291 results about "Filter model" patented technology

The invention discloses a vision-inertia integrated SLAM (Simultaneous Localization and Mapping) method based on a genetic algorithm. The method comprises the following steps: integrating a vision navigation coordinate system with an inertia navigation coordinate system, and calibrating parameters of a binocular camera so as to solve a three-dimensional space coordinate according to an image pixel coordinate; independently calculating by inertia navigation; calculating by vision navigation; integrating vision navigation information with inertia navigation information by using an extended Kalman filter, and building a system filter model; and observing global feature point road signs by using the binocular camera by taking localization locality into account, carrying out data association on map features based on the genetic algorithm, and feeding extended state vectors back to a filter. The method is capable of carrying out long-time and high-accuracy localization; the genetic algorithm is added for improving the data association of the map, so that the simultaneous mapping accuracy is greatly improved.

A device for generating a binaural signal based on a multi-channel signal representing a plurality of channels and intended for reproduction by a speaker configuration having a virtual sound source position associated to each channel, is described. It includes a correlation reducer for differently processing, and thereby reducing a correlation between, at least one of a left and a right channel of the plurality of channels, a front and a rear channel of the plurality of channels, and a center and a non-center channel of the plurality of channels, in order to obtain an inter-similarity reduced set of channels; a plurality of directional filters, a first mixer for mixing outputs of the directional filters modeling the acoustic transmission to the first ear canal of the listener, and a second mixer for mixing outputs of the directional filters modeling the acoustic transmission to the second ear canal of the listener. According to another aspect, a center level reduction for forming the downmix for a room processor is performed. According to even another aspect, an inter-similarity decreasing set of head-related transfer functions is formed.

The invention relates to mobile tracking of vehicle targets in the communication and transportation field. A self-adapting interactive multiple model mobile target tracking method comprises the following steps of establishing mixed initialized input. including a covariance matrix of a mixed initial condition and a mixed initial state. of each model; establishing constant velocity (CV) and constant acceleration (CA) motion models; updating: calculating covariance matrix and innovation of an error according to a Kalman filter formulation; constructing a likelihood function of a target motion module by utilizing the innovation of a Kalman filter result, and calculating a Markov state transition probability matrix; carrying out estimation output after being fused by utilizing the Markov state transition probability matrix as weight of switch among each motion model. According to the self-adapting interacting multiple model mobile target tracking method disclosed by the invention, the problem that the error is increased or the tracking is failed caused by non-matching of a filter model and a target motion model due to the motion of a target in the traditional interactive multiple model algorithm is solved; the self-adapting interacting multiple model mobile target tracking method has the advantages of low calculation complexity and good tracking effect, and can be applied to target tracking of motor vehicles in the communication and transportation field.

Methods and apparatus provide for positioning of a rover antenna from GNSS data derived from multi-frequency signals and correction data derived from a network of reference stations. Rover antenna position and multi-frequency ambiguities are estimated at each epoch. An ionospheric filter models variation in ionospheric bias per satellite. A set of ionospheric carrier-phase ambiguities is estimated at least when the multi-frequency ambiguities have attained a predetermined precision. The estimated ionospheric carrier-phase ambiguities are cached. After detecting interruption of signal at the rover antenna and determining reacquisition of signals at the rover antenna, an ionospheric bias per satellite over an interruption interval is predicted. For each satellite, a cached ionospheric carrier-phase ambiguity is combined with a predicted ionospheric bias to obtain a post-interruption ionospheric ambiguity estimate. The post-interruption ionospheric ambiguity estimates are used to aid estimation of rover antenna position after signal reacquisition.

The invention relates to a distance correction weighted centroid localization method based on reception signal intensity optimization and aims to realize precise localization through moving an unknown node in a complex indoor environment. The method comprises steps that RSSI signal processing optimization strategies are comprised, that is, a Gauss mean value mixing filtering model is utilized to optimize an RSSI value, a problem of severe jittering existing in the RSSI value is solved, and a more reliable and reasonable RSSI value is acquired; timely beacon node combination is carried out, that is, the smallest four nodes are selected by utilizing a bubble sorting method from small to large to carry out combination localization; whether two circles in the combination intersect is determined, if not, a distance correction scheme is utilized to carry out optimization, and fault tolerance capability, adaptability and localization precision of the localization algorithm are improved; a four-side weighted centroid localization algorithm is utilized to carry out timely localization, and the position information of the unknown node is acquired. Through the method, localization precision is improved through the optimized localization method, stronger adaptability and higher fault tolerance capability are realized, and the method is suitable for realizing application and popularization in the complex indoor environment.

A computer-implemented system and method are described for detecting obfuscated words in email messages and using this information to determine whether each email message is spam or valid email (ham). For example, a method according to one embodiment of the invention comprises: providing an obfuscation feature set for detecting obfuscation within email messages, the obfuscation feature set build from a group of obfuscation parameters including a similarity metric, the similarity metric using a set using a set of frequently obfuscated words (FOW) selected from a larger set of obfuscated words; analyzing an email message to detect whether the email message contains features within the obfuscation feature set, wherein the analysis includes determining the similarity of one or more words in the email message with each of the FOWs; generating the similarity metric based on the analysis, the similarity metric providing a relative likelihood that each of the one or more words is obfuscated; firing one or more of the obfuscation detection features based, at least in part, on the value of the similarity metric; analyzing the email message to detect whether the email contains one or more additional spam features unrelated to obfuscation; and determining whether the email message is spam based on the combined obfuscation detection features and the additional spam features.

The invention relates to image process, motion analysis and the like. In order to implement cross-camera tracking of moving targets, the invention uses the technical scheme that the cross-camera tracking method for the multiple moving targets comprises the following steps of: 1, motion filtering for a single camera, specifically, decomposing complex motions of human body into a finite number of combinations of relatively simple motion filtering models, wherein each single motion filtering model is represented by a combination of a linear motion state transition model and a gaussian noise, and the motion filtering method comprises the two following core parts: a motion model: an observation model; zk = Hsk + wk2, cross-camera oriented motion association: (1) creation of a panorama; (2) target similarity measurement; (3) motion trail association; and transforming the problem of cross-camera association of multiple motion trails into the problem of matching of bipartite graphs, and solving the problem through the integer programming under the following constraint conditions. The cross-camera tracking method for the multiple moving targets is mainly used in the image processing, the motion analysis and the like.

The invention discloses a positioning method using laser radar in an indoor unknown environment. Based on traditional laser radar positioning, map construction in the indoor unknown environment is finished, feature matching of line segments and angular points is carried out based on the constructed map, a mixed feature matching filtering model is provided, and real-time accurate positioning in the indoor unknown environment is achieved. The positioning method using the laser radar in the indoor unknown environment is simple, is easy to implement, avoids adverse effects of the indoor environment on the positioning process, and is suitable for engineering application.

A method of updating an all-attitude angle of an agricultural machine based on a nine-axis MEMS sensor includes the following steps: establishing an error model of a gyroscope, an electronic compass calibration ellipse model and a seven-dimensional EKF filtering model, and setting a parameter vector corresponding to a vehicle motional attitude (S1); acquiring data including an acceleration and an angular velocity of a motion of vehicle, and an geomagnetic field intensity in real time (S2); calculating an angle, a velocity, position information, and a course angle of the vehicle by established error model of the gyroscope and the electronic compass calibration ellipse model(S3); data-fusion processing the angle, the velocity, the position information and the course angle of the vehicle by the seven-dimensional EKF filtering model, and updating a motional attitude angle of the vehicle in real time. The steps of the method have a small error, high precision, and reliability.

The invention provides a target tracking method based on multi-characteristic adaptive fusion and kernelized correlation filtering technology. The method comprises steps of according to target position and the dimension of the previous frame tracking, acquiring a candidate region of target motion; extracting histogram characteristics and color characteristics in the gradient direction of the candidate region, fusing the two kinds of characteristics, carrying out Fourier transform so as to obtain a characteristic spectrum and then calculating kernelized correlation; determining the position and the dimension of the target at the current frame, and acquiring a target region; extracting histogram characteristics and color characteristics in the gradient direction of the target region, fusing the two kinds of characteristics, carrying out Fourier transform so as to obtain a characteristic spectrum and then calculating kernelized self-correlation; designing the adaptive target correlation and training a position filter model and a dimension filter model; and using a linear interpolation method to update the characteristic spectrums and the related filters. According to the invention, the discrimination capability of the models is improved; robustness of the target tracking of the target in a complex scene and the appearance change is improved; calculation complexity is reduced; and tracking timeliness is improved.

The invention provides a tracking algorithm based on spatio-temporal context fusion multi-feature and scale filtering. Target position information and size information of a first frame of image are obtained through a video or image sequence; parameter initialization is performed; multi-feature extraction is performed; double-step preprocessing operation is performed, and operation of the cosine window is performed in a feature target area so that the edge effect caused by Fourier transform can be reduced; a filtering template and a scale filtering template are obtained through two times of Fourier transform; a position filtering template and the scale filtering template under the time domain space are obtained through two times of inverse Fourier transform, and the corresponding maximum value, i.e. the target area, is solved; finally the position filtering model, the scale filtering model, an adaptive appearance model and spatio-temporal context information are updated through a new frame; and the process returns to the feature extraction part and target tracking is performed through cyclic operation until the end of the process. According to the method, the tracking accuracy can be enhanced, the appearance change and the scale change of the target in the tracking process can be better adapted and the noise caused by the change of the environment in the tracking process can bereduced.

The invention provides an MEMS pedestrian navigation method based on ZIHR heading angle correction algorithm. The method includes: 1. conducting initial alignment on an MEMS pedestrian navigation system with an accelerometer and a magnetometer at a static moment; 2. working out an inertial navigation calculation equation and an error equation of the MEMS pedestrian navigation system; 3. conductingzero velocity state detection with the output values of a gyroscope and the accelerometer; 4. working out the relationship between the heading angle difference of adjacent moments under a zero velocity state, gyroscopic drift and a heading error angle in ZIHR (zero integrated heading rate) correction algorithm; and 5. establishing a simplified MEMS pedestrian navigation UKF filter model, and performing UKF filtering. The method provided by the invention maximumly utilizes the information of static moment, has uncomplicated calculation amount, and can well inhibit the divergence of the headingerror angle. Use of the UKF filter for real-time feedback correction at a zero velocity moment can well inhibit the problem of navigation parameter error divergence after long-time operation of a low-precision MEMS sensor, and improve the positioning precision of a pedestrian navigation system.

The invention provides a heterogeneous multi-sensor multi-target tracking method, which can rapidly start a target track, reduce calculation amount and improve track tracking accuracy and continuity.The method is realized through the following technical scheme: a track initiation and maintenance module, a track trimming and confirmation module, a hypothesis generation and management module, a track updating and merging module and a track management output module are arranged in a heterogeneous multi-sensor multi-target tracking system, the track initiation and maintenance module roughly classifies measurement data according to attribute characteristics and position grids of the measurement data, and according to the rough classification result, measuring points falling into an existing target track tracking wavegate are sent to the track trimming and confirmation module as candidate point tracks, the tracks confirmed by sequence probability ratio inspection are sent to the hypothesisgeneration and management module for track clustering and hypothesis generation, a target filtering model is determined through the track updating and merging module, the tracks are updated through filtering, and latest covariance information is output through the track management output module.

An echo cancellation invention which does away with the need for an accurate double-talk detector, while maintaining a higher adaptation gain for quicker convergence and also providing increased stability. It does this by operating two filter models of the acoustic path in parallel. One model is adapted continuously to find the most accurate model of the echo path, while the other is not. The echo canceller output is taken from the filter that is not adapted. A comparison of the residual error (echo) is done between the model being adapted and the model being listened to at regular intervals. When the model being adapted has less error (echo) than the model being listened to its filter coefficients are copied to the other model. If the model being adapted has greater error (echo) than the other model (caused by noise diverging the adaptation process) then the adapted model has its coefficients overwritten by the filter that is listened to. This process keeps error from accumulating in the adapted filter by removing divergence at regular intervals (improving overall convergence). Since the output is taken from the filter that is not adapted, higher adaptation gains may be used without fear of the user hearing any effects of adapting with a higher gain in the presence of noise.

Disclosed are methods for ink / toner saving including an ink / toner consumption controller which processes input image data via computer vision (CV) filtering models and other approaches to generate printable raster graphic, vector graphic and font character data on CMYK printing devices. User preferences may be specified using a GUI application. The ink / toner consumption controller manipulates the saving of colorent using dedicated techniques adopted for each type of document object. Models of color human perception are used for the combinations of the required percentages of a printer's CMYK ink / toner. These models are implemented by methods of CV selection rules for preferred objects and image areas, such as human faces, human skin and ROI, for maintaining maximum adherence to original image data. A color balancing method is used to ensure that CMY stacks consumptions are maintained similar throughout their usage. Additionally, methods are applied to compensate for missing CMYK colorants.

The invention discloses a nonlinear compensating method for a pseudo satellite / inertial integrated navigation system and provides a specific implementation mode of the method. The research shows that on the basis of a linearizing method, the method is added with quadratic term amendment, utilizes quadratic term to compensate linearizing error, reduces filtering model error introduced by nonlinearity of the pseudo range of the pseudo satellite to a certain extent and has obvious effect of improving the precision of the integrated navigation system. Simultaneously, the method adopts two-step algorithm to solve the problem of nonlinear calculation brought by the quadratic term, reduces the complexity of the algorithm and has practical value.

The invention discloses an object tracking method based on vision. The method comprises steps: each frame of image acquired by a mobile phone is used as input, a to-be-tracked target is selected by auser, and a multi-scale KCF tracker is operated; during the tracking process, the tracking confidence is calculated by a peak sidelobe ratio, and a tracking filter model is updated in the case of highconfidence; when the confidence is smaller than a certain threshold, the tracked object is proved to be blocked, and a YOLO (You only look once) detection process is operated at the time; and the YOLO detects multiple alternative targets, fine detection is carried out according to the already-learnt tracking filter model and the alternative targets, and the one with the maximum response value isthe tracked target. The method is accurate in target tracking and good in real-time performance.

The invention discloses a project-level and feature-level deep collaborative filtering recommendation algorithm based on an attention mechanism. The algorithm comprises the following steps of S1, counting historical project scores of a user; S2, calculating the feature level content representation of the user on the target project according to the historical project score of the user; and S3, calculating a project-level prediction score of the user for the target project according to the historical project score of the user and the technical result of the S2. According to the algorithm, the recommendation precision is improved to a certain extent by combining attention mechanisms on a project level and a feature level, and compared with the prior art, the algorithm has higher interpretability in analysis of historical preferences of users. The extended DACFs, such as recently proposed neural collaborative filtering and discrete collaborative filtering, will also be considered in othercollaborative filtering models, a higher-order characteristic level attention mechanism is explored for future research, and the theoretical basis of research of the recommendation system is further tamped.

The invention discloses a self-adaptive feature fusion-based multi-scale correlation filtering visual tracking method. The method comprises the following steps: firstly, the correlation filtering is carried out on a target HOG feature and a target color feature respectively by using a context-aware correlation filtering framework; the response values under the two features are normalized; weightsare distributed according to the proportion of the response values and then are subjected to linear weighted fusion, so that a final response graph after fusion is obtained; the final response graph is compared with a pre-defined response threshold value to judge whether the filtering model is updated or not; finally, a scale correlation filter is introduced in the tracking process, so that the scale adaptability of the algorithm is improved. The method can be used for tracking various features. The performance advantages of the features are brought into play, and a model self-adaptive updating method is designed. In addition, a precise scale estimation mechanism is further introduced. According to the invention, the updating quality and the tracking precision of the model can be effectively improved, and the model can be changed in scale. The method is good in robustness under complex scenes such as rapid movement, deformation, shielding and the like.

The invention provides an ordering strategy-based information filtering system, which relates to the technical field of information filtering, and solves the problems of inconsistency between an optimization objective and a filtering problem evaluating indicator, the deviation of a model optimization result and restricted performance in the conventional information filtering model. The information filtering system of the invention consists of a training model, a filter and a feature weight library, wherein a method for identifying a new information unit by the filter comprises the following steps: converting an information filtering problem into an ordering problem; performing optimization aiming at a core evaluating indicator 1-ROCA; establishing an ordering strategy-based information filtering model, wherein the ordering strategy-based information filtering model adopts an ordering logistic regression learning algorithm and comprehensively uses a TONE strategy-based parameter weight updating algorithm and resampling technology to obtain a weight parameter and obtain a prediction score value of the new information unit; and judging the attribute of a new mail according to the result of comparison of the prediction score with a predetermined threshold. The method of the invention can be applied to various information filtering and information push systems.

The invention provides a nonlinear system filter design method based on a self-adaption event triggering mechanism. The method comprises the steps of firstly building a state equation of a non-linear system; setting self-adaption event triggering conditions; building a filter model and a system error model of the non-linear system and obtaining an optimal filter of the non-linear system. The method employs the self-adaption event triggering mechanism for the non-linear system which is characterized by random nonlinearity, multiplicative noise and state constraints, so that the performance of filters is improved.

The invention discloses a self-adaption strong tracking unscented kalman filter (UKF) positioning filter algorithm based on fuzzy logic. The self-adaption strong tracking UKF positioning filter algorithm comprises the steps that: (1) a positioning filter model is built; (2) initial parameters of a filter are set; (3) the state quantity is subjected to filtering by adopting the UKF filter algorithm; (4) a fuzzy logic system is used for solving softening factors in the self-adaption tracking algorithm; (5) the self-adaption factors in the strong tracking self-adaption algorithm are solved; and (6) the epoch moment is increased by 1, the next moment observation is read, and the operation returns to the step (4) until the operation is completed. The strong tracking self-adaption algorithm in introduced on the basis of the UKF filter algorithm, in addition, a novel recursive algorithm is adopted in the strong tracking self-adaption algorithm for estimating the information covariance matrix, and the softening factors in the strong tracking algorithm are solved through a fuzzy logic reasoning system and are estimated in real time according to the work state of an epoch moment filter. The estimation is carried out in satellite navigation user receiver position estimation, and the positioning performance and the capability of carriers adapting to the dynamics can be greatly improved.

The invention relates to a system for automatically generating voice with stressed syllables, which comprises a phoneme positioning module, an acoustic characteristic extracting module, an acoustic characteristic parameter correcting module, a voice reconstructing module and a characteristic transformation parameter library, wherein the phoneme positioning module is used for determining the position of each phoneme in a received voice signal so as to obtain the time boundary of each word, syllable and phoneme; the acoustic characteristic extracting module is used for extracting the characteristics relevant to stress and the spectrum characteristics from the voice signal; the acoustic characteristic parameter correcting module is used for regulating the input characteristic parameter of voice into a corresponding characteristic parameter for representing the stress of given syllables and carrying out smooth treatment on the regulated characteristic parameter; the voice reconstructing module is used for synthesizing the voice anew by adopting the corrected acoustic characteristic parameter of the voice through a source-filter model; and the characteristic transformation parameter library is used for storing the transformation matrix parameter of the statistics of the acoustic characteristic of each phoneme from non-stress to stress.

The invention discloses a spatial-temporal constraint-based target re-identification method. The method includes the following steps that: a visual information probability-based preliminary ranking result is obtained based on the visual information of a target; based on a situation that the target which appears in one camera is impossible to appear in another camera, a Hinge loss function is utilized to construct a filter model to narrow a scope to be inquired; an inter-camera time probability model is realized through Wiebull distribution; inter-camera time preliminary probability is represented by the time interval probability of the target and a query target set, and therefore, an inter-camera time information and visual feature-based intermediate ranking result can be obtained based on the fusion of the inter-camera time probability model and visual information probability; and a target re-identification ranking result can be further optimized based on a situation that the probability of the same target appearing in two cameras which are close to each other spatially is larger than the probability of the target appearing in two cameras which are remote to each other spatially. With the method of the invention adopted, a target re-identification effect can be greatly promoted.

An implementation of a system and method for using inertial sensor data to improve mobile station positioning is provided. Many mobile devices include an inertial sensor. These inertia measurements typically produce low resolution inertial measurements, which may be used to generate an uncertainty value and / or to select a positioning filter model. In some embodiments, this uncertainty value is used as a process noise input signal into a Kalman filter or similar filter. In some embodiments, one position estimate is selected from multiple navigation position filter models operate concurrently, each computing a separate position estimate. In these embodiments, a position estimate is selected based on either the uncertainty value or based on another characteristic of the inertia measurements.

The invention discloses a combination navigation filtering method of a multi-model underwater vehicle. The navigation filtering method provided by the invention comprises the following steps of firstly establishing a state equation, an observation equation and a noise equation of a SINS / DVL / TAN / MCP combination navigation system according to a underwater vehicle combination navigation system; determining a model set according to a system equation and a noise model; selecting characteristic variable from the combination navigation system, and establishing a bayesian network; and correcting the model switching probability in multi-model estimation by adopting a bayesian network parameter according to a multi-model filtering algorithm structure, and calculating the estimation fusion of a filter in a weight sum manner. The data processing and resolving operations of the combination navigation are finished by a navigation computer according to a filtering model and an algorithm flow of the combination navigation system. The navigation filtering method provided by the invention has the advantages of being capable of improving the filtering accuracy of the combination navigation system under a complicated environment, and strengthening an autonomous navigation positioning property of the underwater vehicle.

The invention discloses an unsuspervised automatic extraction method of microblog new words based on repeated word strings. The method includes the steps that firstly, text segmentation is conducted on microblog documents to be processed, texts are segmented through a dynamic programming word segmentation method, the word strings to be recognized are segmented, and word segmentation fragments in the word strings to be recognized are combined into the new words to be recognized; candidate new words are extracted from the word strings to be recognized according to a statistic word selection model, and then the candidate words are filtered through a rule filtering model, and eventually the final new words are acquired. The method has the advantages that the high accuracy rate is effectively guaranteed, the method does not depend on a rule word stock too much, and the extraction speed of the new words is guaranteed.

A model in which filter drivers are managed to receive callbacks for I / O requests in which the filter drivers have registered an interest. Per-volume instances of filter drivers register with a filter manager for pre-callbacks (for I / O to the file system) and post-callbacks (for I / O from the file system), and identify which I / O requests (e.g., create, read, write) they are registering to receive callbacks. The filter manager orders the instances for callbacks. When an I / O request is received, the filter manager converts the I / O request to callback data and calls the interested filters in the callback order, whereby the filter instances can process the I / O data. As the request returns from the file system, filters desiring post callbacks are called in the reverse order. Efficient context management for the filters and other functions, such as non-reentrant file I / O, are also provided by the model.

The invention provides an intelligent simulating method for passively locating and tracking multiple targets in two stations. The method provided by the invention can solve the problem that the traditional method is weak in adaptation of target number and formation, low in locating and tracking precision and too large in calculating amount. The method has the following technical points: (1) a method of differentiating in single station, relating in two stations and cooperating with multiple properties is used for passively locating multiple targets in two stations intelligently; (2) the configurations and cooperated maneuvering strategies of two observation stations are made according to the influence of the system state and deploying manner to the locating precision; (3) the control algorithm is selected based on the filtering model collection and wave self-adaption adjusting algorithm of expert system and machine study; (4) the secondary related algorithm of interrupted track and new starting track and the smoothing algorithm after the relation is succeeded are constructed according to simulation association and inferential mechanisms. The method provided by the invention is suitable for rapidly and accurately locating and tracking multiple moving targets, and has important application values in military and civil aspects.