47 results about "Sparse approximation" patented technology

A parallel dynamical system for computing sparse representations of data, i.e., where the data can be fully represented in terms of a small number of non-zero code elements, and for reconstructing compressively sensed images. The system is based on the principles of thresholding and local competition that solves a family of sparse approximation problems corresponding to various sparsity metrics. The system utilizes Locally Competitive Algorithms (LCAs), nodes in a population continually compete with neighboring units using (usually one-way) lateral inhibition to calculate coefficients representing an input in an over complete dictionary.

The invention aims to provide a polarization SAR image classification method based on an SSAE and an FSALS-SVM. According to the polarization SAR image classification method, a multi-implicit-strata structure of the stacked sparse automatic encoder (SSAE) is used for obtaining depth features which have the capacity for describing original data more intrinsically and are more suitable for classification, the fast sparse approximation least square-support vector machine (FSALS-SVM) which can obtain sparse solutions is used for replacing the Softmax commonly used in traditional deep learning and being combined with the SSAE, the classification accuracy of polarization SAR images is improved, the defect that a traditional polarization SAR image classification method based on pixels is greatly affected by speckle noise is overcome to a certain degree, and therefore coherence of homogeneous areas in classification result images is ensured.

A system and method are provided for taxonomically distinguishing grouped segments of signal data captured in unconstrained manner for a plurality of sources. The system comprises a vector unit constructing for each of the grouped signal data segments at least one vector of predetermined form. A sparse decomposition unit selectively executes in at least a training system mode a simultaneous sparse approximation upon a joint corpus of vectors for a plurality of signal segments of distinct sources. The sparse decomposition unit adaptively generates at least one sparse decomposition for each vector with respect to a representative set of decomposition atoms. A discriminant reduction unit executes during the training system mode to derive an optimal combination of atoms from the representative set. A classification unit executes in a classification system mode to discover for an input signal segment a degree of correlation relative to each of the distinct sources.

The invention provides a discriminant human motion identification method based on a Gaussian process latent variable model and a hidden condition random field. The method mainly comprises that skeletal structure and motion information of the human body are obtained via motion capturing technology or Kinect motion sensing technology when motion data is obtained; when motion characteristics are extracted, the Gaussian process latent variable model added with a dynamic process and sparse approximation is used to obtain structure of high-dimension motion information in low-dimension hidden space and further to represent the motion characteristic; and when the human motion is identified, the discriminant hidden condition random field is used to model characteristics of sequential motion data, and motions are classified. According to the invention, characteristics of the human motions can be visualized, information among the sequential motion data can be used effectively, the human motions can be identified in high precision, and the method is suitable for the field of real-time human motion identification.

The invention relates to a planetary gear box fault diagnosis method based on low-rank sparse approximation. The planetary gear box fault diagnosis method based on low-rank sparse approximation comprises the steps of: acquiring output shaft angle position signals by utilizing a built-in coder at first, then, fitting and differentiating the output shaft angle position signals by taking the mean square error as a loss function, so that instantaneous angle speed signals are obtained, on the basis of the instantaneous angle speed signals, respectively performing time-domain synchronous averaging on a planet wheel, a sun wheel and a gear ring, then, performing low-rank sparse approximation on respective time-domain synchronous averaging signals, so that sparse recovery signals of the planet wheel, the sun wheel and the gear ring are obtained, furthermore, calculating each kurtosis, and judging that the signal, the kurtosis of which is greater than a kurtosis threshold value, fails. According to the planetary gear box fault diagnosis method based on low-rank sparse approximation in the invention, by utilization of the characteristics of a low-rank sparse approximation method, acquired compound signals are decoupled; normal meshing information and fault information of a planetary gear box are separated; fault characteristics are prominent; and early weak fault diagnosis of the planetary gear box is easily realized.

A system and device for solving sparse algorithms using hardware solutions is described. The hardware solution can comprise one or more analog devices for providing fast, energy efficient solutions to small, medium, and large sparse approximation problems. The system can comprise sub-threshold current mode circuits on a Field Programmable Analog Array (FPAA) or on a custom analog chip. The system can comprise a plurality of floating gates for solving linear portions of a sparse signal. The system can also comprise one or more analog devices for solving non-linear portions of sparse signal.

The invention relates to a matrix-filtering-based sparse asymptotic minimum variance direction-of-arrival (DOA) estimation method. A matrix filter is used as a preprocessor to process an array receiving signal, so that a strong interference signal is suppressed and thus a phenomenon that a weak target is shielded by the strong interference signal or the positioning precision of the weak target bythe subsequent DOA estimation algorithm is affected by the strong interference signal is avoided. DOA estimation is carried out base on a sparse asymptotic minimum variance (SAMV) algorithm, so that the high-resolution performance is kept under the circumstances of small snapshot and low signal-to-noise ratio and a direction-of-arrival estimation problem of a coherent signal is solved. Meanwhile,the algorithm only needs to provide a threshold value eta of iterative stopping, thereby avoiding selection of regular parameters and enhancing the practicability of the algorithm. Besides, while thearray manifold change of the interference destroys the structure of the covariance matrix, the DOA estimation of the weak target signal still can be realized by the SAMV algorithm.

A heartrate monitor detects heartbeats in a test signal. A local heartrate and an energy of acceleration are associated with the detected heartbeats. Detected heartbeats are included or excluded from a test set of heartbeats based on the local heartrate and energy of acceleration associated with the respective heartbeats. Anomalous heartbeats in the test set of heartbeats are detected using a sparse approximation model. The heartrate monitor may detect heartbeats in a training heartbeat signal. A reference heart rate and an energy of acceleration are associated with detected beats of the training heartbeat signal and selectively included in a set of training data based on the heart rate and energy of acceleration associated with the detected beat in the training heartbeat signal. A dictionary of the sparse representation model may be generated using the set of training data.

The invention provides a sparse microwave imaging method based on online observation. The method comprises the following steps that S1, according to the radar observation imaging principle, a sparse microwave imaging module based on the sparse approximation theory is constructed; S2, a value estimated for the nth time of a backscatter coefficient of a target scene is set to be x(n); S3, a value x(n+1) estimated for the n+1th time of the backscatter coefficient of the target scene is obtained online by means of observation based on the homotopy algorithm through the backscatter coefficient x(n) of the target scene and newly-added radar observation echo data yn+1; the steps are repeated continuously until a stopping condition is satisfied, so that sparse microwave imaging is achieved. According to the sparse microwave imaging method based on online observation, the smallest sampling number can be determined in a self-adaptive mode according to the characteristics of the scene, an original result can be directly corrected by newly-added sampling data, and therefore recalculation is avoided and the calculated amount is reduced.

The invention provides a method and a device for MDCT domain signal energy and phase compensation. By analyzing the relationship between an MDCT transform radix and an MDST transform radix and by using the sparse approximation matrix, the transformation matrix of the MDCT to the MDST is extracted and simplified; the transformation of an MDCT spectrum to an MDST spectrum is realized by using the transformation matrix, and an MDFT spectrum with constant energy and linear phase is formed by combining the MDCT spectrum and the MDST spectrum; and the spatial parameters are extracted according to the MDCT spectrum. The invention effectively solves the problems of field wobbling and phase deletion of the signal in the MDCT domain and realizes the extraction and synthesis of the spatial parameters in the MDCT domain with low complexity in the spatial audio coding system.

The invention provides a sparse array method for vector hydrophones based on the compressed sensing theory. The method comprises the following steps of: (1) determining functions of the directivity ofa vector hydrophone combination and the directivity of a composite array; (2) constructing a target directional beam of a vector hydrophone array; (3) constructing an array problem model of a sparsevector hydrophone; and (4) reconstructing the vector hydrophone array sparsely. The method is based on the sparse array mechanism of the vector hydrophone array, combines with the convex optimizationand the compressed sensing theory, uses the sensing relationship between the sparse array position and the target array beam, and uses the non-uniform spacing array to reconstruct the target beam, toobtain accurate sparse array results. According to the sparse array method for the vector hydrophones based on the compressed sensing theory, the method can set the number of hydrophones in the desired sparse vector array independently according to the requirements of the system performance, and can control the complexity of the system accurately. The vector sparse array method adopts the orthogonal matching pursuit algorithm; the obtained sparse approximation result is fixed; and the method does not need to be debugged for multiple times, thereby saving a lot of algorithm adjustment time; andthe method is more aligned with the practical application of the engineering.

A signal which is to be quality-improved often suffers from the quality degradation in the spatial high frequency region more than compared to the spatial low frequency region. Accordingly a quality improvement is performed efficiently by combining the signal to be quality improved with a high frequency portion extracted from a sparse approximation of the signal to be quality improved.

The invention discloses a multi-agent centralized region coverage method and system based on global environment prediction. In the method, base stations receive respective main body positions and environment data, predict a global environment density function based on a sparse approximate Gaussian process regression method, and give a global prediction variance; the base station designs an autologous control scheme decision model based on the global prediction variance, assigns tasks to respective subjects through the decision model, designs a finite step prediction priority controller based on the global prediction variance by taking a decision index as a predicted autologous, designing a finite step length coverage priority controller based on an optimal region coverage Lloyd algorithm; and realizing the optimal coverage control through continuous interaction of the self-agent and the base station. According to the method, the base station-multi-autologous system can autonomously capture the environment density function model with high accuracy and high performance. Meanwhile, the multi-autologous system can quickly realize ideal optimal region coverage under the limitation of finite step length.

The invention belongs to the technical field of underwater acoustic equipment design and manufacturing and relates to a method for designing a strong interference suppression beamformer under a multi-target condition. The method comprises the following steps that: S1, a filtering matrix G belonging to CM*M is designed, the output y (t) of a filter is equal to Gx (t) has a spatial filtering characteristic, wherein t is equal to 1,..., N, and a matrix spatial filtering principle should satisfy one of conditions described in the descriptions of the invention; S2, a sparse super-resolution azimuth estimation algorithm is adopted; S3, spatial matrix filtering-based sparse super-resolution azimuth estimation is performed, namely, super-resolution azimuth estimation of a weak target in a strong interference environment can be realized by using spatial matrix filtering and a 1- type sparse approximate minimum variance algorithm; and S4, computer simulation and actual measurement test data verification of the algorithm are performed. The matrix filter with spatial filtering characteristics of the passband and the stop band is designed; a high-resolution sparse azimuth estimation algorithm is adopted, so that interference is effectively suppressed to the greatest extent in a signal processing process, the detection capability of a sonar on the weak target is improved, the output signal-to-noise ratio of the matrix is improved, and the remote detection capability of the sonar is enhanced.