624results about How to "Reduce computational cost" patented technology

An implementation of SVM functionality improves efficiency, time consumption, and data security, reduces the parameter tuning challenges presented to the inexperienced user, and reduces the computational costs of building SVM models. A system for support vector machine processing comprises data stored in the system, a client application programming interface operable to provide an interface to client software, a build unit operable to build a support vector machine model on at least a portion of the data stored in the system, based on a plurality of model-building parameters, a parameter estimation unit operable to estimate values for at least some of the model-building parameters, and an apply unit operable to apply the support vector machine model using the data stored in the system.

The invention discloses a monocular natural vision landmark assisted mobile robot positioning method. The method comprises the following steps: establishing a natural vision landmark feature library at multiple positions in a navigation environment in advance; matching the acquired monocular image and the vision landmark in the library through a robot in the positioning process by utilizing an inertial navigation system; establishing an online image rapid matching frame based on combination of GIST global features and SURF local features, and meanwhile correcting the vehicle course by combining the motion estimation algorithm based on the monocular vision; finally, effectively fusing the positioning information acquired through vision landmark matching and the positioning information acquired through the inertial navigation system by utilizing Kalman filtering. According to the method, positioning precision and robustness are high under the condition that the global position system (GPS) is limited, and the inertial navigation error caused by noise can be effectively corrected; the operation amount is greatly reduced by employing the monocular vision.

The invention belongs to the radiation source signal identification technology field and particularly relates to a radar radiation source signal identification method based on the one-dimensional convolutional neural network. The method comprises steps that (1), a radar radiation source signal data set is generated; (2), data pre-processing is carried out; (3), the convolutional neural network is constructed; (4), parameters are set, and the convolutional neural network is trained; (5), prediction classification is carried out; (6), precision calculation is carried out; and (7), the result is outputted. The method is advantaged in that 1, a network structure of the convolutional neural network is utilized to carry out characteristic extraction of signals, and a manual characteristic design process in a traditional algorithm can be avoided; 2, signals with a signal to noise ratio lower than -10dB can be accurately identified; and 3, the method can be simply realized.

The invention discloses a planning method for a mixed path of a mobile robot under a multi-resolution barrier environment. The invention aims to solve the problems of blindness of initial planning, environment modeling lack of flexibility and poor real-time obstacle avoidance capability of the present method. According to the technical scheme, a self-adapting inhomogeneous polar-radius polar coordinate modeling method is adopted for performing environment modeling on the motion space of the mobile robot; a gravity particle swarm searching method is adopted for planning an initial overall path from starting point to ending point; according to the initial overall path, a modified artificial potential field method is adopted for performing local dynamic obstacle avoidance by estimating the minimum safe distance and safe collision-preventing angle and for arriving at each initial overall path point in turn; and a final overall collision-free path is output after arriving the planning end point. According to the planning method provided by the invention, the blindness of initial overall planning and the environment modeling flexibility can be effectively improved, the real-time obstacle avoidance capability for dynamic unknown barrier is strong, and the method is high in speed, high in precision and strong in adaptability.

A computer simulation enhanced exercise device is provided which engages the user by directly relating the users exercise motion in real time to a visual simulation or interactive game. The exercise device may comprise any variety of machines including, stationary bikes, rowing machines, treadmills, stepper, elliptical gliders or under desk exercise. These exercise devices are configured with sensors to measure physical movements as the user exercises and are coupled to computer hardware with modeling and virtualization software to create the system. These sensor measurements are then sent to a computer for use in the physics based modeling and real-time visual simulation. The computer simulation enhanced exercise device is further provided with features including manual and automatic adjustment of resistance levels, visualization of accurate caloric bum rates and correlation to everyday food items, and network connectivity providing for multiplayer network simulations and directed advertising.

The invention relates to a rapid three-dimensional measurement method based on colorful grating projection in a three-dimensional scanning system. The method comprises the following steps: selecting four pure colors of green, yellow, cyan and white so as to carry out color strip encoding, wherein the G component values of the four pure colors are 255, and the R component values and the B component values of the four pure colors are respectively 0 or 255; after finishing color encoding, carrying out sinusoidal modulation on the G component of strip color to finally obtain a projective grating; on one hand, resolving G component information to obtain a folding phase value by a method based on Fourier transform, on the other hand, carrying out binaryzation on the R component and B component of an image by an iteration threshold method in the color image segmentation link, and automatically assigning 255 to the G component; and segmenting an image decipher to obtain phase period information so as to unfold the phase. By utilizing the method, inconvenience for cutting brought by periodic occurrence of light and shade strip-shaped areas in the image caused by G component sinusoidal modulation can be well avoided, and a train of thought for realizing three-dimensional dynamic measurement can be provided on the premise of not influencing the measurement accuracy.

The invention discloses target detection system and method based on covariance and a binary-tree support vector machine. The system comprises a video data collecting unit, an image preprocessing unit and a background modeling vehicle partitioning and displaying unit. The video data collecting unit is used for acquiring information in real time and carrying out digitization and system conversion on analog videos. The image preprocessing unit comprises an FPGA (Field Programmable Gate Array) and a DSP (Digital Signal Processor), wherein the FPGA is used as a coprocessor; and the DSP is used as a main processor to accomplish the background modeling of video images, the partitioning and the extraction of vehicle targets and the realization of a model identification algorithm. Utilizing the combination of the FPGA and the DSP, the invention can realize multi-video real-time model identification through combining with the model identification algorithm based on the covariance features of the images and the support vector machine. The invention can be widely used for a plurality of fields of intelligent traffic management, intelligent video monitoring and the like.

An embodiment of the invention discloses a video class identification method and device, a data processing device and an electronic device. The method comprises the following steps: carrying out subsection on a video to obtain a plurality of segmented videos; carrying out sampling on each segmented video in the plurality of segmented videos to obtain an original image and an optical flow image of each segmented video; processing the original image of each segmented video by utilizing an airspace convolution neural network to obtain an airspace classification result of the video; processing the optical flow image of each segmented video by utilizing a time domain convolution neural network to obtain a time domain classification result of the video; and carrying out fusion processing on the airspace classification result and the time domain classification result to obtain a classification result of the video. The video class identification method and device can improve video class identification accuracy.

The invention discloses a mobile robot route planning method based on a distance grid map. The mobile robot route planning method includes selecting front grids by a moving window method in the process of moving of a robot, updating the distance grid map, managing the front grids to be updated by an array management method during updating, and updating a control model of the mobile robot by an A*search algorithm in the mobile robot route planning method based on the distance grid map. The mobile robot route planning method based on the distance grid map is of great fundamental significance in high-ranking tasks like generating topological maps and planning a route in the unknown environment.

The invention discloses a travel time prediction method based on multi-modal data fusion and multi-model integration. The travel time prediction method comprises a multi-modal data preprocessing module which extracts taxi travel data from taxi GPS track data according to the passenger carrying state; a multi-modal data analysis, feature extraction and feature fusion module which is used for extracting corresponding feature sub-vectors from the fields of taxi track data, weather data, driver portrait data and the like and completing feature splicing; and a multi-model integration module which is used for respectively establishing a gradient improvement decision tree model and a deep neural network model, and integrating prediction results of the models by using the decision tree model. According to the travel time prediction method, by fusing the multi-modal data such as taxi track data, weather data and driver portrait data, the factors influencing travel time are fully extracted and mined, and an integrated model based on a decision tree is established, so that higher travel time prediction accuracy is obtained at lower calculation cost.

The invention discloses an on-line calculation method of probability power flow based on BP neural network, which mainly comprises the following steps: 1) establishing a BP neural network power flow model. 2) initializing the basic parameters of BP neural network power flow model. 3) acquiring training sample data. 4) identifying training objectives; using the training sample data, the BP neural network power flow model being trained to obtain the trained BP neural network power flow model. 5) obtaining a calculation sample. 6) inputting the training sample data obtained in the step 3 into theBP neural network power flow model trained in the step 4 at one time to obtain the training target, thereby judging the solvability of the power flow of all the training samples; calculating the tidal current value of the solvable sample. 7) performing statistic probability power flow index. That invention can be widely apply to on-line calculation of probability power flow of a power system, andis particularly suitable for the situation that the uncertainty of the power system is enhance due to high proportion access of new energy sources.

The invention discloses a method for clustering gene expression data based on manifold learning, and the method provided by the invention comprises the following steps: acquiring a gene expression data matrix A through an acquisition system, and preprocessing the gene expression data matrix A by using a local linear smoothing algorithm; introducing the preprocessed data matrix A, and constructing a weighted neighborhood figure G in a three-dimensional space; taking the shortest path between two points as the approximate geodesic distance between two points; calculating a two-dimensional embedded coordinate by using an MDS (minimum discernible signal), and mapping the three-dimensional data matrix A to a two-dimensional visual space; and carrying out clustering on the two-dimensional visual space subjected to mapping by using a k-mean clustering algorithm so as to obtain the clustering result. The clustering method has the characteristics of low calculating cost, capability of eliminating high-order redundancies, suitability for pattern classification tasks, and the like; and by using the method disclosed by the invention, the current states of cells, the effectiveness of medicaments to malignant cells, and the like can be discriminated effectively according to the clustering result. The invention also provides a system for clustering gene expression data based on manifold learning.

The invention discloses an efficient condition privacy protection and security authentication method in an internet of vehicles. The efficient condition privacy protection and security authentication method comprises the following steps of: system initialization, generation of pseudonym identities and signature private keys of vehicles, signing and authentication of a message and tracing of real identities of the vehicles. The vehicles carry out cooperative communication with surrounding vehicles and roadside units deployed at both sides of a road by on-board units assembled on the vehicles, driving security of the vehicles can be effectively improved, and vehicle users can more conveniently and rapidly acquire related traffic services. The efficient condition privacy protection and security authentication method disclosed by the invention not only can meet security requirements in the internet of vehicles, but also optimizes the computing process of signature generation and verification in the communication. The efficient condition privacy protection and security authentication method is greatly improved on the aspect of efficiency of computing cost, communication cost and the like, and is more applicable to communication and application in the internet of vehicles.

The invention discloses an image / video demisting method based on the combination of a dark primary color and atmospheric scattered light. The method comprises the following specific steps of: 1, inputting an original misty image I into an image processing system of a computer, and acquiring a dark primary color image Idark of the original misty image I; 2, according to the obtained dark primary color image Idark, evaluating the atmospheric light values A, i.e. airR, airG and airB of red, green and blue (RGB) channels of the original image; 3, evaluating the atmospheric scattered light value V (x,y) of the original misty image I; and 4, evaluating a demisted restored image J according to images of the RGB channels of the original misty image I, the atmospheric light values A and the atmospheric scattered light value V to obtain a demisted restored image J finally. According to the image / video demisting method based on the combination of the dark primary color and atmospheric scattered light, a clear demisted image is obtained by calculating the dark primary color image of the original image and restoring the original image with the atmospheric light values and the atmospheric scattered light value.

The invention discloses a traffic signal identification method and device, vehicle navigation equipment and an unmanned vehicle. The method comprises the steps of obtaining a current image in a videoshot by the unmanned vehicle; when the image detection condition is satisfied, detecting a region of interest of a traffic signal lamp in the current image through a traffic signal lamp detection model; otherwise, based on the region of interest of the traffic signal lamp in the previous frame of image of the current image in the video, determining the region of interest of the traffic signal lampin the current image by using the target tracking model; using a trained traffic signal lamp state identification model; identifying a traffic signal represented by a traffic signal lamp in a regionof interest of the current image; in view of the fact that the calculated amount of the target tracking algorithm is smaller than the calculated amount of the model detection. The region of interest of the traffic signal lamps in the continuous frames are identified in combination with the model detection and the target tracking, recognition of each frame of image can be achieved, the possibilityof missing detection is reduced, and the calculation cost for recognizing the continuous frames can be effectively reduced.

The invention discloses a secret key updating method for cloud storage and an implementation method of a cloud data auditing system, and belongs to the technical field of network security. The secret key updating method for cloud storage comprises the steps that when a cloud user needs to update a secret key, a CA server is requested to generate a new secret key, and a new file label and a new data block label are generated based on a file label and a data block label downloaded from a cloud server, the old secret key and the new secret key at present, are uploaded to the cloud server, and are used for replacing the corresponding old file label and the corresponding old data block label in the cloud server. Meanwhile, the invention further discloses the implementation method of the cloud data auditing system on the basis of zero-knowledge verification. When the cloud user needs to update the secret key, the corresponding file label and the corresponding data block label on the cloud server are updated based on the secret key updating method for cloud storage. The secret key updating method for cloud storage and the implementation method of the cloud data auditing system are used in a cloud network, the communication cost, caused by changing of the secret key, between the cloud server and the cloud user can be remarkably reduced, the calculation cost of the operation that the cloud user calculates the labels again is reduced, and the data privacy can be effectively protected in the auditing process.

The invention discloses a fake fingerprint detecting method based on a SVM and sparse representation. The method comprises the steps of collecting hundreds of true and fake fingerprint images, extracting characteristic data of statistical characteristics, frequency domain characteristics and the like, uniformizing the characteristic data, carrying out supporting vector machine (SVM) training to obtain classification models of the SVM, extracting images of fingerprints needing to be detected, extracting and uniformizing the same characteristic data, classifying the classification models of the SVM to obtain true or fake SVM classification results, randomly extracting subimages of the fingerprint images, training a sparse representation dictionary, randomly extracting subimages of the fingerprint images needing to be detected, carrying out sparse representation to judge the subimages as true subimages or fake subimages, and finally using the classification and judgment results to carry out compound decision. The method needn't improve fingerprint collecting hardware, is quick in computation speed and high in accuracy and has important application value in improving safety of a fingerprint recognizing system.

The invention discloses a trusted storage verification method and a trusted storage verification system for cloud storage and belongs to the technical field of computer software. In the method and the system, before a file is transmitted to an untrusted cloud storage server, a series of random positions are generated according to the key held by a user and other generated related verification parameters, the contents at the random positions in the file are read, a plurality of verification labels are generated for the file, and all necessary parameters are stored and maintained; and when the storage state of the file is required to be checked, a user can initiate an interaction process with a cloud storage system according to related parameters, and the cloud storage system can generate new verification labels again according to parameters corresponding to the verification. In the method disclosed by the invention, a higher verification reliability can be acquired at a lower computingcost, the contents at different positions in the file are selected at each time of generation of the file verification labels, and different keys are adopted to prevent a server from generating a correct signature by using a stored correct signature or by storing the file contents at a specific position.

The present invention discloses a ship safety forewarning method based on a classification detection of collision. The ship safety forewarning method provided by the invention comprises the following steps: (a) acquiring state information of a ship on the basis of an established three-dimensional virtual scene of a ship navigation channel; (b) classifying the collision types according to the driving direction angle with respect to the ship in the three-dimensional virtual scene; (c) calculating the minimum safe distance among ships through different determination methods of collision types; and (d) performing collision determination and emitting safety warning to the ships in a dangerous situation. The ship safety forewarning method based on classification detection of collision provided by the invention is capable of visually and accurately determining the ships in a potential collision danger in the navigation channel three-dimensional virtual scene in advance, and has the advantages of good accuracy, low operation cost, low interference rate and good application prospects.

The invention discloses an illumination-classification-based adaptive image segmentation method which is used for accurately segmenting a target object under different illumination conditions. The illumination conditions are divided into two types, namely a frontlighting type and a backlighting type, by extracting color characteristics of an image to be processed in a red, green and blue (RGB) space and a hue, saturation and value (HSV) space and adopting a minimum euclidean distance classifier; a proper color characteristic quantity serving as a segmenting parameter is extracted from the image in the two illumination types and imported into a two-dimensional histogram; neighbor information of each pixel point is increased, so the interference resistance capacity is improved; and the acquired image is subjected to intelligent illumination judgment and precise segmentation. In the illumination-classification-based adaptive image segmentation method, a mode of judging the illumination condition first and then selecting a segmenting algorithm is adopted, so the algorithm has higher pertinence and the effectiveness of the algorithm is improved; meanwhile, illumination correction is not required, so the computing cost is reduced greatly; and a favorable condition is created for the subsequent image processing and analysis.

The invention belongs to the technical field of computer network security, and relates to an anti-attack reliable wireless sensor network node positioning method. In ordinary positioning methods, nodes to be positioned are easy to attack by malicious nodes, so the positioning effects of the nodes are influenced. By the reliable positioning method provided by the invention, Sybil attacks and wormhole attacks can be resisted at the same time. The method comprises that: a beacon node estimates the smallest hop number between the beacon node and other beacon nodes; the beacon node estimates an average hop distance among the nodes; the nodes cooperate with one another to detect attack nodes in a network; and unknown nodes estimate own coordinates. Compared with the prior art, the invention improves the attack resistance of the wireless sensor nodes in a positioning process, reduces positioning errors in an environment with the malicious nodes, and reduces the communication cost and calculation cost of the nodes as much as possible in the positioning method by fully combining the characteristic that the energy of the wireless sensor nodes is finite.

The invention relates to a forecasting method for a multi-stage differential evolution protein structure based on abstract bulge estimation. The method comprises the following steps: firstly, calculating the distance from each conformation individual in a current colony to a new conformation and performing ascending sorting according to the distance; then selecting the part of the new conformation individual close to a abstract bulge lower-limit estimation support surface of the conformation individual, thereby acquiring an energy lower-limit estimation value of the new conformation individual; calculating an average estimation error between the energy lower-limit estimation value of all the new conformation individuals and a practical energy value; dividing the whole algorithm into a plurality of optimizing stages according to the change in the average estimation error; judging the stage of the present iteration according to the average estimation error in the last iteration; and designing different strategies for all the stages and generating the new conformation individual. The forecasting method for the multi-stage differential evolution protein structure based on the colony abstract bulge estimation provided by the invention is high in forecasting precision and low in calculation cost.

The invention belongs to the technical fields of statistical pattern recognition and image processing, in particular to a super-resolution method based on an artificial neural network. In the invention, the artificial neural network is used for expressing the function mapping relation among low-resolution images and high-resolution images. The method of the invention comprises the following concrete steps: creating a training set; establishing a BP neural network for training; bonding high-resolution images which are obtained by training according to the corresponding relation; and then, obtaining super-resolution images. The invention overcomes the disadvantage of time consumption of the original super-resolution algorithm based on manifold learning, and obtains better effect.

The invention relates to a mobile edge computing auxiliary vehicle task unloading method and belongs to the field of vehicle communication, and vehicle tasks can be divided into unloadable tasks and non-unloadable tasks according to attributes of the vehicle tasks. Selecting a local computing task for the task which cannot be unloaded; For an unloadable task, the vehicle utilizes the local computing resources and the MEC computing resources to process the task together, and the purpose of minimizing the task computing cost is achieved. the optimal offload decision depends on a comparison between benefits brought by the task in local processing and benefits brought by offloading to the MEC. In the task calculation process, the data packet queue is dynamically changed, data packets arrive and leave the queue, and meanwhile packet loss is generated due to time delay. Under the condition that the queue is kept stable, the packet loss rate of the task is optimized, the task calculation costcan be reduced, and the vehicle user data transmission experience is improved.

A high-resolution remote sensing image scene classifying method based on unsupervised feature learning comprises the steps that original input high-resolution remote sensing images are divided to obtain scenes, a plurality of training image blocks are randomly extracted from each scene, and the training image blocks are gathered for conducting preprocessing operation; the low dimension manifold representation of all the training image blocks is calculated, and a set of clustering center is obtained by clustering; intensive sampling is conducted on each scene to obtain local image blocks, each local image block is subjected to preprocessing operation and then is mapped to the same low dimension manifold space, and then encoding is conducted to obtain all local features of the scenes; the local features of all the scenes are gathered to conduct feature quantization, the local feature column diagrams of all the scenes are counted to obtain the global feature representation of the scenes; a plurality of scenes are randomly selected to be used as training samples, the predicted class labels of all the scenes are obtained through a classifier, and thus the labeling task of the original high-resolution remote sensing scenes is achieved.

The invention discloses a self-adaptive spectral subtraction real-time speech enhancement method. The method comprises: establishing a dynamic threshold including discrimination of voice and no voice of noise voice, providing noise spectrum time varying update principles according to the dynamic threshold; making full use of correlation extraction information among adjacent frames, and realizing a pure voice spectrum smooth estimation method. Aimed at a practical problem that voice signals are difficult to extract in unstable noise and strong background noise, the self-adaptive spectral subtraction speech enhancement method is provided. The method uses a rapid tracking noise algorithm to perform smooth update on the unstable noise frame by frame, and can preferably estimate noise spectrums. The algorithm can effectively restrain background noise, and improves voice quality and intelligibility after noise reduction. The method is low in calculation cost, and is easy to realize, and has good real-time property. The method provides a new approach for denoising of strong background noise and detection of weak signals.

The invention provides a human face recognition method based on supervision isometric projection. The human face recognition method comprises the human face sample training process and the human facesample testing process. The human face sample training process comprises the following steps: firstly carrying out pretreatment on a human face training image, adopting Gabor wavelet for filtering theimage, proposing a new distance formula for calculating an adjacency matrix of a training sample, calculating a shortest path distance matrix D in the training sample by the adjacency matrix DG of the training sample, calculating a low-dimensional projection matrix describing data of the human face training sample, calculating the projection of the training sample in low-dimensional space througha projection conversion matrix A and the like; and the human face sample testing process further comprises the following steps; carrying out the pretreatment on a human face testing image, adopting the Gabor wavelet for filtering the image, calculating the projection of the testing image in the low-dimensional space, adopting a nearest neighbor algorithm for judging the type of a testing sample and the like. The human face recognition method is characterized by stronger description of the structure of the sample data, elimination of high-order redundancy and small calculation cost, thereby being more applicable to mode classification tasks and the like.

The invention provides a stream cipher key control method for fusing a neural network with chaotic mappings. A sending party and a receiving party are configured to have neural network weight synchronization models with same parameters and three chaotic mapping functions with the same initial value, and the sending party and the receiving party are set to have the same initial value of the chaotic functions. The method comprises the following steps: (1) determining a hybrid stream key generator based on three chaotic mappings; and (2) realizing a chaotic function initial value update mechanism based on neural network weight synchronization. The method provided by the invention can be used for updating the chaotic initial values and strengthening the safe application of stream ciphers.

The invention discloses a layer classification method for converting architectural drawings into a three-dimensional BIM model, comprising the following steps: a, importing CAD architectural drawingsinto BIM modeling software, and converting graphic element information in CAD architectural drawings into a basic graphic element information database and a text information database; B, retrieving the layer information in the CAD architectural drawings, traversing the pretreated CAD architectural drawings, and putting the basic graphic elements of the same layer into a collection; C, taking the image layer to be identified as the target image layer, and sequentially calculating the total matching degree score of the image layer data set of each image layer, thereby obtaining the target imagelayer; carrying out the next round of matching degree score calculation on the layer data set of each layer until all the target layers are obtained; D, after obtaining all the target layers, identifying and modeling on the designated target layers. The invention automatically classifies layers of CAD architectural drawings and improves the efficiency of recognizing and reconstructing 3D BIM models of CAD architectural drawings.