375 results about "Support vector machine classification" patented technology

The invention provides a fault diagnosis method for a rolling bearing based on a deep learning and SVM (Support Vector Machine). The method comprises using a manure learning algorithm in a deep belief network theory to complete a characteristic extraction task needed by fault diagnosis; automatically extracting the substantive characteristics of data input independent of manual selection from simple to complicate, from low to high, and automatically digging abundant information concealed in known data; in addition, classifying and identifying a test sample by adopting an SVM classification method, seeking and finding a global minimum of a target function through an effective method previously designed, so as to solve the problem that a deep belief network may be trapped into a locally optimal solution. According to the fault diagnosis method for the rolling bearing based on the deep learning and SVM provided by the invention, the accuracy and effectiveness of the fault diagnosis method for a rolling bearing can be improved, and a new effective way can be provided to solve the accuracy and effectiveness of the fault diagnosis method, therefore the fault diagnosis method can be extensively applied complex systems in chemistry, metallurgy, electric power, aviation fields and the like.

The invention relates to a recognition method of digital music emotion, belonging to the field of computer pattern recognition; the recognition method solves the problem that the existing recognition method of digital music emotion can not recognize sampling-based digital music format, the sorting technology based on a multi-class support vector machine is adopted, acoustic characteristic parameters and music theory characteristic parameters are combined, so as to carry out emotion recognition of digital music; the recognition method comprises the following steps: (1) pretreatment; (2) characteristic extraction; (3) training the multi-class support vector machine; (4) recognition. The music emotion is classified into happiness, impassion, sadness and relaxation, the emotion recognition is carried out based on a sampling-based digital music format file, the common acoustic characteristics in the speech recognition field are not only extracted, and a series of music theory characteristics are extracted according to the theory characteristics of music; meanwhile, the sorting method based on the support vector machine is adopted, the leaning speed is rapid, the sorting precision ratio is high and the recognition efficiency is improved.

The invention discloses a full-view monitoring robot system, which comprises a monitoring robot, a wireless interaction unit and a remote monitoring terminal, wherein the monitoring robot comprises a robot housing, an image acquisition unit, a sensing unit, a processor and a moving unit; the image acquisition unit comprises a plurality of cameras which surround the robot housing at intervals for acquiring all-around images on the four sides of the monitoring robot; the sensing unit comprises a sensor network on the robot housing; the processor comprises an image detection unit and a motion controller, wherein the image detection unit extracts characteristics of a directional gradient column diagram from the images acquired by the image acquisition unit, classifies linearly supported vector machine, detects human body images according to the classification result and generates a control command when the human body images are detected; and the motion controller receives the control command and controls the travel unit to travel according to the control command. The system can perform 360 degree full-view monitoring and improve monitoring efficiency. Besides, the invention also provides a monitoring robot for use in the full-view monitoring robot system.

The invention discloses an indoor passive positioning method based on channel state information and a support vector machine. The method comprises the following steps: firstly preprocessing the acquired channel state information data, performing de-noising and smoothness through the adoption of a density-based spatial clustering of applications with noise and a weight-based moving average algorithm, and then using the principal component analysis algorithm to extract the features. The data after the preprocessing and feature-extracting can reflect the signal change more accurately and the dimension is greatly reduced. The passive positioning adopts two-stage positioning. In the training stage, the large positioning space is divided into sub-regions, the support vector machine classification and regression model is established for each sub-region so as to acquire a statistic model for accurately representing the nonlinear relationship between the position and the signal. The two-stage positioning firstly determines the sub-regions through the classification of the support vector machine, and the precision position is determined in the sub-region through the regression of the support vector machine. The method disclosed by the invention has the beneficial effects that the passive positioning can be performed in the absence of the active participation of the target, and the indoor positioning precision is improved to sub-meter level.

The invention discloses a method and device for extracting a pulmonary lobe from a chest CT image and relates to the technical field of a computer. The method can obtain pulmonary lobe tissues through a pulmonary parenchyma extraction process based on 3D region growth, a left and right lung adhesion trachea removal process based on region features, a pulmonary vascular root removal process, a pulmonary vascular center path extraction process based on topological thinning and a pulmonary lobe segmentation algorithm based on support vector classification. The method can accurately extract the pulmonary lobe from the chest CT image, can accurately finish quantitative evaluation of illness degree of each pulmonary lobe and is more accurate and effective for diagnosis and treatment of lung diseases.

The invention relates to a search method of related feedback images. The search method comprises the following steps: initially searching, selecting feedback samples, establishing a separating hyperplane of a support vector machine, establishing transformation mapping by biased discriminant analysis, constructing a piecewise similarity measurement function, and performing multi-feedback turn combination and feedback iteration. The method helps cause efficient intelligent retrieval effect in the case of small samples and can help meet real-time requirements by performing three operations of support vector machine classification, the biased discriminant analysis transformation mapping and the multi-feedback turn combination.

The invention discloses a medical insurance fraud detection method based on multiple features. The method comprises: aimed at treatment histories of all patients suffered from the same disease in medical insurance declaration data, in combination with medicine classification knowledge, adopting probability statistics, mixture Gaussian modeling, feature fusion and other techniques to extract multiple secondary feature data with higher distinction degree; then, vectorizing the treatment histories of the patients based on the secondary feature data; and then, after carrying out clustering analysis on treatment data marked with 'normal', adopting a non-linear support vector machine classification technique to establish multiple classification hyperplanes for each type of normal treatment data subjected to clustering and treatment data marked with 'fraud', so that the fraud detection can be carried out on non-marked medical insurance data. The method can be used for quickly and effectively detecting the fraud data existent in the medical insurance data, and has relatively high accuracy.

The invention discloses a vehicle classification method based on a convolution neural network. The vehicle classification method is concretely implemented according to the following steps that step 1: learning samples are acquired, and category tags are marked on the samples; step 2: the acquired learning samples act as training data to train a convolution neural network model so that excellent network model parameters are obtained; step 3: the characteristics of training data are extracted by using the trained convolution neural network model, and tenfold cross training is performed by using a support vector machine classification model constructed by a liblinear classifier so that the support vector machine classification model is obtained; and step 4: the characteristics of vehicle models under classification are extracted by suing the convolution neural network model, and then the vehicle categories to which vehicle images under classification belong are obtained by using the support vector machine classification model. All the connection layer output of the convolution neural network is used as the characteristic representation of the vehicle images and then the vehicle images are classified by using a SVM classifier so that the great accuracy of vehicle classification can be obtained.

An expression recognition method based on AVR and enhanced LBP in the technical field of model recognition comprises the following steps: collecting an original image; expanding virtual samples; carrying out wavelet decomposition on human image; extracting local binary pattern characteristic LBP; calculating an enhanced variance ratio AVR characteristic value and adding with a penalty factor, then extracting a plurality of groups of characteristic values of different dimensionality discriminated by AVR value, carrying out support vector machine classification accuracy test, and taking the characteristic dimensionality with the highest accuracy and corresponding characteristic value as the LBP characteristic. The method of the invention integrates image acquisition, human face test and human eye test, enhances the LBP characteristic by wavelet decomposition, and effectively improves the accuracy by adopting AVR method to extract effective characteristic.

The invention discloses a partial discharge detection identification method based on ultrasound and ultraviolet information fusion. The method comprises the following steps: S1. uniformly setting sensing detection circuits in surrounding space of a detected object; S2. real-timely collecting an ultrasonic signal and an ultraviolet signal generated when partial discharge is generated in a detected area, after the signals are processed, sending to a digital signal processing circuit; S3. carrying out digital filtering processing; S4. extracting dual density wavelet transform wavelet coefficient Shannon entropy x from the ultrasonic signal generated in S3, extracting wavelet packet wavelet coefficient Shannon entropy y, and sending a characteristic vector x and a characteristic vector y to a detection system host via a communication module; S5. carrying out characteristic fusion by the detection system host so as to obtain the characteristic vector z after the fusion; S6. classifying the vector z obtained in S5 after the fusion by using support vector machine classification trees. By using the method of the invention, application of a detection identification technology based on the ultrasound and ultraviolet information fusion can be promoted; high accuracy, reliability and instantaneity of the partial discharge detection can be ensured.

The invention discloses a weed image segmentation method under the rape field environment. Multiple weed/rape RGB image samples are randomly acquired in the rape field; a visual attention model is established, the color characteristics, the brightness characteristics and the direction componential characteristics are extracted, each characteristic graph is acquired and each characteristic channelsaliency map is generated so that a total saliency map is acquired and the area of interest is acquired; the shape characteristics and the texture characteristics of the area of interest are extractedto perform support vector machine classification training so as to acquire the rape area; and the miscellaneous image samples and all the rape area images are fused so as to acquire the final inter-strain weed area distribution information. The area of interest is acquired through fusion of the improved visual attention model with combination of the region growth algorithm, and the whole algorithm process does not require grayscale transformation or threshold segmentation so that the processing link and the computing amount can be reduced; and the segmentation efficiency is further enhanced by extracting the characteristic parameters of the area of interest and support vector machine classification model judgment so that weed image segmentation under the background of the rape field can be realized.

An incremental greedy method to feature selection is described. This method results in a final classifier that performs optimally and depends on only a few features. Generally, a small number of features is desired because it is often the case that the complexity of a classification method depends on the number of features. It is very well known that a large number of features may lead to overfitting on the training set, which then leads to a poor generalization performance in new and unseen data. The incremental greedy method is based on feature selection of a limited subset of features from the feature space. By providing low feature dependency, the incremental greedy method 100 requires fewer computations as compared to a feature extraction approach, such as principal component analysis.

The invention discloses an indoor positioning method based on manifold learning and an improved support vector machine. The method comprises a step of determining a positioning area, dividing the positioning area according to an indoor structural characteristic and a layout characteristic, and obtaining a classification result, a step of obtaining offline training data, and collecting hotspot RSS signal values which can be received by the reference points in different classification area as a training data set, a step of using an isometric mapping algorithm to carry out training data characteristic extraction, a step of using the training data to carry out support vector machine classified training, using a taboo search algorithm to carry out support vector machine classification hyper parameter searching, and establishing the support vector regression model of each category at the same time, a step of carrying out online positioning, collecting the RSS signal value of each hotspot of a target, using a support vector machine classification model to carry out classification, and obtaining the rough positioning area of the target, and a step of carrying out the accurate positioning of the target by using the support vector regression model according to the classification result. According to the method, the time-varying characteristic of the wireless signal intensity is effectively suppressed, and the precision is obviously improved.

The invention relates to the field of medical image processing and pattern recognition, and provides a computer-assisted lump detecting method based on a mammary gland magnetic resonance image. The computer-assisted lump detecting method based on the mammary gland magnetic resonance image aims at solving the problems that in the prior art, the lump partition effect is not good, and the accuracy, the sensitivity and the specificity in a classification test are not high. The computer-assisted lump detecting method includes the following steps: S1, extracting an interest area from the mammary gland magnetic resonance image; S2, extracting an initial lump area from the interest area in a separated mode, and determining the contour line of the initial lump area; S3, calculating the weight distribution of characteristic parameters of the initial lump area; S4, selecting the characteristic parameters, with the weight coefficients larger than a standard weight coefficient, of the initial lump area, and carrying out training classifying to obtain optimized characteristic parameters; S5, inputting the optimized characteristic parameters into a classifier, analyzing the optimized characteristic parameters with a support vector machine classification method, determining a final lump area, and displaying the final lump area for a user. The detecting method has the good lump partition effect, the accuracy, the sensitivity and the specificity in the classification test are effectively improved, the detecting result serves as a second opinion to be provided for a radiologist, and the misdiagnosis rate and the missed diagnosis rate of the radiologist can be effectively reduced.

The invention relates to an automobile hub temperature anomaly detection and early warning method and system. The system comprises a surveillance camera, a thermal infrared imager, a control host and a variable information board. The surveillance camera is installed above a road, the thermal infrared imager is installed on the road side, and the variable information board is arranged on the downstream of the surveillance camera. When a vehicle arrives at a detection area, the surveillance camera and the thermal infrared imager collect images of the vehicle and send the images to the control host; the control host carries out vehicle license plate recognition on the image shot by the surveillance camera; the control host extracts and fuses the SBDD, HOG and LBP features of a hub in the thermal imaging image and then judges whether the temperature of the hub is abnormal or not in a classified mode through an incremental support vector machine; when it is found that the temperature of the hub is abnormal, vehicle information is sent to the variable information board for early warning reminding. By means of the automobile hub temperature anomaly detection and early warning method and system, whether the temperature of the hub of an automobile is abnormal or not can be effectively detected, and early warning can be effectively carried out, so that traffic accidents caused by brake failure, fire breakout, tire burst and the like are avoided.

The invention relates to a water-injected meat detection method and water-injected meat detection device. The method comprises the steps that: water content data of a meat product is acquired according to a near infrared spectrum of the meat product; meat PH value is measured; meat image is collected, and meat color information is acquired according to the collected meat image; the meat product water content data, PH value, and color information are adopted as input characteristics of a support vector machine classification prediction model, and an output value of the support vector machine classification prediction model is obtained; and whether the meat product is water-injected meat is determined according to the output value from the support vector machine classification prediction model. With the method provided by the invention, water-injected meat can be rapidly and accurately detected.

The invention relates to a method for analyzing the service state of numerical control equipment, belonging to the technology for monitoring the service state of major equipment and forecasting the service life of the major equipment. The method comprises the following steps: firstly, recognizing the service states of a plurality of the major parts of the numerical control equipment through the characteristic physical quantity acquired by a multisensor; then, forecasting the service state of the complete machine through a classification model of a support vector machine established by the statistical learning theory; and calculating the residual service lives of the major parts and the complete machine through a 'hidden semi-Markov' random process model. The method of the invention not only can be used for recognizing the current running states of the parts but also can be used for forecasting the residual service life of the parts. The current running state and the residual service life of the complete machine are obtained by the classification forecasting method of the support vector machine according to the operation result of each part. The invention provides a new method for decision support of preventive maintenance.

The invention relates to an image processing technology and particularly relates to an interactive liver tumor region segmentation method based on watershed transform and classification through a support vector machine. The method comprises the following steps: 1) performing segmentation pretreatment on a liver tumor region; 2) performing watershed transform on an image of the pretreated liver region which is obtained in the step 1) and dividing the image of the pretreated liver region into numerous reception basins; 3) calculating four-dimensional characteristic vectors of all the reception basins which are generated by the watershed transform, marking tumor and normal liver tissues in the image of the liver region in an interactive manner and adopting a support vector machine process to classify the reception basins in a characteristic space; and 4) adopting communicating region detection to eliminate a false positive tumor region generated by the classification, and applying morphological operation to fill voids and smoothen edges. The region class is classified, and user marks are further utilized for training parameters of a classifier, thereby effectively improving the segmentation speed and the precision. The method has important application value in the fields of liver surgical planning and the like.

The invention discloses an SVM (support vector machine)-based nonlinear damage removing device of a coherent optical communication system. The device disclosed by the invention comprises a chromatic dispersion compensating unit, an SVM array unit, an SVM training unit, a logical processing unit and a sign deciding unit, wherein the chromatic dispersion compensating unit is used for compensating optical fiber dispersion applied to receiving signals; the SVM array unit is used for carrying out binary classification on the receiving signs according to different classification rules via a plurality of SVMs; the SVM training unit is used for defining a classification hyper plane of each SVM in the device according to a certain training sequence; the logical processing unit is used for performing logical operation on classification results of each SVM to acquire type signs corresponding to the signals; and the sign deciding unit is used for decoding the signal type signs to binary sequences corresponding to the signals. The SVM-based nonlinear damage removing device of the coherent optical communication system can be used for removing optical nonlinear damage to the signals by virtue of the nonlinear classification characteristic of the SVM. Without requirements of knowing the nonlinear characteristic of optical fibers, the nonlinear damage removing effect is ensured and the processing complexity is only decided by the number of a small quantity of support vectors.

The invention relates to a method and a device for classifying remote images by integrating space information and spectral information. The method includes the steps: performing pixel-wise support vector machine classification and area-based image segmentation for the remote images so as to obtain classified images and segmented images respectively; and combining the classified images with the segmented images, and performing majority voting so as to obtain final classified images. Since classified results of the remote images are obtained based on combination of the area-based image segmented results and the pixel-wise classified results, the method and the device are better in robustness, low in computing cost, easy to perform and capable of avoiding the problem of dimensional proportion selection. Besides, the defect that traditional partitional-clustering-segmentation-based classification of the hyperspectral images by integrating the space information and spectral information is non-robust is overcome.

The invention relates to a method and a system for graded measurement of voice. The method comprises the following steps: carrying out voice recognition for a received voice signal, and acquiring a voice feature sequence of state alignment according to a reference text and a reference model; correcting a distribution parameter of the reference model according to the voice feature sequence of state alignment, and generating a voice template vector based on the reference model for the voice signal; using a support vector machine classification decision tree to carry out classification decision for the voice template vector, and then, obtaining the classification grade mapped by the voice template vector. A support vector machine is provided by the invention to build a model for a language classification boundary, and the model is applied to language learning with the following steps: fetching the voice feature of the received voice signal, and carrying out state alignment between the received voice signal and the reference model; correcting the distribution parameter of the reference model and generating a corresponding voice template vector; and using the support vector machine classification decision tree to decide the voice template vector, thus, the complexity of the classification decision of voice is effectively reduced, and the accuracy of the graded measurement of voice is improved.

The invention discloses a multi-kernel support vector machine classification method for remote sensing images, and belongs to a support vector machine classification method for the remote sensing images. The method comprises the following steps of: performing principal component transform on original data; taking first four principal components to represent spectral information, performing wavelet texture feature extraction on the first principal component, and combining the spectral feature and spacial feature by adopting two independent radial basis functions; and finally performing classification by utilizing a multi-kernel support vector machine method. The wavelet texture feature and the spectral feature are combined thorough a plurality of basis functions, so the spectral feature extracted by principal component analysis is fully utilized, the wavelet texture feature is fused, the support vector machine is optimized, and the limitation that the traditional method separately adopts the spectral feature for classification is overcome; therefore, the classification accuracy is effectively improved. The method has the main advantage of improving the classification accuracy by combining the spectral information and the spacial information through the plurality of basis functions.

In accordance with various embodiments of the disclosed subject matter, a method and a system for detecting multiple objects from real-time images are provided. The method comprises: performing, using a CPU host, an image segmentation process to divide real-time input images into a plurality of image partitions; performing, by multiple GPUs, a fast block-wise registration process, a mark setting process, a background generation process, a foreground generation process based on a Hyper-Q computation infrastructure, and a support vector machine classification process; and generating, by the CPU host, visualization classification images.