37 results about "Kernel Fisher discriminant analysis" patented technology

The invention relates to a method for extracting a curb of a road using a laser range finder and a method for localizing of a mobile robot using curb information of a road. The method for extracting the curb of the road using the laser range finder includes extracting a road surface and line segments from scan data of the laser range finder, extracting a plurality of curb candidate line segments among the line segments on the basis of an angle between the road surface and the line segment, extracting a plurality of curb candidates having a plurality of curb properties, wherein each of the plurality of curb candidates is generated by combining the couple of the curb candidate line segments, and applying the plurality of the curb candidates to a Kernel Fisher Discriminant Analysis to extract a final curb.

A image verification method, medium, and apparatus using a local binary pattern (LBP) discriminant technique. The verification method includes generating a kernel fisher discriminant analysis (KFDA) basis vector by using the LBP feature of an input image, obtaining a Chi square inner product by using the LBP feature of an image registered in advance and a kernel LBP feature and projecting to a KFDA basis vector, obtaining a Chi square inner product by using the LBP feature of a query image and a kernel LBP feature and projecting to a KFDA basis vector, and obtaining the similarity degree of the target image and the query image that are obtained as Chi square inner product results, and projected to the KFDA basis vector. According to the method, medium, and apparatus, the KFDA based LBP shows superior performance over conventional LBP, KFDA, and biometric experimentation environment (BEE) baseline algorithms.

The invention relates to a method for extracting a video object under a dynamic background based on fisher linear discriminant analysis. The method comprises the following steps of: dividing a K frame (namely a current frame) into a plurality of 8*8 blocks; obtaining a motion vector field of the K frame by performing block matching motion estimation on the K frame and a (K-1) frame; selecting probable background pixel blocks in the K frame as characteristic blocks; acquiring global motion model parameters from motion vectors of the characteristic blocks by a least square method; filtering exterior points by adopting Fisher linear discriminant analysis; performing global motion compensation for the current frame according to the global motion parameters; performing inter-frame difference between a rebuilt frame of the K frame and the (K-1) frame to extract a motion object. Tests prove that the extraction of the video object in a dynamic-background video sequence is realized, and the segmentation accuracy after compensation is obviously improved.

The invention provides a gait recognition method based on multi-linear mean component analysis. The gait recognition method comprises a training stage and a recognition stage. The training stage refers to performing dimension reduction processing on a half-cycle gait sequence image subjected to linear interpolation through the following algorithms, and training to obtain a projection conversion matrix of the algorithms: projecting the half-cycle gait sequence image into a low-dimension multi-linear subspace through general tensor discriminant analysis; performing further feature extraction by adopting the multi-linear mean component analysis; enabling a training tensor sample to be finally projected into a low-dimension vector space by adopting linear discriminant analysis. According to the recognition stage, conversion matrixes obtained through training the to-be-recognized half-cycle gait sequence image sample with the general tensor discriminant analysis and multi-linear mean component analysis algorithms are subjected to twice projection in a tensor space, the conversion matrix obtained through training the to-be-recognized half-cycle gait sequence image sample with the linear discriminant analysis algorithm is projected in the vector space, and a nearest neighbor classifier is adopted in the vector space during recognition. The gait recognition method can be used for improving the accuracy of gait recognition and has high robustness under different environments.

The invention discloses a building seismic damage analysis method, and relates to the field of seismic damage assessment. The method comprises the following steps of: setting a contour sampling interval, extracting a point cloud set of different height of a building, then, obtaining a contour section profile line through a convex hull algorithm, and converting three-dimensional point cloud of the building into a two-dimensional contour polygon sequence; extracting shape discrete parameters of the building; and calculating standard deviation of building shape parameters, and through a Fisher discriminant analysis, performing statistics on the standard deviation obtained in a step S22 or a step S23, thereby obtaining a discrimination function corresponding to three types of degree of damage of the building. According to the invention, edge extraction, shape clustering and discriminant analysis are fused, and problems in building contour polygon sequence extraction, shape discrete parameter calculation, irregular building block segmentation and damage analysis are solved effectively.

Visual tracking over a sequence of images is formulated by defining an object class and one or more background classes. The most discriminant features available in the images are then used to select a portion of each image as belonging to the object class. Fisher's linear discriminant method is used to project high-dimensional image data onto a lower-dimensional space, e.g., a line, and perform classification in the lower-dimensional space. The projection function is incrementally updated.

The embodiment of the invention discloses a lithofacies division method for the shale based on Fisher discriminant analysis. According to mineral compositions, a sedimentary structure and organic matter contents, mud rock lithofacies division is carried out to obtain a mud rock lithofacies division class result; a logging identification mode pattern of the mud rock lithofacies is established and awell logging curve class reflecting the lithology of the rock is selected based on the logging identification mode pattern; Fisher discriminant analysis is performed by SPSS software; on the basis ofcore logging data, a mud rock lithofacies class is determined and, the well logging curve reflecting the lithofacies characteristics is selected, and a lithofacies determination classification equation is established; according to a well logging curve of an unknown lithofacies, a function value of the lithofacies discriminant classification equation is obtained; and on the basis of the function value, a mud rock lithofacies division class to which the unknown lithofacies is determined. According to the invention, for the well section being lack of data like the core, logging, and diffractionanalysis, the mud rock lithofacies can be divided in details by using the lithofacies determination equation based on the well logging data, thereby improving the division precision and promoting theoil exploration.

The invention provides a gustatory induction signal variation feature extraction method based on kernel linear discriminant analysis. The method comprises the following steps: obtaining sensor response sequential signals by detecting tea samples by use of an electronic tongue; according to the response sequential signals, analyzing and rejecting abnormal samples by use of a main component residual error and Mahalanobis distance method; optimizing parameters of a kernel linear discriminant analysis method, and taking a Longjing tea quality grade correct recognition rate as a basis, selecting parameters of the kernel linear discriminant analysis; obtaining taste features of tea samples by performing nonlinear feature extraction on the sensor response signals by use of the kernel linear discriminant analysis method; and inputting the taste features of the tea samples into a classifier, and carrying out teat quality grade determination. According to the invention, abnormal value rejection is performed on the tea samples, nonlinear features of the tea samples with different grades can be represented better by use of the kernel linear discriminant analysis method after parameter optimization, and signal variation of the samples after nonlinear mapping in a high-dimensional feature space is improved.

The invention discloses a face identification method on the basis of duplex multi-kernel discriminant analysis of color features. A multi-kernel subspace learning technology is applied to duplex discriminant analysis of the color face features; three different nonlinear kernel mappings are respectively used for three color components and then feature layer duplex discriminant analysis is carried out. For features acquired by duplex multi-kernel discriminant analysis, a nearest neighbor classifier on the basis of the Euclidean distance measurement is used for classification and identification. According to the face identification method, an identification effect is better and after the features of three color namely R (red), G (green) and B (blue) components are subjected to duplex multi-kernel discriminant analysis, classification capacity of the identifying features is obviously reinforced.

The invention discloses an INS/GPS integrated navigation fault detection and diagnosis method based on an integrated neural network. The method provided by the invention comprises the following steps:firstly, obtaining INS and GPS fault data of integrated navigation through simulation; then, performing data processing by means of kernel Fisher discriminant analysis (KFDA), and achieving balance of mode data; then, carrying out One-Class SVM model training on the data, and detecting a fault condition of the nonlinear data by applying a nuclear skill; and finally, fusing the fault diagnosis results of the BP neural network and the improved dynamic particle swarm optimization (PSO) BP neural network by using a D-S evidence theory to realize fault detection and diagnosis of INS/GPS integratednavigation. According to the method, sudden change faults and slow change faults of the small sample integrated navigation system can be effectively detected, the fault detection rate reaches 90% orabove, the error rate of fault detection is effectively reduced, and an accurate comprehensive decision result is obtained.

The invention provides a hyperspectral image classification method based on wavelet kernel local Fisher discriminant analysis, and relates to the technical field of hyperspectral images. The method comprises the following steps: firstly, reading hyperspectral data as sample data, and normalizing a sample data set; mapping the data from a low-dimensional original space to a high-dimensional featurespace by adopting a wavelet kernel function; performing feature extraction on the sample data by using a local Fisher discriminant analysis method; dividing the data set after dimension reduction into training data and test data, and inputting the training data into an SVM classifier to obtain an optimal parameter value; inputting the test data into a classifier to obtain a classification result;and performing analysis and precision evaluation on a classification result. According to the hyperspectral image classification method based on wavelet kernel local Fisher discriminant analysis provided by the invention, a very good classification effect is obtained, and the method can be applied to the fields of agricultural monitoring, environmental management, disaster assessment, mineral mapping and the like.

The invention discloses a fault diagnosis method through combination of nested iteration Fisher discriminant analysis and relative change. Two types of faults-bias and data volatility often simultaneously exist in one type of fault data. According to the method, fault information is extracted through application of combination of nested iteration Fisher discriminant analysis and relative change analysis for the two types of faults of the same fault data, and then a reconstruction model is determined according to the fault information of each type of fault data and used for online fault diagnosis. According to the method, the disadvantages that the fault characteristics cannot be fully extracted by single method can be overcome, the performance of online fault diagnosis can be greatly enhanced and the faults can be accurately and rapidly restored through assistance so that process safety can be guaranteed and production benefit can be enhanced.

The invention discloses an exponent regularization and null space linearity discriminant analysis-based fault diagnosis method. Regularization discriminant analysis is combined with null space discriminant analysis, advantages of NSLDA and RLDA in terms of mode identification are integrated, a regularized intra-class sample matrix Sw1 is used for replacing a class sample matrix Sw in null space discriminant analysis, small sample problems can be further solved; in discrimination criteria for discriminant analysis, an exponential function is introduced, the regularized intra-class sample matrix Sw1 and an inter-class sample matrix Sb are respectively subjected to exponential operation, and therefore more characteristic information can be obtained; faults can be effectively and accurately identified, fault diagnosis precision can be effectively improved, and a new train of thought is put forward for small fault diagnosis based on data driving.

The invention discloses a human body behavior recognition method based on sparse tensors. The human body behavior recognition method comprises the following steps: obtaining a human body behavior silhouette sequence construction sample, and representing each sample with a three-order tensor; using sparse tensor local Fisher discriminant analysis to train a sparse projection matrix group; projecting the training sample and the to-be-tested sample to a tensor subspace by using the obtained sparse projection matrix group; and calculating the similarity between the to-be-tested sample and the training sample, and constructing a nearest neighbor classifier based on the tensor distance to identify the to-be-tested sample in the low-dimensional space. According to the tensor local Fisher discriminant analysis algorithm of the invention, the problems of eigenvalue solving and eigenvector solving are converted into a series of linear regression problems, so that the objective of tensor local Fisher discriminant analysis can be satisfied, and the sparsity of an obtained projection matrix can be ensured.

The invention discloses an SAR image target identification method based on enhanced kernel sparse representation, and the method comprises the steps: firstly, providing a multi-scale unicast feature extraction method which is used for simultaneously extracting the spatial and frequency domain information of an SAR image target; secondly, designing a classifier based on enhanced kernel sparse representation for target identification. The method is different from a traditional kernel sparse representation classifier. The designed classifier based on enhanced kernel sparse representation firstlyadopts kernel principal component analysis (Kernel Princture Analytics), and then the kernel Princture Analytics is used as the main component of the classifier. The method comprises the following steps of: calculating an enhanced pseudo-transformation matrix by using KPCA (kernel Fisher Discriminant Analytics) and KFDA (Kernel Fisher Discriminant Analytics); secondly, proposing a discriminative feature mapping method based on an enhanced pseudo-transformation matrix, and carrying out dimension reduction on the features in a kernel space; And finally, calculating a sparse coefficient by minimizing an L1 norm, and identifying the target category based on an error of sparse reconstruction. According to the method, the SAR target is identified based on the multi-scale unicast signal theory and the enhanced kernel sparse representation classifier, and a good classification and identification effect can be achieved.

An exemplary method includes receiving an image data set that comprises a multidimensional property space and data classifiable into data classes, determining a projection vector for data of the data set wherein the projection vector maximizes a ratio of between-class scatter to within-class scatter, selecting a reference for the vector, projecting at least some of the data onto the vector, measuring distances from the reference to at least some of the data, classifying at least some of the data into data classes based on a nesting analysis of the distances, eliminating the classified data from the image data set to produce a modified image data set and deciding whether to determine another projection vector for the modified image data set. Various other exemplary methods, devices, systems, etc, are also disclosed.

The invention discloses a urinary sediment detection method based on unbalanced local Fisher discriminant analysis. The urinary sediment detection method includes the steps: firstly, extracting aggregation channel features from an input urinary sediment visible component image; secondly, performing channel filtering on each channel by using a Haar-like template to extract an intermediate layer feature; then, grouping the features of a single channel, randomly selecting a plurality of groups of features to carry out linear weighting combination to form a new candidate feature; secondly, considering the imbalance of sample distribution, proposing an imbalance local Fisher discriminant analysis method to learn a weighting coefficient; and finally, connecting the candidate features of all thechannels in series to form a final feature vector, conducting training in combination with an Adaboost classifier based on a decision tree, and training different detectors for different urinary sediment visible components. According to the urinary sediment detection method, the local information fusion of the urinary sediment tangible image and the imbalance of sample distribution are considered,and the influence of noise is effectively reduced, and the accuracy is high, and the calculation speed is high, and the urinary sediment detection method has very important practical value.

The invention discloses a logistics enterprise customer classification method based on semi-supervised kernel Fisher discriminant analysis. The logistics enterprise customer classification method is characterized by comprising the following steps: (1) determining customer classification indexes and classification conditions commonly used by logistics enterprises; (2) collecting logistics enterprise customer information according to the customer classification indexes determined in the step (1); (3) standardizing the data sample set in the step (2); (4) constructing a consistency hypothesis matrix for the normalized customer sample data matrix obtained in the step (3), and calculating local inter-class and intra-class Laplace matrixes; (5) calculating a regularization term Laplacian matrixby utilizing the consistency hypothesis matrix obtained in the step (4), integrating the regularization term Laplacian matrix into a Fisher discriminant analysis target function, and obtaining an optimal projection matrix by solving a minimized target function; (6) calculating the projection coordinates of the normalized customer sample in the step (2) on the projection matrix; and (7) classifyingthe projection coordinates by using a nearest neighbor algorithm to determine the customer category. The logistics enterprise customer classification method is applied to classification of logisticsenterprise customers.