52 results about "Sparse representation classifier" patented technology

The invention discloses a robust speech emotion recognition method based on compressive sensing. The recognition method includes generating a noisy emotion speech sample, establishing an acoustic feature extraction module, constructing a sparse representation classifier model, and outputting a speech emotion recognition result. The robust speech emotion recognition method has the advantages that effects of noise on emotion speech in the natural environment are fully considered, and the robust speech emotion recognition method under the noise background is provided; validity of feature parameters in different types is fully considered, extraction of feature parameters is extended to the Mel frequency cepstrum coefficient (MFCC) from prosodic and tone features, and anti-noise effects of the feature parameters are further improved; and the high-performance robust speech emotion recognition method based on the compressive sensing theory is provided through the sparse representation distinguishing in the compressive sensing theory.

The invention discloses a CFAR (Constant False Alarm Rate) and sparse representation-based high-resolution SAR (Synthetic Aperture Radar) image ship detection method, which mainly solves the problems of large data quantity to be processed and low real-time property existing in the conventional method. The method comprises the following detection steps: selecting a ship target training sample in a high-resolution SAR image and determining the size of a CFAR sliding window by the training sample; down-sampling the high-resolution image, performing image segmentation and land elimination on the high-resolution image, detecting in a low-resolution image by using the CFAR method and performing preliminary identification, and mapping a detected pixel point to a potential target region in the original high-resolution image; outputting potential target region slices obtained by all detection; and finally, extracting characteristic vectors of the potential target region slices respectively and identifying through a sparse representation classifier to obtain a final ship detection result. The CFAR and sparse representation-based high-resolution SAR image ship detection method has the advantages of high detection speed, high detection rate and low false alarm rate, and can be used for fishery supervision, maritime safety management and the like.

The invention discloses a super-pixel polarimetric SAR land feature classification method based on sparse representation. The method comprises: inputting polarimetric SAR image data to be classified, processing the image, and thereby obtaining a pseudocolor image corresponding to Pauli decomposition; performing super-pixel image over-segmentation on the pseudocolor image to obtain a plurality of super-pixels; extracting features, which are seven-dimensional, of radiation mechanism of the original polarimetric SAR image as features of every pixel; performing super-pixel united sparse representation to obtain sparse representation of each super-pixel feature; classifying by using a sparse representation classifier; working out the mean value of each super-pixel covariance matrix, then performing super-pixel complex Wishart iteration by using the classifying result in the last step, and at last obtaining a final classifying result. According to the super-pixel polarimetric SAR land feature classification method based on sparse representation, the problem that traditional classifying areas based on the single pixel are poor in consistency is solved, and operating speed of the algorithm is greatly increased on basis of improving accelerate.

The invention discloses a hyperspectral image classification and wave band selection method based on multi-target immune cloning. The hyperspectral image classification and wave band selection method based on multi-target immune cloning comprises the following steps that a sparse representation classifier is used for classifying hyperspectral remote sensing images so as to obtain classified images based on pixels; the hyperspectral remote sensing images are preprocessed by means of mean shift, the processed images are segmented into a plurality of clustering results through a multi-target immune cloning clustering and wave band selection algorithm, and the optimal clustering results are selected from the clustering results so as to constitute a segmentation image; the obtained classified images and the obtained segmentation image are fused by means of the maximum voting rule so as to obtain a final result image. The hyperspectral image classification and wave band selection method based on multi-target immune cloning has the advantages that the very high accuracy rate and Kappa coefficient values can be obtained under the circumstance that few training samples exist, region consistency is well kept, the hyperspectral image classification and wave band selection method is suitable for multiple hyperspectral data, and parameters are adjusted easily and conveniently.

The invention provides an SAR image search method based on sparse coding classification. The SAR image search method aims at the defects of an existing image search system and method. Through extraction of characteristics and construction of an over-complete dictionary, solution is carried out through sparse representation based on a dual Memetic algorithm, a sparse representation classifier is trained, the classification process with supervision can be achieved in the classification process, the sparse solution with global optimum can be solved fast, and then search results are output from high to low according to similarity. When the problem of image classification is solved, the method achieves the good improvement effect on classification accuracy, search content similarity, calculating complexity and result robustness.

The invention discloses an image classification method based on combination of SRC and MFA, which is mainly used for solving the problem that the image classification result is not ideal because only the reconstruction relationship or local discriminant structure is considered and the sample information cannot be accurately described in the conventional feature extraction method. The method comprises the following steps: 1. inputting a training sample and a test sample, constructing the same kind and different kinds of sample matrixes, and initializing a projection matrix; 2. projecting the training sample, respectively taking the same kind and different kinds of samples as dictionaries, solving sparse representation coefficients of the samples, and constructing the same kind and different kinds of sparse weight matrixes; 3. constructing an objective function to solve a novel projection matrix; 4. iterating the steps 2 and 3 until the cycle index is larger than the set initial value, outputting the final projection matrix, and projecting the test samples; and 5. classifying the test samples by utilizing a sparse representation classifier. According to the method disclosed by the invention, the accuracy of image classification is enhanced, and the method can be used for discriminating identity of characters or searching objects during image shooting in a police work system.

The invention discloses a facial emotion recognition method based on a local sparse representation classifier. The method is characterized by including: collecting facial emotion images; using Gabor wavelet transformation to construct the feature vectors of the facial emotion images; using a feature selection algorithm MFCS to select features; using the local sparse representation classifier to recognize emotion classification. Compared with the prior art, the method is high in emotion classification recognition accuracy, fast in emotion recognition, insensitive to faces, and the like.

The invention discloses a facial expression recognition method. According to the method, texture feature extraction is carried out on a facial expression image by using a multi-scale-parameter MB-LBPoperator and unified-mode pixel histogram statistics is carried out; HOG feature extraction is carried out on facial image samples; series-connection feature fusion is carried out on MB-LBP features and the HOG features in a same-kind mode; in fused feature space, a training sample is extracted randomly and the rest of samples are used as testing sample; PCA dimensionality reduction calculation iscarried out on the training sample to obtain a projection matrix W, the training sample is projected to low-dimensional sub space to obtain a feature expression of an expression image in the low-dimensional sub space; and the testing samples are projected to the low-dimensional sub space by the projection matrix W and the features of the testing samples are classified by using a sparse representation classifier to obtain the types of the testing samples. Therefore, an expression feature with texture and shape information is expressed by using a low dimension number, so that a high expressionrecognition rate is obtained; and the recognition accuracy is high.

The invention provides a pest image classification method based on context sensing dictionary learning. The method comprises the following steps that: context sensing information of pest images in the known category is added into a pest image sample base to obtain a plurality of types of training samples, a learning function is constructed, and the training samples are used for completing pest image redundant dictionary learning; the pest images to be classified are subjected to preprocessing to obtain test samples; the test samples are subjected to sparse representation dimensionality reduction processing; the test samples subjected to the sparse representation dimensionality reduction processing are read into a sparse representation classifier, and the residual error of the context sensing information of the test samples and various types of the training samples is calculated according to a redundant dictionary obtained through learning; and the residual error of the context sensing information of the test samples and various types of the training samples is analyzed, and the categories of the test samples are determined. The pest image classification method has the advantages that the precision and the efficiency of the pest image classification in complicated scenes can be improved, and a traditional crop pest diagnosis mode is improved.

The invention requests protection of a distributed compressed sensing data classification method based on a sparse representation classifier, relating to the field of wireless networks. Specific to that the characteristics of overload in data transmission and processing of sending nodes and data sparseness relevancy are not fully considered when the traditional classification method is applied to a sensor network with larger scale or more applications, a classification method suitable for relevant sparse data is designed, wherein a common part and a special part of data sparse coefficients are taken as classification basis so as to process sensor data reasonably. The data classification method provided by the invention is more accurate in data classification result of the sensor network and meanwhile has the advantages of effectively reducing network energy consumption by using a sparse coefficient strategy, improving the algorithm efficiency and the data processing performance, and facilitating user management as well as network application and scale expansion.

The invention discloses a pedestrian tracking method of a single camera. The method adopts a hierarchical tracking strategy. Firstly, detection targets are linked into stable and reliable short segments of movement tracks, and then a final tracking track is formed by matching the short segments of the movement tracks and filling blank segments between the short segments of the movement tracks. The hierarchical tracking strategy helps solve the problem of track interruption caused by short time of blocking of pedestrians in the tracking process. In the constructions of the short segments of the movement tracks, the method proposes a two-step matching method based on a cascading idea. Firstly, a high-efficient histogram matching method is used to deal with a condition of small matching difficulty, and then a robust sparse representation classifier is adopted to solve the aliasing problem caused by target apparent similarity. In the matching of the short segments of the movement tracks, a method of trajectory distance measurement based on sparse representation is proposed, which has the preferable robustness to the apparent change of target caused by environment change and target deformation. Furthermore, a track matching method, with small computation, based on hierarchical clustering, is also proposed, thereby improving the efficiency of the global multi track matching.

The invention discloses a BIT (built-in test) intermittent-fault diagnosis method based on sparse representation. The method specifically includes the following steps: (1) collecting different equipment history BIT data of which types include a normal data type, an intermittent-fault data type and a permanent-fault data type, selecting 70% of data from each data type to use the same as a trainingsample set, and removing labels of the remaining data to use the same as a test sample set; (2) utilizing a k-singular-value-decomposition (K-SVD) learning algorithm to carry out learning training onthe training sample set, and constructing an overcomplete dictionary D; (3) utilizing the overcomplete dictionary D to carry out sparse representation on the obtained test sample set to obtain sparsecoefficients; (4) utilizing a sparse representation classifier (SRC) to carry out sparse reconstruction on the sparse coefficients to obtain reconstruction residuals; and (5) categorizing to-be-testedsamples into categories of smallest reconstruction residuals, and thus obtaining diagnosis results of to-be-tested BIT data. The method can highly effectively and accurately diagnose BIT intermittent-faults, has higher diagnosis precision, and decreases a BIT false-alarm rate.

The invention discloses a spectrogram feature extraction algorithm based on kernel principal component analysis, and on the basis, a sparse representation classifier is applied to analyze an astragalus membranaceus spectrogram, so that a method for quickly identifying the production place of astragalus membranaceus is realized. The chemical analysis method adopted by the invention mainly comprisesan ion mobility spectrometry technology and an ultraviolet fluorescence spectrum analysis technology, has the advantages of rapid spectrum collection, good discrimination and the like, and is an important method for rapidly identifying the quality of the Chinese herbal medicine. According to the technical scheme, the method mainly comprises the steps of collection of astragalus membranaceus spectrogram information, extraction of Chinese herbal medicine spectrogram characteristics based on kernel principal component analysis and identification of the astragalus membranaceus production place based on a sparse representation classifier. Experiments show that the method is high in operation speed and high in recognition rate, provides a technical solution idea and scheme for rapid quality identification of Chinese herbal medicines, and has wide application significance.

The invention discloses a cell classifying method based on EMD feature extraction and sparse representation, and relates to an EMD-based cell feature extraction method. The method comprises steps: an orthogonal subspace projection (OSP) method is firstly used for carrying out band section on a medical hyperspectral image, dimensions are reduced, and data redundancy is reduced; and then, a two-dimensional EMD method is used for carrying out feature extraction on the data after dimension reduction, and the data are decomposed into a series of IMF components whose frequencies are arranged from high to low. A sparse representation classifier (SRC) is used for classifying the data, samples are classified through comparing residual, and when the residual obtained through calculation is small, the sample is divided into the class. In the cell classifying method based on EMD feature extraction and sparse representation, the EMD shows good time frequency characteristics, and an obvious potential and advantages exist in hyperspectral data feature extraction. Meanwhile, as the sparse representation classifier (SRC) is used, the classifying precision is more greatly ensured.

The invention discloses a facial expression recognition method based on MB-2DPCA features. The method comprises the following steps: carrying out local texture feature extraction on a facial expression image through a multi-scale parameter MB-LBP operator; carrying out feature extraction on expression features and constructing an MB-LBP feature space of the image; calculating a projection matrix W through a 2DPCA algorithm in training samples, and mapping all of the training samples to a subspace through the projection matrix W, and obtaining feature expression of the expression image in the subspace; and mapping a test sample to the subspace through the projection matrix W, and carrying out classification on the features of the test sample through a sparse representation classifier to obtain the category of the test sample. The method prevents the problem of large covariance matrix calculation dimensions due to conversion from an image matrix to the image vector, and the problem of small samples due to fewer samples; and the calculation time is less.

The invention discloses a 3D palmprint identification method. In the method, only the local features of blocks of a 3D palmprint, namely, a phase and surface type block histogram, are adopted. A 3D palmprint feature model is proposed. The identification effect is improved, and moreover, the use of a 2D palmprint image is avoided so that the whole system is free from the influence of light intensity and scratch marks. The local features of the 3D palmprint adopted are robust to small translation, rotation and even zooming of images, so that there is no need to use multiple translation, nearestiteration, cross-correlation and other methods for image alignment, and the identification efficiency is improved. An intermediate term is added in a sparse representation classifier, and an improvedsparse representation classifier is proposed by improving the sparse coefficient. The sparse coefficient can be calculated before 3D palmprint classification, and a subspace method is used to compareimages in the classification process. Thus, when there are a lot of samples in the training library, there is no need to compare testing samples and training samples one by one. The data redundancy, computation and processing time are reduced.

The invention relates to a design method of an elastic network constraint self-interpretation sparse representation classifier. The method comprises the following steps: training samples are read, the training samples are linearly transformed to a high-dimensional kernel space, each type of the training samples are learnt in the high-dimensional space, a contribution (i.e., a weight) made by each individual in the type of the training sample to constructing a sub-space of the type of the training samples is found, and a dictionary is constructed by a product of the type of the training samples and a weight matrix; and elastic network coefficient coding of test samples in the kernel space is obtained through training the obtained sparse representation dictionaries, and finally, the test samples are fitted by use of each type of the dictionaries and the elastic network sparse coding corresponding to the dictionaries, fitting errors are calculated, and the type of minimum fitting errors are the type of the test samples. According to the invention, the method is integrated with the advantages of ridge regression and lasso regression, sparse coding features of the samples are enabled to sparse, the fitting errors are also quite small, classification errors are effectively reduced, and the identification performance of a classifier is improved.

The invention provides a plant identification method based on elliptical Fourier descriptors and weighted sparse representation. The plant identification method is mainly and technically characterized by comprising the steps that leaf images are preprocessed, wherein all the colored leaf images are converted into grayscale images, the leaf images are separated from the background through an Otsu segmentation algorithm and converted into binary images, and small holes of the leaf images are eliminated through an erosion algorithm; edge detection is conducted by adopting a Canny edge detector; centroids of boundaries are calculated; the Fourier descriptors are calculated; a complete dictionary is constructured, wherein Fourier descriptor vectors of all leaf image data sets are divided into training sets and test sets, and the complete dictionary is composed of the Fourier descriptor vectors of all the training sets; and optimization is conducted by a weighted sparse representation classifier. According to the plant identification method, by adopting the elliptical Fourier descriptors, good robustness is achieved on noise and other factors, and by applying the weighted sparse representation classifier (WSRC) to plant species identification and particularly to a low-dimension space, the identification rate is obviously increased.