506 results about "Eigenvalue computation" patented technology

Object detection apparatus includes storage unit storing learned information learned previously with respect to sample image extracted from an input image and including first information and second information, first information indicating at least one combination of given number of feature-area/feature-value groups selected from plurality of feature-area/feature-value groups each including one of feature areas and one of quantized learned-feature quantities, feature areas each having plurality of pixel areas, and quantized learned-feature quantities obtained by quantizing learned-feature quantities corresponding to feature quantities of feature areas in sample image, and second information indicating whether sample image is an object or non-object, feature-value computation unit computing an input feature value of each of feature areas belonging to combination in input image, quantization unit quantizing computed input feature value to obtain quantized input feature value, and determination unit determining whether input image includes object, using quantized input feature value and learned information.

A picture signal encoding apparatus on which, even if a picture is not uniform in picture pattern, the picture can be rendered substantially uniform in picture quality, the volume of bits occupying a virtual buffer is stabilized, deterioration in picture quality is rendered less visible and in which stable rate control is assured. The picture signal encoding apparatus, dividing a picture making up moving picture signals into plural macro-blocks MB and encoding the signals from one macro-block to another, includes a relative coding difficulty calculating circuit for calculating the relative coding difficulty from one macro-block to another MB, and a characteristic value calculating circuit for calculating the weighting coefficients (characteristic values) specifying visibility degree of picture quality deterioration from one macro-block MB to another. The picture signal encoding apparatus also includes a bit rate controller for calculating the quantization scale for encoding each macro-block MB using the complexity degree and the volume of allocated bits for macro-block-based encoding.

The invention discloses a filtering method of recursive Chinese junk E-mail, which is based on Logistic. The method comprises the following steps: first, analyzing E-mails, extracting E-mail titles, E-mail main bodies and accessory relative information, second, segmenting words for version information which is extracted, third, accounting word frequencies of entries in E-mails, calculating weights of words through utilizing TF-IDF pattern, presenting the E-mail to be characteristic vector which is weighted, fourth, utilizing an LIBLINEAR tool kit to exercise the sample of the E-mail to get an Logistic recursive module, fifth, utilizing the Logistic recursive module to classify for new E-mails, getting the probability value whether the E-mails which are got are junk E-mails. The utility which utilizes the Logistic recursive module has the advantages of simple module, little amount of parameter, and high classifying accuracy in a data set whose text number and characteristic number are both bigger, the accuracy and efficiency of filtering junk E-mails are improved through dimension reduction and improved characteristic value calculating method, and meanwhile, the problem of choosing module exercise parameter which is faced in filtering junk E-mails is effectively solved.

The invention discloses a method for calculating time-lag electric power system eigenvalues and discriminating stability based on Pade approximation. The method uses Pade approximation to make wide area feedback time-lag approximate to a rational polynomial. Though connecting with an electric power system without time-lag and a wide area damping controller, a linearized model of the time-lag electric power system is established. Finally, part of characteristic roots of the time-lag system is directly obtained according to a system state matrix, and then time-lag stability of the system is determined. According to an eigenvalue calculating result of a four-machine two-region example system, the method can relatively accurately solve part of the eigenvalues and eigenvectors which are corresponding to dynamic elements in the time-lag system. The method can conveniently and accurately determine time-lag stability of the system, correctly solve the number of the eigenvalues which are corresponding to a time-lag link and calculate precision which is related to order of the rational polynomial. In addition, the method is advantaged by small calculated amount, and short calculating time.

The invention discloses a witness and evidence integration recognition method and system based on a deep convolutional neural network. By constructing the human face convolution neutral network model, a most difficult distinguishing triple and iterative optimization method is used for training the human face convolutional neural network model, and the deep convolutional neural network model is obtained, the model is used for calculating certificate image and human face image feature values, an Euclidean distance is calculated according to the feature values, and then whether the certificate image is consistent with the human image or not is judged. Compared with the prior art, the method and system has the good robustness for background, illumination, gesture and other changes, constant iteration training can be performed by means of depth learning, the human face recognition performance and the characteristic extraction accuracy are improved, and the system performance is effectively enhanced.

A pattern shape predicting method comprising: predicting, with simulation, an intensity distribution of a pattern image concerning a pattern shape of a pattern on substrate formed on a substrate based on pattern data; calculating a first pattern edge position from the intensity distribution of the pattern image; calculating a feature value of the intensity distribution of the pattern image in a predetermined range including the first pattern edge position; calculating a fluctuation amount of the first pattern edge position from the feature value using a correlation; and predicting a second pattern edge position taking into account the fluctuation amount with respect to the first pattern edge position.

A learning apparatus includes an image generator, a feature point extractor, a feature value calculator, and a classifier generator. The image generator generates, from an input image, images having differing scale coefficients. The feature point extractor extracts feature points from each image generated by the image generator. The feature value calculator calculates feature values for the feature points by filtering the feature points using a predetermined filter. The classifier generator generates one or more classifiers for detecting a predetermined target object from an image by means of statistical learning using the feature values.

An operation part (50) in a pattern inspection apparatus (1) comprises a defect candidate image generator (511) for generating a binary defect candidate image representing a defect candidate area in an inspection image by comparing the inspection image with a reference image, in an inspection image masking part (521) the inspection image is masked with the defect candidate image to obtain a masked inspection image. In a feature value calculation part (531), an autocorrelation feature value is obtained from the masked inspection image, and outputted to a classifying part (54). The classifying part (54) comprises a classifier (541) outputting a classification result on the basis of the autocorrelation feature value and a classifier construction part (542) for constructing the classifier (541) by learning. It is thereby possible to easily perform the high accurate classification of defect candidate using the autocorrelation feature value which is hard to characterize as compared with geometric feature value or feature value representing density.

A facial feature analysis system for command applications for users with physical disabilities, like impaired users of wheelchairs, patient beds or other appliances is provided. The disclosed system includes a virtual filter bank and a virtual discriminator. The virtual filter bank comprises a feature localization main module having an ancillary data bank which supplies a special signal Sg for control functions. The system calculates face localization based on parameters of a holistic face-model, calculates feature localization based on parameters of an adaptive face graph, calculates feature extraction using stored feature values corresponding to selected validation and provides output signals using a signal delivery main module controlled by static and dynamic classification. The virtual discriminator bank calculates a user adapted allocation based on a face feature of the user and provides a periphery allocation for at least one command modus. and for calculating execution signals for said manipulators.

The invention discloses a meta-path-based link prediction method for aligned heterogeneous social networks, mainly solves the problem of link prediction (mainly including friend relationship prediction and position prediction) in the aligned heterogeneous social networks, and relates to related technologies of the aligned heterogeneous social networks, feature selection, link prediction and the like. The method mainly comprises six steps of (a) modeling the aligned heterogeneous social networks; (b) automatically extracting a meta-path; (c) defining a meta-path-based eigenvalue calculation method; (d) dividing data sets according to timestamps; (e) proposing a two-stage gradual forward greedy feature selection algorithm for performing feature selection; and (f) training a decision tree classifier based on a feature selection result to perform the link prediction. Based on the method, the aligned heterogeneous social networks consisting of Foursquare and Twitter are subjected to the link prediction; and the method has practical application values for entity recommendation, accurate marketing, criminal gang discovery and the like in the social networks.

Embodiments of the present application relate to a method, device, and system for allocating resources in a server. The method includes obtaining first resource usage information associated with the first host computing system and second resource usage information associated with the second host computing system, computing a first characteristic value and a second characteristic value, wherein the first characteristic value is computed based at least in part on the first resource usage information, the second characteristic value is computed based at least in part on the second resource usage information, obtaining a first comparison result based on comparing the first characteristic value to a resource usage threshold value of the first host, and a second comparison result based on comparing the second characteristic value to a resource usage threshold value of the second host computing system, and adjusting resource allocations for the first host computing system or the second host computing system.

The movie image retrieving apparatus includes an image input device 13 to which the movie images are inputted in a time-series manner, a feature value calculation device 14 which includes a feature value deriving section 16 for deriving the feature value and a quantization section 17 which quantizes the feature value with a predetermined quantization width to produce the feature value information, a comparative information selection device 15 for deriving the comparative feature value information from the data-base, and a matching device 18 for matching the feature value information and the comparative feature value information using a quantization error. The matching result is outputted from the output device 19. Load on the hardware is reduced and the time required for the search is shortened.

An image processing apparatus includes a feature value calculating section for setting a certain pixel in a processing target image as a pixel of interest, and for calculating a curvature of a curved surface which is obtained from density distribution of adjacent pixels which are located within a predetermined range from the pixel of interest, as a feature value of an image area within the predetermined range from the pixel of interest.

In an image output device a test chart 18 including tests charts 6 of each color is used, and a tone correction table is calculated. The test charts 6 are patterned in a checkered pattern, each including continuous areas 10 in which the tones increase in steps and reference areas 11 for comparison with the continuous areas 10. The reference areas 11 are formed of white ground of paper in a highlight proof part 7, a solid pattern with a maximum density in a shadow proof part 8, and halftone dot-concentrated or line screens with a lower resolution than in the continuous areas in a middle proof part 9. Tone correction values are calculated from characteristic values of the highlight, shadow and middle parts, obtained from visual data of the test chart 18.

The invention provides a feature value analysis system for small-interference stability of a large-scale power system, and belongs to the technical field of power system simulation and analysis. The feature value analysis system comprises a BPA data interface module for achieving seamless read-write operation of a BPA data file, a sparse matrix calculation module for achieving relevant sparse matrix processing in small-interference stability feature value analysis, a power flow calculation module, a system modeling and linearization module, a state matrix calculation module, an all feature value calculation engine module, a partial feature value calculation engine module and an oscillation mode extraction and analysis module. Omni-bearing and missing-free feature value simulation analysis for small-interference stability of an actual large-scale power grid is achieved by the adoption of TTQRE and IRAM and JDM with multiple calculation schemes.

The invention discloses a self-adaption multi-pose face recognition method based on hidden least square regression. The self-adaption multi-pose face recognition method includes the multi-pose face recognition method based on the hidden least square regression. The method includes the steps of detecting a region size and a region position of an input facial image; correcting the detected facial image, obtaining a corrected facial image; extracting facial characteristic values from the corrected facial image; estimating a pose type of the corrected facial image according to the extracted facial characteristic values; selecting a corresponding transformational matrix of a pose type and a corresponding offset vector of the pose type according to the pose type, and calculating to obtain an identity characteristic vector of the facial image according to the transformational matrix, the offset vector and the extracted identity characteristic vector; and searching for a known facial image which has the highest similarity with the identity characteristic vector of the input facial image in a known facial image search library, and returning identity information of the known facial image to be used as a recognition result.

The invention provides a detection method for identifying the surface quality of a continuous laser seam of a metal workpiece online, wherein the method comprises the following steps: using a high resolution color area array camera, a high magnification micro lens and double LED illuminating systems to acquire an image of the surface of the seam; afterwards, when dividing a seam region and calculating the characteristic value, translating or rotating the workpiece to shoot an image for other positions of the seam, continuously shooting the image of the seam and completing real-time calculation until reaching a seam welding end or an overlapping point; finally integrating the calculation results of the images, and completing online judgement on the size, the position and the surface defect of the continuous seam. The method is used to complete online detection of the quality of a laser straight welded joint, a curve welding seam or a space welding seam of various ferrous metals and nonferrous metals, and belongs to the category of non-contact visual inspection; a device is simple and compact, high in response speed and strong in anti-jamming capability, has the characteristics of stability, reliability and convenience, and can be widely applied to the field of laser welding of metal materials.

Provided is an image analyzing apparatus for efficiently performing detection of an object and tracking of a specified object, including a feature value recording section that records a plurality of reference feature values different in type from each other; a feature value extracting section that extracts a plurality of feature values different in type from each other, from each of a plurality of moving image constituent images included in a moving image; an object extracting section that extracts an object from the moving image constituent images, based on a degree of matching of the plurality of extracted feature values with respect to the plurality of reference feature values recorded in the feature value recording section; a reference feature value calculating section that calculates, from the plurality of reference feature values recorded in the feature value recording section, a plurality of reference feature values adjusted to the feature values of the extracted object, to a predetermined degree corresponding to the type; and a feature value updating section that updates the plurality of reference feature values recorded in the feature value recording section, with the plurality of reference feature values calculated by the reference feature value calculating section.

The conventional ESPRIT method is accompanied by the problem of very long signal processing time. The radar device of the invention includes a signal vector-forming unit for forming signal vectors based on waves reflected from an object and received by using a plurality of receiving antennas; a submatrix-forming unit for forming submatrices based on the signal vectors; a regular matrix operation unit for calculating a regular matrix from the submatrices; an eigenvalue decomposition unit for calculating an eigenvalue of the regular matrix; and an angle calculation unit for calculating an angle at where the object is present from the eigenvalue.

The invention relates to the technical field of radars, and aims to solve the problems existing in the prior art. The invention provides an airport runway foreign matter detection method and a device based on the characteristics of a characteristic spectrum. According to the invention, the FOD detection at a low false alarm probability is realized. Runway benchmark background data are adopted as clutter map reference data, and the clutter map constant false-alarm rate (CFAR) treatment is carried out on runway radar data. After the clutter map constant false-alarm rate (CFAR) treatment of the runway radar data, obtained data are classified and divided into a background clutter signal and an FOD echo including a false alarm signal. The characteristic values of the background clutter signal and the FOD echo signal are respectively calculated, and then corresponding characteristics are extracted to form corresponding characteristic vectors according to corresponding characteristic values. A characteristic vector corresponding to the background clutter signal and the label of the background clutter signal are trained by a classifier, and then the parameters of the classifier are obtained. Whether an FOD exists in the FOD echo or not can be judged according to a characteristic vector corresponding to the characteristic value of the FOD echo signal and the parameters of the classifier. Therefore, the FOD detection of a runway is realized.