471 results about "Linear discriminant analysis" patented technology

An input signal is converted to a feature-space representation. The feature-space representation is projected onto a discriminant subspace using a linear discriminant analysis transform to enhance the separation of feature clusters. Dynamic programming is used to find global changes to derive optimal cluster boundaries. The cluster boundaries are used to identify the segments of the audio signal.

A face recognition apparatus and method using Principal Component Analysis (PCA) learning per subgroup, the face recognition apparatus includes: a learning unit which performs Principal Component Analysis (PCA) learning on each of a plurality of subgroups constituting a training data set, and then performs Linear Discriminant Analysis (LDA) learning on the training data set, thereby generating a PCA-based LDA (PCLDA) basis vector set of each subgroup; a feature vector extraction unit which projects a PCLDA basis vector set of each subgroup to an input image and extracts a feature vector set of the input image with respect to each subgroup; a feature vector storing unit which projects a PCLDA basis vector set of each subgroup to each of a plurality of face images to be registered, thereby generating a feature vector set of each registered image with respect to each subgroup, and storing the feature vector set in a database; and a similarity calculation unit which calculates a similarity between the input image and each registered image.

Feature attributes are extracted from an observation space to create feature vectors for each class to be identified. A linear transformation matrix is used to reduce the dimension of the feature vectors. A numerical optimization algorithm maximizes a geometric criterion function in order to calculate the linear transformation matrix, where it exploits the geometry of the class contours of constant density. Next, a classifier based on the feature vectors in a lower dimension is generated and a class is determined for the data represented.

A system and method for acoustic detection of coronary artery disease (CAD) are provided. The system includes a transducer for acoustically detecting heart signals of a patient and a computer system which executes detection software for processing the detected heart signals to identify the presence of CAD from the heart signals. The software allows for the automatic definition of a diastolic “window” of the acoustic signal for analysis, and automatically edits the sampled acoustic signal to eliminate unwanted artifacts and / or noise in the acoustic signal. The edited signal is then processed by a plurality of signal processing algorithms, including spectral analysis algorithms, time-frequency algorithms, global feature algorithms, kurtosis algorithms, mutual information algorithms, negenthropy algorithms, and principal component analysis algorithms, to generate a disease vector. The disease vector is then classified to determine whether CAD is present in the patient. Classification can be accomplished using linear discriminant analysis or a support vector machine.

The invention discloses an automatic vehicle body color recognition method of an intelligent vehicle monitoring system. The method comprises the following steps: firstly detecting a feature region on the behalf of a vehicle body color according to the position of a plate number and the textural features of the vehicle body; then, carrying out color space conversion and vector space quantization synthesis on pixels of the vehicle body feature region, and extracting normalization features of an obscure histogram Bin from the quantized vector space; carrying feature dimension reduction on the acquired high-dimension features by adopting an LDA (Linear Discriminant Analysis) method; carrying out various subspace analysis on the vehicle body color, then carrying out vehicle body color recognition of the subspaces by utilizing the recognition parameters of an offline training classifier, and adopting a multi-feature template matching or SVM (Space Vector Modulation) method; and finally, correcting color with easy intersection and low reliability according to the initial recognition reliability and color priori knowledge, so as to obtain the final vehicle body recognition result. The automatic vehicle body color recognition method is applicable to conditions of daylight, night and sunshine and is fast in recognition speed and high in recognition accuracy.

The invention discloses an optimized face recognition method and a device thereof, which improve the rate of facial image recognition. The technical proposal is that: the method and the device combine principal component analysis and linear discriminant analysis to solve the recognition problem; namely, the principal component analysis is carried out first before the conduction of the linear discriminant analysis so as to obtain relatively low dimensional space, then the space is utilized to carry out the linear discriminant analysis, and therefore, an intra-class dispersion matrix can not be caused and the process of the linear discrimination is effective. The invention firstly adopts the principal component analysis to obtain the best feature description, based on which, optimum identifying features are obtained by adopting the linear discriminant analysis, thereby greatly reducing the space dimension of facial features; finally, a minimum distance method is adopted to carry out classification and identification and therefore the rate of recognizing human faces is evidently increased. The method and the device of the invention are applied to face recognition.

An apparatus and method for human activity and facial expression modeling and recognition are based on feature extraction techniques from time sequential images. The human activity modeling includes determining principal components of depth and / or binary shape images of human activities extracted from video clips. Independent Component Analysis (ICA) representations are determined based on the principal components. Features are determined through Linear Discriminant Analysis (LDA) based on the ICA representations. A codebook is determined using vector quantization. Observation symbol sequences in the video clips are determined. And human activities are learned using the Hidden Markov Model (HMM) based on status transition and an observation matrix.

A non-invasive tool for skin disease diagnosis would be a useful clinical adjunct. The purpose of this study was to determine whether visible / near-infrared spectroscopy can be used to non-invasively characterize skin diseases. In-vivo visible- and near-infrared spectra (400-2500 nm) of skin neoplasms (actinic keratoses, basal cell carcinomata, banal common acquired melanocytic nevi, dysplastic melanocytic nevi, actinic lentigines and seborrheic keratoses) were collected by placing a fiber optic probe on the skin. Paired t-tests, repeated measures analysis of variance and linear discriminant analysis were used to determine whether significant spectral differences existed and whether spectra could be classified according to lesion type. Paired t-tests showed significant differences (p<0.05) between normal skin and skin lesions in several areas of the visible / near-infrared spectrum. In addition, significant differences were found between the lesion groups by analysis of variance. Linear discriminant analysis classified spectra from benign lesions compared to pre-malignant or malignant lesions with high accuracy. Visible / near-infrared spectroscopy is a promising non-invasive technique for the screening of skin diseases.

The invention discloses a speech emotion classifying method for emotion-based characteristic optimization, which comprises the following steps: (1), acquiring speech data of basic emotion states: a happy state, an angry state, an alarmed state and a sorrowful state; (2), extracting speed emotion characteristics; (3) matching emotion pairs; (4) compacting and selecting the characteristics: (4-1) reducing dimensions of linear discriminant analysis and performing respective LDA conversion by using the respective projection vector of each emotion pair, and (4-2) selecting a characteristic selecting method which is based on fisher discriminant criterion; and (5) fusing discriminations on the basis of two types of classifiers: (5-1) recording input emotion speech data, (5-2) outputting Ci and j according to the discriminations of the two types of classifiers, (5-3) obtaining the confidence coefficients wi and j of the two types of classifiers by using a formula (3), and (5-4) making a discrimination by using a correlation decoding method, wherein the discrimination result is the emotion type corresponding to the maximum correlation value.

The invention discloses a face recognition method based on PCA image reconstruction and LDA and belongs to the technical field of computer image processing and pattern recognition. The face recognition method is based on a principal component analysis algorithm; an intra-class covariance matrix serves as a generation matrix for acquiring face feature subspaces of individuals; then an image to be recognized maps the feature subspaces to extract features; the image is reconstructed according to feature values; a residual image is subjected to the linear discriminant analysis; and finally, face recognition is realized by a minimum distance classification recognition algorithm. Compared with the prior feature subspace method, the face recognition method can better extract the face features of different people, and the face recognition rate is increased effectively. In addition, when a face database is required to be expanded, only the newly added faces are required for feature face training; not all the face feature subspaces are retrained; and the face recognition method also has good extendibility.

The invention provides a voiceprint authentication processing method and apparatus. The method comprises steps of extracting the first characteristic vector of each voice in a training set by using a gender-mixing deep neural network (DNN) voiceprint baseline system; training a gender classifier according to the first characteristic vector of each voice and pre-marked gender labels; training DNN models of different genders according to the voice data of different genders in the training set; and training unified background models, characteristic vector extraction model, and probability linear discriminant analysis models of different genders according to the DNN models of different genders and the voice data of different genders in the training set. The method and the apparatus establish a voiceprint authentication processing model capable of distinguishing genders in order to increase voiceprint authentication efficiency and accuracy.

The invention discloses a method of improving performance of a convolutional neural network based on a linear discriminant analysis criterion, which belongs to the field of deep learning convolutional neural networks, the field of brain cognition and the field of computer vision image classification. The method comprises the following steps: 1) to-be-processed image sets are divided into a training set, a verification set and a test set; 2) a convolutional neural network model is selected; 3) one layer in the convolutional neural network model in the second step is selected, regularization constraints based on the linear discriminant analysis criterion are carried out on features of the selected layer, and a new convolutional neural network model is formed; and 4) according to mini-batch-based stochastic gradient descent method, the training set is used for training the new convolutional neural network model, and after the new convolutional neural network model is well trained, the well-trained convolutional neural network model tests to-be-classified images, and classified prediction is completed. Experimental results show that the method of the invention can significantly improve the convolutional neural network image classification precision.

The present invention relates to an iris recognition system and method using a multifocus image sequence, the iris recognition system including: an eye image input unit calculating the distance between an image obtaining apparatus and a user using a distance sensor to receive a multifocus eye image sequence having focus information of the user's eye varied at regular time intervals within a predetermined distance using the image obtaining apparatus; an eye image pretreatment unit calculating a fuzzy entropy for the multifocus eye image sequence input from the eye image input unit to select eye images of N number that can effectively express eye image data of various focusing levels, and extracting a multifocus iris image sequence by separating an iris image between pupil and collarette that has various iris patterns, from which an area unnecessary for configuring a feature vector is removed, from the selected eye images; a feature vector extraction unit multi-dividing the multifocus iris image sequence extracted in the eye image pretreatment unit, forming a wavelet feature vector from the multi-divided images using a subband including only a low-frequency area, and generating a wavelet feature vector reduced to a dimension having a high discrimination performance and a dimension-reducing transform function P; a recognition unit discriminating the user's authenticity by measuring a dissimilarity between the feature vector extracted from the feature vector extraction unit and a feature vector already registered; a register unit registering the reduced wavelet feature vector extracted from the feature vector extraction unit and the dimension-reducing transform function P in a storage unit; and a storage unit storing the dimension-reducing transform function P and the reduced wavelet feature vector. Accordingly, the iris recognition system and method using a multifocus image sequence can effectively recognize low-resolution images and poorly focused images as well as high-resolution images and well focused images. Moreover, the iris recognition system and method using a multifocus image sequence can extract the feature vector having a high discrimination performance, enhance the efficiency of a storage device by effectively reducing the dimension of the feature vector, and minimize recognition errors even using a simple recognition apparatus using a wavelet analysis and a direct linear discriminant analysis.

The invention relates to a multi-class motor imagery brain electrical signal classification method based on phase synchronization. According to the method, firstly, phase synchronization features of a training sample and a test sample are calculated respectively through a phase locking value; secondly, correlation coefficients of the training sample and the test sample are calculated and arrayed from large to small after an average value is removed and an absolute value is obtained; thirdly, brain electrical signals are roughly classified according to the arrayed correlation coefficients, and then disaggregated classification is conducted according to the brain electrical signals which are roughly classified, wherein the process is involved in a shared airspace mode feature extraction method and a linear discriminant analysis and classification method. The method comprises the steps of brain electrical signal collection, data pre-processing, filtering, calculation of the correlation coefficients of the phase synchronization features, feature extraction and classification and classification accuracy calculation. Classification results show that by the adoption of the brain electrical signal classification method based on phase synchronization, the classification results are good, the rough class where the test sample belongs can be efficiently and accurately determined through brain electrical signal rough classification based on phase synchronization, and the calculated amount is reduced.

The invention relates to a semi-supervised learning-based multi-gesture facial expression recognition algorithm which comprises the steps of acquiring n front expression images and n side expression images of n persons to form a training set X and a testing set S, segmenting face regions of the front expression images and the side expression images, and carrying out illumination compensation on the face regions by using a histogram equalization method; then extracting expression characteristics of the images by adopting a linear discriminant analysis method, carrying out expression recognition on samples in the testing set S; marking each unmarked sample in the training set X by using marked samples in the training set X by adopting an Euclidean distance nearest neighbour method; re-sampling the training set X by adopting a round-robin mode to obtain a new training set Xr; scheduling a basic classifying device to calculate a mark ht of each sample in the training set X at the tth circle by using the new training set Xr, and calculating a mark ft of each sample in the testing set S at the tth circle by using the new training set Xr; and finally, calculating a classifying error rate epsilon t of the basic classifying device to side samples in the training set, and updating weights of all training samples in the training set X until reaching the circle ending condition.

The invention discloses a health assessment and fault diagnosis method for rotating machinery based on fisher discriminant analysis and a mahalanobis distance, which belongs to the technical field of condition-based maintenance of the rotating machinery. The method comprises the steps of extracting an energy eigenvector based on wavelet packet decomposition, constructing a discriminant analysis function, conducting health status assessment, conducting fault detection on the rotating machinery, and finally conducting fault diagnosis on the rotating machinery. The method constructs a comprehensive frame integrating the status assessment, the fault detection and the fault diagnosis, solves the hotspot problem in comprehensive health management of the rotating machinery at present, achieves intelligent maintenance of the rotating machinery, can establish an assessment and diagnosis model without full life status monitoring data of the rotating machinery, reduces the dependence on historical data, and is very high in engineering applicability.

The invention relates to a road boundary detection and tracking method based on a three-dimensional laser radar. The method comprises the following steps: converting polar coordinate data of a laser radar scanning point into rectangular coordinate data; Extracting ground points by using a random sampling consistency plane fitting method and a principal component analysis method; Roughly extractingroad boundary feature points by adopting a multi-feature, wide-threshold and hierarchical feature point extraction method; Dividing the roughly extracted road boundary feature points into a left roadboundary feature point and a right road boundary feature point by using a projection-based method by referring to the idea of linear discriminant analysis; Respectively filtering the left road boundary point and the right road boundary point by using a random sampling consistency method to obtain final road boundary points; Fitting the left road boundary line and the right road boundary line by adopting a quadratic polynomial random sampling consistency method; And tracking the road boundary line by adopting a Kalman filtering method based on amplitude filtering. The method can be used for processing various complex scenes and working conditions, and can be widely applied to unmanned driving.

The invention discloses a method and a device for creating a sample library for large-scale face mode analysis. The method comprises the following steps: obtaining pictures in batches through a network; performing face detection and screening on the obtained pictures by adopting a Viola-Jones face detection algorithm; performing normalization processing on the pictures with faces based on a face ratio; performing gender classification on cut pictures by adopting PCA (principal component analysis) and LDA (linear discriminant analysis) algorithms; comparing similarity of grey histograms of the pictures to remove repeated pictures; performing secondary screening and artificial marking on the face sample library as needed and storing the face sample library into a local memory. The device comprises a network communication interface, display equipment, input equipment, a memory and a processor, wherein the processor is used for executing the method for creating the sample library for large-scale face mode analysis. The method and the device reduce the labor intensity of scientific research personnel during creation of the face sample library used for face mode analysis to a relatively great extent, increase the creation speed of the sample library, and also can be used for creating sample libraries for related scientific research of other mode identification.

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.

A variation image generation means generates a plurality of variation images having different postures, facial positions, and sizes with respect to a normalized image. A characteristic extraction means extracts a frequency characteristic from the plurality of variation images. A discriminant space projection means projects the frequency characteristic on a discriminant space having high discriminant ability that is obtained by linear discriminant analysis. A reference person comparison means performs a reference person comparison to extract a highly discriminant characteristic. A discriminant characteristic is extracted for a match image using the characteristic extraction means and the discriminant space projection means. A score computation means uses a discriminant axis obtained from a registered image, and the discriminant characteristic obtained from the match image to output a match score. A match determination means determines whether the person is the same person by comparing the match score with a threshold value.

The invention discloses a brain cognitive state judgment method based on polyteny principal component analysis (PCA). The method includes the following steps of firstly, inputting sample sets, and processing input data; secondly, calculating characteristic decomposition of training sample sets, determining an optimal feature transformation transformational matrix, and projecting training samples into tensor characteristic subspace to obtain feature tensor sets of the training sets; thirdly, vectorizing lower dimension feature tensor data which are subjected to dimensionality reduction as input of linear discriminant analysis (LDA), determining an LDA optimal projection matrix, and projecting the vectorized lower dimension feature tensor data into LDA feature subspace for further extracting discriminant feature vectors of the training sets; and fourthly, classifying features, subjecting the discriminant feature vectors obtained by projection of training images and test images to feature matching, and further classifying the features . According to the brain cognitive state judgment method, PCA is utilized to directly perform dimensionality reduction and feature extraction to multi-level tensor data, the defect that structures and correlation of original image data are destroyed and redundancy and structures in the original images can not be completely maintained due to the fact that traditional PCA simply performs dimensionality reduction is overcome, and space structure information of functional magnetic resonance image (fMRI) imaging data is kept.

The invention discloses face authentication method and device, and belongs to the field of biological recognition. The method comprises the following steps: sequentially extracting multiple levels of feature vectors from a to-be-authenticated face image and a face image template with a multi-level depth convolutional network which is subjected to multi-layer classified network joint training in advance; sequentially mapping the multiple levels of feature vectors into unified dimension feature vectors through a unified dimension linear mapping matrix; connecting the unified dimension feature vectors into joint feature vectors in series; carrying out dimension-reducing mapping on the joint feature vectors through a linear dimension-reducing mapping matrix; and normalizing cosine values with absolute values through linear discriminant analysis, and carrying out comparison and authentication on the obtained comprehensive feature vectors of the to-be-authenticated face image and comprehensive feature vectors of the face image template. Compared with the prior art, the face authentication method disclosed by the invention is high in anti-jamming capability, good in expandability and high in authentication accuracy rate.

The system and method for detecting seizure activity combines signal traces from both an electroencephalogram (EEG) and an electrocardiogram (ECG) in order to detect and predict a seizure event in a patient. Determination of a seizure classification of the combination is based on Dempster-Shafer Theory (DST) to calculate a combined probability belief. Prior to combination, classification of the EEG and ECG data is performed by linear discriminant analysis (LDA) or naïve Bayesian classification to provide a seizure event classification or a non-seizure event classification.

Disclosed is a method for speech speaker recognition of a speech speaker recognition apparatus, the method including detecting effective speech data from input speech; extracting an acoustic feature from the speech data; generating an acoustic feature transformation matrix from the speech data according to each of Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), mixing each of the acoustic feature transformation matrixes to construct a hybrid acoustic feature transformation matrix, and multiplying the matrix representing the acoustic feature with the hybrid acoustic feature transformation matrix to generate a final feature vector; and generating a speaker model from the final feature vector, comparing a pre-stored universal speaker model with the generated speaker model to identify the speaker, and verifying the identified speaker.

A method, system, and computer program product for detecting at least one nodule in a medical image of a subject, including identifying, in the medical image, an anatomical region corresponding to at least a portion of an organ of interest; filtering the medical image to obtain a difference image; detecting, in the difference image, a first plurality of nodule candidates within the anatomical region; calculating respective nodule feature values of the first plurality of nodule candidates based on pixel values of at least one of the medical image and the difference image; removing false positive nodule candidates from the first plurality of nodule candidates based on the respective nodule feature values to obtain a second plurality of nodule candidates; and determining the at least one nodule by classifying each of the second plurality of nodule candidates as a nodule or a non-nodule based on at least one of the pixel values and the respective nodule feature values. True-positive nodules are identified using linear discriminant analysis and / or a Multi-MTANN.

The invention discloses a voiceprint model training method, a voice recognition method, devices, equipment and a medium. The voiceprint model training method includes the following steps: using training voice data in at least two business scenarios to carry out universal background model training to obtain a universal background model; using the universal background model to carry out feature extraction on intra-domain voice data and out-of-domain voice data to obtain corresponding intra-domain voiceprint features and out-of-domain voiceprint features; carrying out probability linear discriminant analysis model training based on the out-of-domain voiceprint features to obtain an out-of-domain probability linear discriminant analysis model; and adaptively training the out-of-domain probability linear discriminant analysis model based on the intra-domain voiceprint features to obtain a target probability linear discriminant analysis model. The voiceprint model training method overcomes the problem that it is impossible to obtain multiple pieces of voice data of a single person in some business scenarios, and ensures the recognition effect of the target probability linear discriminantanalysis model.

The invention provides a face authentication method and device, and belongs to the field of image processing and pattern recognition. The method comprises the following steps: extracting Gabor texture features of an acquired face image sample pair through a Gabor filter, and separating real parts from virtual parts of the extracted Gabor texture features to obtain a plurality of feature graphs; summarizing the plurality of obtained feature graphs through an LBP (Local Binary Pattern) statistical histogram to serve as face feature expression of the face image pair; performing feature selection on the face feature expression of the face image sample pair through an AdaBoost (Adaptive Boosting) algorithm; and performing feature dimension reduction, feature judgment processing and the like on the face feature expression of the face image sample pair subjected to the feature selection through a PCA (Principal Components Analysis) algorithm and an LDA (Linear Discriminant Analysis) algorithm in sequence. Compared with the prior art, the face authentication method provided by the invention has the advantages of full extraction of sample texture information, low sample quantity demand, short algorithm time and low space complexity.

The invention relates to a hand action identification method based on surface electromyography (sEMG) decomposition. The method comprises the following two parts including sEMG decomposition and a hand action identification method based on MUAPT (motor unit action potential train), wherein the sEMG decomposition is carried out to obtain the MUAPT and consists of sEMG signal preprocessing, sEMG peak detection and Gaussian mixture model (GMM) clustering; and the hand action identification based on the MUAPT comprises characteristic extraction, principal component analysis (PCA) dimensionality reduction, LDA (Linear Discriminant Analysis) classification and the like. Under a condition that only one channel sEMG is used, motor unit action potential information obtained by the sEMG signal decomposition is used for identifying hand action, an identification rate of single-channel sEMG is effectively improved, and the hand action identification method has important theoretical significance and practical application value.

Methods and apparatus are provided for generating a decision trees using linear discriminant analysis and implementing such a decision tree in the classification (also referred to as categorization) of data. The data is preferably in the form of multidimensional objects, e.g., data records including feature variables and class variables in a decision tree generation mode, and data records including only feature variables in a decision tree traversal mode. Such an inventive approach, for example, creates more effective supervised classification systems. In general, the present invention comprises splitting a decision tree, recursively, such that the greatest amount of separation among the class values of the training data is achieved. This is accomplished by finding effective combinations of variables in order to recursively split the training data and create the decision tree. The decision tree is then used to classify input testing data.

The invention provides a gait recognition method based on a partitioned matrix. Firstly, extracting single-frame images from a video, then carrying out grey scale transformation on the single-frame images, using the background subtraction method to extract person body targets, using mathematical morphology to fill the holes of binary images, and extracting profiles of the person by means of single connection analysis so that the person bodies are positioned in the middle and are uniformly in the size of 64 * 64 pixels; observing the periodic change of the gait according to elliptical short axis and eccentricity fitted in image regions after the standard centralization of each frame image in a gait video sequence; using a gait energy diagram to extract the integral characteristic of the gait in the one period, dividing GEI into sub-blocks by means of the partitioned matrix, eliminating the sub-blocks which are useless to classification in a self-adapting manner, and adopting the method, which combines the two-dimensional principal component analysis of a sub-block mode with the two-dimensional linear discriminant analysis, to further extract local characteristics; and integrating the characteristics of each effective sub-block into a whole during the classification recognition, and adopting a nearest neighbor classifier to perform identification judgment. The method is effective for the recognition of the gait of knapsack change.