole image is taken as an image characteristic; and finally, determining the image quality by combining a support vector with a two-step frame in image quality evaluation. The method provided by the invention has the advantages of being small in time complexity, high in subjective consistency, low in image feature dimension and good in universality; the method can be applied to applications related to small computer devices or image quality, and is high in practical value.

The invention relates to a Chinese variation text matching recognition method. The method comprises the following steps of: performing special encoding conversion on a target text and a mode character string to improve the similarity of texts; adding proper wildcard characters into the converted mode character string according to the structural characteristics of Chinese characters in the mode character string; and matching by using a precise character string matching algorithm (namely a CV-BM algorithm). The method solves the problems of the replacement of characters with similar shapes in the variation text and the matching difficulty based on the replacement of wrongly written characters with similar shapes. The problems of the replacement of the characters with the similar shapes in the variation text and the matching difficulty based on the replacement of wrongly written characters with the similar shapes can be solved. Compared with an image partitioning recognition method, the method has smaller time and space complexities, is more suitable for the quick matching of Chinese character information in a high-speed network data transmission environment, and can be widely applied to systems needing to match the Chinese character information, such as an intrusion prevention system, an information retrieval system and the like.

The invention discloses a quick and stable cell nucleus location method in order to solve the problem that the processing speed of the traditional cell nucleus location algorithm can hardly meet the requirement of large-scale digital pathological full-slice computer-aided analysis. First, a color de-convolution algorithm is used to extract the cell nucleus component in a pathological image; then, the center position of the cell nucleus is highlighted by a mixed Gaussian filter; and finally, the local maximum value is found in the filtered cell nucleus image, and the position of the cell nucleus is determined through adaptive threshold judgment. The method has the advantages of clear process, easy implementation, low time complexity, fast execution and stable location effect, and is of great significance in digital pathological full-slice computer-aided analysis.

The invention discloses a power grid instantaneous frequency measuring and tracking method based on an orthogonal sub-band. The method comprises the steps that discrete sampling is carried out on an analog input signal; the signal after discrete sampling is filtered by a constructed orthogonal filter group, and is divided into two sine sub-band signals with the phase difference of 90 degrees; the mathematical relationship among three random and continuous sampling points in the sine sub-band signals is used to respectively calculate the instantaneous angular frequency of each sub-band; linear optimization integration based on a Gaussian noise model is carried out on the measurement results of two sub-bands, so as to acquire an instantaneous angular frequency measurement value with small error variance; Kalman filter estimation is carried out on the measurement value; and the instantaneous frequency of a power grid is tracked. According to the method provided by the invention, measuring and tracking accuracy is not affected by a non-periodic noise signal, power grid voltage / current amplitude variation and a harmonic signal; the method has the advantages of short data window and simple algorithm calculation; and the instantaneous frequency of a power grid fundamental wave can be accurately measured and tracked in real time.

The invention discloses a realization method for a wireless mesh network and nodes. The realization method for the wireless mesh network comprises steps: obtaining nodes and route information needing to be monitored in the network, according to the associated relationship between the nodes and the route, obtaining a monitoring acting candidate set in the network which covers maximal link and has minimum nodes, selecting a node set with larger residing weighted value of the nodes from the acting candidate set as a monitoring agency; adopting the selected agency to monitor the routing transmission of the wireless mesh network, wherein the anticipated transmission time SMETT of the route is measured in the transmission process, thus forming the route measuring result; and according to the result, controlling packet routing or call forwarding. The invention can increase network performance and reduce resource consumption.

The invention relates to the vision identification method of a chip, especially to a solder ball positioning and parameter identification method for a ball grid array pin chip. With the method, problems of poor robustness and low chip solder ball identification precision of the existing method can be solved. An original chip image is processed to obtain a binary solder ball image and connected domain marking is carried out on each binary solder ball; grayscale connected domain extraction is carried out on each binary solder ball after the connected domain marking in a corresponding neighborhood range on the original image and a grayscale information list is established; a background image is established and a local analysis is carried out; boundary straight-line fitting is carried out to solve a deflection angle and a center position of the chip; a linear equation of equivalent solder balls at each row and each column is determined, equivalent solder balls searching is carried out to obtain an equivalent solder ball set of each row and an equivalent solder ball of each column, and then correlated dimensions of the solder balls are solved; and linear fitting is carried out, and all fitting straight line spaces at adjacent rows are used as solder ball standard row intervals and fitting straight line spaces of adjacent columns are used as solder ball standard column intervals. For identification and positioning of a chip with a 500-solder-ball-pin scale, the time is less than 200ms.

The invention discloses an image encryption method which encrypts a clear text image through combination of chaotic mapping and DNA encoding. The image encryption method comprises the following steps:generating an initial key; generating a chaotic sequence and a scrambling vector by use of the initial key; performing scrambling encryption on the clear text image by use of the scrambling vector toobtain a scrambling image; performing DNA encoding on the chaotic sequence to obtain a first DNA sequence, and performing DNA encoding on the scrambling image to obtain a second DNA sequence; performing diffusion encryption on the DNA sequence by use of the first DNA sequence to obtain a third DNA sequence; and performing DNA decoding on the third DNA sequence to obtain a cipher text image.

The invention discloses a vehicle safety control method based on tire condition information and a vehicle. The vehicle safety control method includes the steps that a reference line is determined according to an adhesion slip curve between an adhesion coefficient and the slip rate and the slip slope of tires under different road conditions; the adhesion coefficient is calculated according to the current slip slope of the tires and tire condition information, and the adhesion coefficient and the slip slope are taken as coordinates of data points; a predetermined number of reference points are selected on the reference line, and the similarity between the reference points and the data points is calculated; and maximum adhesion coefficients corresponding to the data points are calculated according to the similarity between the reference points and the data points, and the maximum adhesion coefficients are taken as control parameters to control the vehicle. According to the technical scheme, the maximum adhesion coefficients of the tires in the vehicle are calculated in real time and accurately, thus the maximum effective braking force or effective driving force is obtained, the driving safety of the vehicle is improved, and energy consumption of the vehicle is decreased.

The invention provides a real-time digital surface model generation method, which takes the current frame image taken by a camera carried by an unmanned aerial vehicle and the current frame feature points and point clouds outputted by the SLAM as input data, and generates a DSM in real-time. Firstly, the airborne camera captures the real-time surface image and processes it with real-time SLAM to obtain the feature points of the current frame and the constructed map point cloud; and the feature points of the current frame and the map point cloud are pretreated in advance; Secondly, the DSM texture and regular mesh of the current frame are generated. Finally, DSM texture and regular mesh fusion are carried out respectively. The invention can meet the real-time requirement, has the advantages of fast generation speed, small time complexity, low memory consumption, high robustness and precision, and good splicing effect.

The invention relates to a symmetric encryption method with an infinite key space, which adopts a group of public computer file sequences as a key; the plaintext of a file to be encrypted takes a fixed-length of m Bits as base unit in a physical layer in which the plaintext is stored; the plaintext is divided into a plurality of plaintext fragments in sequence; when the remaining length is less than m Bits, '0s' are added to cause the length to be m Bits; every file of the key divides a ciphertext into a plurality of key fragments in sequence with the same fixed-length of m Bits as the base unit in a physical layer and gives out an address coding of every key fragment in a form of 'file name / address offset'; if the remaining length of the last key fragment is less than m, '0s' are added to cause the length to be m Bits; the plaintext fragment is presented by any one address of the same key fragment. A code representation is carried out on all plaintext fragments in the plaintext by using the same key fragment so as to obtain the ciphertext.

The invention provides a method and device for path planning of mixed airspace heterogeneous aircrafts and belongs to the technical field of aircrafts. A path planning device includes an information acquisition module, a conflict detection module, a collision risk assessment module and a path planning module. The information acquisition module is used to obtain information of all aircrafts, the conflict detection module is used to determine whether a conflict occurs between the two aircrafts and whether the aircrafts pass through a dangerous area, and the collision risk assessment module is used to evaluate the safety of the aircrafts under certain uncertainties. The path planning module optimizes paths of each aircraft through a particle swarm optimization algorithm. The method for path planning includes the steps of first establishing a path planning model for the aircrafts in a mixed airspace; adopting a Monte Carlo method to measure the influence of the uncertainties on air safety,wherein obtained results are closer to practical flight and the planned path is more reasonable; and finally adopting the particle swarm optimization algorithm to optimize the path of each aircraft.The algorithm is small in time complexity, and the path planning is faster.

The invention provides an image color cast correction method based on channel gain coefficient weighting. The image color cast correction method comprises the steps of obtaining digital image data having color cast, calculating the gain coefficients of three channels of red (R), green (G) and blue (B) of the color-cast image and corresponding weights thereof according to formulas, further calculating the weighted new gain coefficients of the three channels of R, G and B, performing point-by-point traversing on the color-cast image by use of the weighted new gain coefficients, and calculating the pixel values of each pixel point in the three channels after correction, thereby realizing the color cast correction process and obtaining the corrected digital image. The image color cast correction method based on channel gain coefficient weighting is applicable to the color cast correction of digital images, has great significance for accurately and efficiently overcoming color distortion of images, and has wide market prospect and application value.

The invention discloses an image watermarking method based on phase singular value correction. The method comprises steps of: firstly embedding a watermark into a low-frequency sub-band transformed by a host image PDTDFB; secondly, extracting the high-frequency sub-band phase matrix of a to-be-detected image subjected to quaternion two-dimensional wavelet transform, subjecting the phase matrix to singular value decomposition (SVD), selecting the first four singular values in the main singular value as the eigenvector of a training sample image, and subjecting the sample to the training learning of geometric transformation parameters such as scaling, translation and rotation so as to obtain an LS-SVR training model; then calculating a mathematical model of the geometric correction of the watermarked to-be-detected image, and determining the model parameters by using the obtained LS-SVR training model to perform geometric correction; and finally, extracting the watermark information from the corrected watermarked image.

The invention discloses an image quality evaluation method based on information entropy. The method is used for solving the problem that an existing image quality evaluation method cannot effectivelyevaluate images of multiple distortion and deblurring distortion types. According to the method, an image library containing deblurring distortion is constructed, and then on the basis of the image library, a support vector machine is used for carrying out classified training according to extracted two-dimensional space entropy and spectrum entropy characteristics of images of the image library toobtain a distortion probability vector model; and then, score fitting is carried out on the characteristics of each type of images and the full reference quality evaluation index VSI of each type ofimages by utilizing support vector regression to obtain a specific distortion quality vector model. By means of the two models, a quality prediction score can be obtained for any image according to the information entropy characteristics of the image, and finally a final evaluation score is obtained by combining distortion detection and evaluation of ringing, residual blurring, noise and the likeof the deblurred image. Compared with the prior art, the method has the advantages of being more objective in evaluation and good in universality. In an image restoration experiment, rapid optimization of a plurality of parameters of the deblurring algorithm is realized by using the method, and the quality of the obtained restored image is obviously improved. The method can be embedded into a DaVinci system or other application systems related to image quality, and has very high practical value.

The invention relates to the field of computer database processing, in particular to a big data real-time enquiry system load balancing method based on copy selection. The method comprises the processes of node load information collection and node load balancing, and the node load balancing process comprises the stages of preprocessing and the copy selection. The method has the advantages that the problems that an existing big data real-time enquiry system load balancing method is too simple and the current state of a machine is not considered are solved, the new big data real-time enquiry system load balancing method based on the copy selection is provided, the load balancing effect is superior to that of the existing big data real-time enquiry system, time complexity is low and is O (n2), and n is the number of blocks; the method is suitable for heterogeneous distributed systems and the conditions of operating other tasks in the systems.

The invention discloses an unstructured road identification method, which comprises the steps of: S1, acquiring a color image in front of a mobile robot in the running process, and preprocessing the acquired color image; S2, performing region segmentation on the image obtained by preprocessing; S3, performing edge detection on the preprocessed image; S4, and performing fusion extraction on the image obtained after edge detection, so as to obtain road regions. The unstructured road identification method is implemented by the steps of: firstly, acquiring the color image in front of the mobile robot, and preprocessing the acquired color image; secondly, subjecting the image to region segmentation, and extracting edge features; and finally, fusing region segmentation and edge feature information to obtain the road regions. According to the unstructured road identification method, an OTSU algorithm is adopted in region segmentation, a multi-directional sobel operator edge detection algorithm is adopted in edge detection, the time complexity is small, the algorithms have good identification effect for a new environment, the robustness thereof is greatly enhanced, and the precision and validity of calculation are improved.

The invention discloses an uncertainty multicast transmission method. The uncertainty multicast transmission method comprises the following steps of: generating a topological graph in a software-defined network, and determining a transmission path among sending ends, receiving ends and forwarding nodes; determining limitation of the number of the sending ends selected by the receiving ends, adjacent edge selection limitation, transmission path selection limitation and multi-wave transmission limitation; and, according to the transmission path among the sending ends, the receiving ends and the forwarding nodes, the transmission cost among any two sending ends, the receiving ends and the forwarding nodes, limitation of the number of the sending ends selected by the receiving ends, adjacent edge selection limitation, transmission path selection limitation and multi-wave transmission limitation, calculating the minimum-cost forest for uncertainty multicast transmission. By means of the uncertainty multicast transmission method disclosed by the invention, uncertainty multicast transmission can work in a software-defined network architecture; and furthermore, the transmission cost and the time complexity are relatively low.

The invention relates to a face recognition method based on a manifold self-adaptive kernel. The method comprises the specific steps that a face image is represented in a vector mode through a representation method based on the appearance; the graph Laplacian is calculated; a non-parameter kernel matrix K is calculated; the graph Laplacian and the non-parameter kernel matrix K are calculated so that a manifold self-adaptive kernel function tightly related to the face data can be obtained; an optimized objective function of the kernel NMF is constructed, a Lagrange function is built, and a multiplicative updating rule is obtained; the kernel matrix calculation process is achieved through the low rank approximate technology of a Nystrom method; the optimized W and the optimized V are calculated, and for a new tested face image Ztest, low dimensional characteristic representation after the dimension reduction of the kernel NMF is obtained; an optimized objective function of an SVM is built; an optimized solution alpha is calculated. The face recognition accuracy of the manifold self-adaptive kernel is better than that of the existing algorithm, dimension reduction serves as the preprocessing algorithm, and then the performance of the face recognition algorithm is enhanced effectively.

The invention discloses a video deblurring method based on a motion vector and a CNN, and the method comprises the steps: S11, carrying out the block processing of all video frames of a to-be-restoredvideo, and generating a plurality of image blocks; S12, performing fuzzy detection on a plurality of image blocks of the video frame to obtain fuzzy image blocks to be restored; S13, searching a clear image block corresponding to an adjacent frame of the to-be-restored blurred image block; S14, inputting the blurred image block to be restored and the clear image block into a CNN deblurring network; S15, the CNN deblurring network generating a deblurring result; and S16, completing the deblurring of the to-be-restored video. According to the method, the effect of efficient video blurring restoration is achieved, the method is different from the mode that deblurring is conducted on the whole video frame, and only the blurred part in the video frame is restored. In addition, the advantages of the CNN in the aspect of computer vision processing are further utilized, and a clear video restoration result is obtained.

The present invention provides a multi-resource partial order scheduling policy. A large number of tasks with a multi-resource demand and a complicated structure exist in NFV, and in actual task scheduling, task scheduling completion time affects user experience; traditional scheduling mostly focuses on stream scheduling study, and is incapable of considering a task structure, however, when a task has a partial order structure, the stream scheduling cannot ensure task scheduling time to be optimized; according to the multi-resource partial order scheduling policy provided by the present invention, in an NFV network architecture, a multi-resource task scheduling problem based on a partial order structure is formalized into a model of the multi-resource task scheduling problem based on the partial order structure; and a policy of simultaneously optimizing average and tail scheduling time is provided, and according to the policy, a task scheduling order is determined in a dominant resource priority (DRP) based mode, and resources are allocated with a maximum utilization allocation (MUA) method; and the DRP policy involved in the present invention has polynomial time complexity, a resource utilization rate close to 100%, and excellent fairness.

The invention discloses a test case set reduction method based on a program slice, and relates to the field of software testing. The test case set reduction method is based on a program slice algorithm and a GRE algorithm and is recorded as a GRE-R-T algorithm. Test requirement reduction and test case set reduction are combined. Firstly, test requirements are subjected to slice calculation, an obtained slice and a program associated with the test requirements are combined to generate a test case set corresponding to statements in the slice, and finally, the test case set is subjected to reduction according to the GRE-R-T method. Compared with a traditional algorithm, the GRE-R-T algorithm is characterized in that the test case set is effectively subjected to the reduction, an amount of detected bugs is not lowered, and test time and workloads are greatly saved.

The invention provides a power flow transfer identification algorithm based on multi-branch removal and deviation paths, which includes the following steps: I, analyzing the composition of a power flow transferred to a non-fault branch, wherein, under the condition of multi-branch removal, the transferred power flow on a branch k can be seen as the superposition of power flow transferred to the branch k after multiple branches with an equivalent power flow as an initial power flow are separately broken; II, based on a deviation path algorithm in graph search, searching for the first K shortestpaths between the two end nodes of removed branches, and determining the regional scope with large power flow transfer; and III, giving a transmission section selection principle by considering the real-time running state of the power grid, identifying a branch set with large increase in actual power flow, and finding out lines of which the power flow increases reversely. Important improvement ismade in power flow transfer estimation, K value selection and section criterion for removed branches.

The invention discloses an object prediction area optimization method applicable to target identification, comprising the following steps: expanding an object prediction area; performing image segmentation; performing super pixel bounding box calculation and super pixel significance evaluation; and finally, getting an optimized object prediction area based on sliding window search of super pixels. By controlling the size of super pixels, the scale of image segmentation is increased, the search scope of super pixels is reduced, and therefore, the time complexity of the algorithm is low and is only related to the number of super pixels in an image; as the pixels in the super pixels are consistent and segmentation of the local edge is better, a better locating effect can be produced by taking super pixels as the basic elements of sliding window search compared with single pixels; the search area is reduced effectively, and the traditional pixel-based sliding window search algorithm is accelerated; and in addition, by cascading the method, the target identification accuracy of the existing target identification algorithm can be improved.

The invention discloses a mobile application program behavior identification method based on spectral clustering and a random forest algorithm, and belongs to the technical field of network security.The method comprises the following steps: acquiring encrypted data traffic of mobile application program equipment; performing feature extraction statistics on the encrypted traffic; performing spectral clustering preprocessing on the encrypted data stream feature data set; and finally, modeling the clustered data cluster data set through a random forest algorithm. According to the method, spectral clustering is used as preprocessing to reduce the opportunity of over-fitting of a classifier, and then a random forest ensemble classifier is used to accurately identify various behaviors of an application program.

The invention discloses a method for hiding reversible information of a security encryption domain based on high-level overturn prediction. The high-level overturn prediction method is used for only testing original images and not changing values and positions of pixels, so that a general image exclusive OR / AND scrambling operation can be applied to encryption of images, and meanwhile, protectionof pixel values and pixel positions can be realized, and the security of the encrypted images is improved; According to the testing method based on the high-level overturn prediction disclosed by theinvention, correlation of the images is fully utilized, so that the steganographic capacity is effectively improved; at a receiving end, a receiver only needs an image encryption secret key and does not need a steganographic secrete key, decryption images identical to the original images and plaintext images which contain additional information and are similar to the original images can be acquired; and on the other hand, the receiver only needs an information steganographic secrete key and does not need the image encryption secrete key, the same additional information can be extracted separately from a plaintext domain containing the additional information and a ciphertext domain. According to the method disclosed by the invention, realization of an algorithm is simple, the security is high, and the steganographic capacity is large.