614 results about "Model matching" patented technology

System and methods for intelligent routing of requests from customers to agents where a request is received at a service center from a customer. Identification information related to the customer is accessed. Background information on the customer is retrieved from a storage facility. Task and attitude information about the customer is gathered. A model of the customer is created. Models for at least two agents are created. A performance optimizing calculation is performed that matches the customer model and the models for the at least two agents. A best match agent from the at least two agents is determined based on the matching. The customer request is routed to the best match agent.

Systems and methods for automatically and generically matching models are provided, such as may be provided in a matching application or matching component, or provided in a general purpose system for managing models. The methods are generic since the methods apply to hierarchical data sets outside of any particular data model or application. Similarity coefficients are calculated for, and mappings are discovered between, schema elements based on their names, data types, constraints, and schema structure, using a broad set of techniques. Some of these techniques include the integrated use of linguistic and structural matching, context dependent matching of shared types, and a bias toward subtree, or leaf, structure where much of the schema content resides.

A character generator and its method are suitably applied to such a video title generator for adding title images to a video program. Specifically, the character generator according to the present invention analyzes the form of characters inputted by an operator through a keyboard, and generates a polygon model matching the inputted characters so that the inputted characters look a three-dimensional image. The generated polygon model is subjected to rendering processing and displayed on a computer display. The polygon model displayed on the display can be changed in real time by using three-dimensional parameter and so on, for a shadow or perspective, which is set by the operator on the GUI screen. Thereby, the three-dimensional character generated by transforming the inputted character into the three-dimensional form can be generated with simple constructions and operations and in real time.

The invention provides an automatic synthesis method for a facial image. A user inputs two facial source images, a target image and semantic information of an area to be edited; a model matching module utilizes an active appearance model to automatically search an image and provide facial shape information and facial texture information of the facial image; then, the shapes of the two facial images are aligned by using the characteristics of the model, and the shapes of the source images in the edited area are aligned to a corresponding area in the target image by virtue of the deformation of a sheet sample band; a characteristic synthesis module is used for characteristic synthesis for the shape information and the texture information of the aligned source images and the target image, so as to generate a synthetic image automatically; an occlusion processing module is used for occlusion processing the synthetic image, and the occlusion areas of the source images are divided, matched and seamlessly integrated into the target image. Compared with the traditional facial synthesis, the invention integrates the characteristic areas of the two images and solves the problem of image distortion caused by partial face occlusion in the target image.

A method for profiling network traffic of a network. The method includes obtaining a signature library comprising a plurality of signatures each representing first data characteristics associated with a corresponding application executing in the network, generating, based on a first pre-determined criterion, a group behavioral model associated with the signature library, wherein the group behavioral model represents a common behavior of a plurality of historical flows identified from the network traffic, wherein each of the plurality of signatures correlates to a subset of the plurality of historical flows, selecting a flow in the network traffic for including in a target flow set, wherein the flow matches the group behavioral model without being correlated to any corresponding application of the plurality of signatures, analyzing the target flow set to generate a new signature, and adding the new signature to the signature library.

The invention relates to a pedestrian re-identification method based on global features and local features of an attention mechanism. The pedestrian re-identification method comprises the steps of respectively extracting the global features and the local features of pedestrians; in the global feature branch, taking the whole pedestrian feature image as input, sending the pedestrian feature image into a space attention mechanism module and a channel attention mechanism module, and fusing the feature representations of the two modules; in the local feature branch, horizontally and averagely dividing the pedestrian feature map into three parts, and inputting the three divided parts into a channel attention mechanism module to obtain the local feature of each part; sending the global feature and the local feature into a feature vector extraction module to obtain a feature vector for pedestrian prediction; and training the whole network to obtain a pedestrian re-identification model. According to the method, the global features and the local features of the pedestrian images are fully utilized, the attention mechanism is effectively fused, the pedestrian features have better discrimination ability, a good pedestrian re-identification result is obtained, and the model matching accuracy is improved.

A method for recognition of an input human motion as being the most similar to one model human motion out of a collection of stored model human motions. In the preferred method, both the input and the model human motions are represented by vector sequences that are derived from samples of angular poses of body parts. The input and model motions are sampled at substantially different rates. A special optimization algorithm that employs sequencing constraints and dynamic programming, is used for finding the optimal input-model matching scores. When only partial body pose information is available, candidate matching vector pairs for the optimization are found by indexing into a set of hash tables, where each table pertains to a sub-set of body parts. The invention also includes methods for recognition of vector sequences and for speech recognition.

Embodiments of the invention provide a personalized video recommendation method and apparatus. The method comprises the steps of obtaining a user portrait characteristic corresponding to a user identifier; according to the user portrait characteristic, obtaining a video characteristic model corresponding to the user portrait characteristic; and according to the video characteristic model corresponding to the user portrait characteristic, obtaining a video matched with the video characteristic model corresponding to the user portrait characteristic from a video database and performing recommendation. According to the method and apparatus, a user can quickly and accurately obtain the video interested by the user, so that the video acquisition efficiency is improved, the personalized video demand of the user is met, and the experience of user video search and recommendation is greatly enhanced.

The invention discloses a power grid multi-source data fusion method and system based on a multi-element heterogeneous model. The method comprises the following steps of multi-source data access, metadata storage, unified device modeling and data extraction, model matching degree calculation and multi-element data fusion. The system mainly comprises the five modules of multi-source data access, metadata storage, unified device modeling and data extraction, model matching degree calculation and multi-element data fusion. A unified model including multi-source system data is established by deeply analyzing the multi-source system data, automatic association is performed through model traversal comparison and model matching degree calculation according to the matching degree, automatic data integration and fusion of 90% or above is achieved, unfused data details and detailed causes are accurately analyzed, integration and fusion of various service data are achieved, and the overall fusion quality is effective managed and controlled.

The invention is applicable to the technical field of identity authentication, and provides a voiceprint recognition method and device. The method comprises the steps of preprocessing input voices toobtain a valid voice in each voice; extracting MFCC acoustic features of each voice and outputting first and second feature matrices including the MFCC dimension and the voice framing number; constructing a short- and long-time recurrent neural network model, and taking the first feature matrix as the input; training feature extraction matrices by using training parameters of the neural network model and speaker characteristics of the voices, each feature extraction matrix corresponding to a speaker model; and selecting the speaker model matching the second feature matrix, with the speaker output corresponding to the matched speaker model as a voiceprint recognition result. The method and device in the invention adopt a supervised learning method to train a voiceprint background model andcan excavate more suitable acoustic features from the training voices so as to be able to distinguish the difference characteristics of speakers more accurately and learn a more robust speaker model for a better voiceprint recognition result.

The invention discloses a pole tower model matching and visual navigation-based power unmanned aerial vehicle and an inspection method. In an unmanned aerial vehicle, a depth image of a front end of the unmanned aerial vehicle is acquired by a dual-eye visual sensor, distance between the unmanned aerial vehicle and a front object is further measured, a surrounding image is acquired by a cloud deckand a camera, the object is further identified, and the flight gesture of the unmanned aerial vehicle is controlled by flight control of unmanned aerial vehicle. The method comprises the steps of performing pole tower model building on different types of power transmission line pole towers; automatically identifying the power transmission line pole towers and pole tower types by the unmanned aerial vehicle during the flight process, matching and loading a pre-built pole tower model; performing visual positioning on the power transmission line pole towers by the unmanned aerial vehicle, and acquiring relative positions of the unmanned aerial vehicle and the pole towers; and performing flight inspection by the unmanned aerial vehicle according to optimal flight path. By the unmanned aerialvehicle, the modeling workload is greatly reduced, and the model universality is improved; and the inspection method does not dependent on absolute coordinate flight, the flexibility is greatly improved, the cost is reduced, and the power facility safety is improved.

A method for predicating remaining life of a turbine engine based on degradation model matching relates to a remaining life predication method for the turbine engine, and solves the problem that the predication effect cannot reach the predication requirement when a universal RUL (Remaining Useful Life) is adopted to predicate the remaining life of the turbine engine. The method comprises the specific steps as follows: step one, pre-treating data, that is, extracting a running state variable from the acquired data, acquiring a feature vector for a sensor, and fusing the running state variable and the feature vector to form a health factor; step two, building a degradation model base, that is, building degradation models by using the health factor, and forming the degradation model base by the degradation model groups; step three, evaluating the similarity, that is, matching a degradation track with the models in the model base and giving an RUL estimation to each model; and step four, fusing the RUL, that is, fusing by means of similarity weighting to obtain the final remaining life predication value according to the tested matching degree of the turbine engine and the model. The method for predicating remaining life of the turbine engine based on degradation model matching is applicable to the predication of the remaining life of the turbine engine.

Disclosed herein is a 3D urban modeling apparatus and method. The 3D urban modeling apparatus includes a calibration unit for calibrating data about a translation and a rotation of at least one capturing device at a time that input aerial images and terrestrial images were captured. A building model generation unit generates at least one 3D building model based on the aerial images and the terrestrial images to which results of the calibration have been applied. A terrain model generation unit generates a 3D terrain model by converting an input digital elevation model into a 3D mesh. A texture extraction unit extracts textures related to the building model and the terrain model from the aerial images and the terrestrial images. A model matching unit generates a 3D urban model by matching the building model with the terrain model, which are based on the textures, with each other.

A system and method for fusing the outputs from multiple LiDAR sensors on a vehicle. The method includes providing object files for objects detected by the sensors at a previous sample time, where the object files identify the position, orientation and velocity of the detected objects. The method also includes receiving a plurality of scan returns from objects detected in the field-of-view of the sensors at a current sample time and constructing a point cloud from the scan returns. The method then segments the scan points in the point cloud into predicted clusters, where each cluster initially identifies an object detected by the sensors. The method matches the predicted clusters with predicted object models generated from objects being tracked during the previous sample time. The method creates new object models, deletes dying object models and updates the object files based on the object models for the current sample time.

The invention discloses a bus load prediction method. The method comprises the following steps: correcting abnormal values in historical load data by use of a transverse comparison method, and determining key influence factors of bus load by use of a grey relation projection method; putting load curves with similar features in the same category by use of an improved K-Means clustering method to get a plurality of typical load patterns, building a random forest classification model, and establishing the mapping relationship between influence factors and clustering results; for each load pattern, training a plurality of prediction models by use of a multivariate linear regression method; and determining the category of a day under test, and selecting a matching regression model to realize load prediction. A data mining method is introduced to analyze the changing rules of bus load, and a prediction model library is built. Model matching is realized based on the category of a day under test. The accuracy and real-time performance of short-term bus load prediction are improved. More accurate decision support is provided for power grid planning and real-time dispatching.

The invention discloses a front-vehicle ranging method based on a monocular vision and image segmentation under a vehicle-borne camera. The method comprises a step of extracts a two-dimensional bounding box and a three-dimensional bounding box from a target vehicle based on a depth learning algorithm to obtain corresponding position information, a step of matching a 3D CAD vehicle model based on the length, width and height of the three-dimensional bounding box and obtaining a similar 3D vehicle model of a corresponding vehicle, a step of extracting vehicle model classification information ona vehicle in a picture based on the two-dimensional bounding box, a step of sending 3D information and vehicle model information corresponding to the vehicle to an example segmentation network, and calculating the absolute depth value of the vehicle in the image is calculated according to the size information of different vehicle models according to the camera imaging principle. According to the method, the time of the depth calculation is saved, a driving view is ensured, and so, a driver can intuitively observe a specific distance value of a front vehicle and make a reasonable driving judgment.

The invention discloses a feature and model mutual matching face tracking method based on on-line increment principal component analysis. The method includes the following steps: off-line modeling is performed on a plurality of face images to obtain a model matching (CLM) model A; key point detection is performed on each frame of a face video to be tracked, and a set of all key points and robust descriptors of the key points are combined to form a key point model B; key point matching is performed on each frame of the face video to be tracked on the basis of the key point model B to obtain an initial face gesture parameter set in each frame of the face images; the model A is used for performing CLM face tracking on the face video to be tracked; re-tracking is performed according to the initial face gesture parameter sets and initial tracking results; the model A is updated, the steps are repeated, and final face tracking results are obtained. The feature and model mutual matching face tracking method based on the on-line increment principal component analysis solves the problem of tracking losing occurred when variation between adjacent frames in a target image is large during CLM face tracking, thereby improving tracking accuracy.

The invention relates to the technical field of image processing, in particular to a paper currency recognition method and system. On the basis of a speeded-up robust feature (SURF) and model matching, each infrared anti-counterfeiting feature area of paper currency to be tested is subjected to SURF point detection, and then, the infrared anti-counterfeiting feature area is matched with a corresponding feature model of a real paper currency sample so as to realize a purpose of judging the authenticity of the paper currency to be tested. Compared with the prior art, the invention avoids detection to areas which do not exhibit practical meanings to a detection result on the paper currency, interference factors are reduced, and a detection result is more accurate. Meanwhile, detection calculated amount is reduced, and detection speed is quickened.

The invention discloses an indoor positioning method based on three-dimensional environment model matching. Indoor environment modeling, position analysis and information calibration are carried out by three-dimensional information acquired by a sensor; a user collects current scene information and performs quick comparison and precise feature matching on the current scene information and three-dimensional environment model information to realize indoor positioning of the user. The positioning method cannot be finished until a data acquisition device and a data analysis platform for indoor positioning are respectively built. The data acquisition device comprises but not limited to an RGB-D (red, green and blue-D) sensor, an inertia measurement sensor and the like and can finish acquiring space information such as three-dimensional information, the attitude and a deflection angle at any indoor position; the data analysis platform can carry out quick storage, position analysis and information calibration on the environment model by the acquired three-dimensional information and feed back an analyzed positioning result to a user; therefore, the indoor positioning function is realized.

The invention provides a parking lot management system based on video technology. The system comprises an entrance and exit control system, a parking guidance system, a vehicle identification system, a security monitoring system and a user charge system. The entrance and exit control system realizes vehicle detection, card issuing, card reading, the obtainment of the image information of a vehicle, vehicle characteristic information matching, parking space information display and the control of a gate through an underground detection coil, and the automatic entering and leaving of the vehicle are realized. The parking guidance system finds an available parking space number suitable for the parking of a user according to the vehicle type of a parking user and gives a path indication to the user. The vehicle identification system provides vehicle model information for parking guidance and provides basic information for vehicle license plate comparison and vehicle model matching and color matching in vehicle release through calling the video collected by entrance and exit video acquisition equipment. The security monitoring system monitors the state in a parking lot in real time. The user charge system is responsible for the charging of parking fees.

The invention discloses a commodity recommendation method based on model matching, which mainly comprises the following steps of: establishing a commodity description model, wherein the commodity description model comprises two parts, i.e. a character introduction and description part and an attribute information part; establishing a user preference model, ensuring that the user preference model is based on the commodity description model and fusing a plurality of commodity description models through the user preference model; calculating the similarity between the commodity description model and the user preference model; and ranking according to the similarity to recommend to a user. According to the commodity recommendation method, the potential interest of the user can be fully explored, and the accuracy of commodity recommendation can be realized.

The present application discloses a method and a device for searching a target in an image. The search method includes: locating a target in an image to be searched (S101); classifying the target and extracting a local feature of the target to generate a local feature code of the local feature (S102); searching an image database for suspected images of the same classification as the target through a tree classification index structure, comparing pre-stored local feature codes of the suspected images with the local feature code of the target, and outputting a suspected image with a similarity larger than a first threshold (S103). By locating a target in an image to be searched, performing at least two levels of classification on the target, generating local feature code of the target, and searching according to the classification of the target when searching and sorting by similarity, thus the target to be searched is model-matched within the same target classification, enhancing the searching accuracy and efficiency.

Waveform analysis is performed to identify and characterize power-consuming devices operating on a building electrical circuit. Current waveforms are measured from the building circuit with electrical devices operating thereon. The waveforms are separated into wavelets and analyzed to identify a representative wavelet model which is transmitted to a server for analysis. The server compares the representative wavelet model to a predictive model built from waveform signatures of known electrical devices operating on a circuit. When the predictive model matches the representative wavelet model, the electrical devices contributing to the representative wavelet, their operating mode(s) (e.g., “on”, “off”, “paused”, “hibernating”) and / or their performance state(s) (e.g., normal operation, deterioration, or failure modes) can be identified. This information can be communicated as feedback to the consumer to facilitate more efficient and more cost-effective energy usage.

The invention discloses an interactive video retrieval method based on unsupervised learning and semantic matching features, which comprises the steps of extracting image bottom layer features and model matching sequence features from the video image frame level of a video database; extracting semantic matching features from the high-level semantic level of image bottom layer features; performing unsupervised learning on the extracted model matching sequence features and semantic matching features, establishing retrieval and direct retrieval based on unsupervised learning, and forming an interactive interface by related feedback. The middle layer features, high layer features, unsupervised retrieval mechanism and interactive mechanism of an integrated video compose a set of new complete video retrieval system, which precisely measures the time-space sequence information of a video object, therefore better retrieval effect is achieved, the semantic understand of a sports video theme is developed, the online calculating complexity and retrieval time of a system are reduced, and the interactive interface greatly improves the retrieval performance of the system.

The invention discloses a mark point automatic registration method based on model matching. The method comprises the following steps of (1)obtaining image data containing mark points; (2) constructing a mark point model; (3) reading in the obtained image data, conducting anisotropic diffusion filtering on images, and automatically extracting a skin three-dimensional grid from the images; (4) adopting an ICP algorithm to be matched with the mark point model and the skin three-dimensional grid, and obtaining the coordinate of the center of each mark point on the skin three-dimensional grid in an image space coordinate system; (5) adopting the ICP algorithm to be matched with the center of each mark point of the image space and the center of each mark point of the actual space, obtaining a rotating matrix R and a translational vector quantity T between the image space coordinate system and the actual space coordinate system, and completing registration of the mark points. According to the mark point registration method, the multiple mark points can be rapidly registered, the obtained rotating matrix R and the obtained translational vector quantity T of the image coordinate system and the actual coordinate system are more accurate, manual interference is reduced, the precision of the registration of the mark points is improved, and the mark point automatic registration method has good robustness.

The invention discloses a model predictive control performance evaluation and diagnosis method. The method includes the steps: calculating the real-time performance value Ji and the average form Jnew of a system; selecting a segment of a data set and making the segment a historical performance benchmark value J<hist><*>; comparing the Jnew with the J<hist><*> and obtaining a system performance index [gamma]<k><*>, determining that the system performance is good if the [gamma]<k><*> is close to 1, and moving to the next step if the [gamma]<k><*> is close to 0; calculating an interference error e<0>(k), a predictive error e(k) and a model quality index [eta], and determining that the reason of deterioration of the system performance is external factors or a controller factor, otherwise, determining a system model mismatch and moving to the next step; detecting autocorrelation of an information sequence e(k), and moving to the next step if the autocorrelation of the e(k) exists, otherwise, determining that the model matching degree is good; and since n corresponding to the minimal loss function is the class of the e(k), determining a process model mismatch if the class of the e(k) is greater than the class of a process model, otherwise, determining an interference model mismatch. The overall performance of the system can be evaluated and deterioration sources of the system performance can be positioned only through closed loop input and output data.

The invention relates to a collision case playback system of the cars and two-wheeler in the computer application technical field, wherein an accident site reconstitution module collects the accident field data and picture, reconstitutes and reverts the site; a two-wheeler driver modularization dummy customizing module builds a dummy model matching with the body shape of the driver, extracts the damaged sensitive parts of the dummy model, sets up a finite element model, becomes a new conbination dummy model combined with the multi-rigid body dummy main body model; a collision process emulating module performs the scaling of model matching with the sufferer bodily form characteristics, meanwhile builds the integral accident site three-dimensional model before the accident, and simulates and computes the primary collision result; an optimization recurrence module builds the optimization flow, optimizes the collision simulation; an accident recurrence result post-processing module displays the optimized simulation result in the collision process animation and the collision parameters. The invention can realizes the site information collection, fast simulation and modelization, short treating period, high precision and high reliability.