225 results about "Sparse model" patented technology

A method for accurately estimating a pose of the human head in natural scenes utilizing positions of the prominent facial features relative to the position of the head. A high-dimensional, randomly sparse representation of a human face, using a simplified facial feature model transforms a raw face image into sets of vectors representing the fits of the face to a random, sparse set of model configurations. The transformation collects salient features of the face image which are useful to estimate the pose, while suppressing irrelevant variations of face appearance. The relation between the sparse representation of the pose is learned using Support Vector Regression (SVR). The sparse representation, combined with the SVR learning is then used to estimate a pose of facial images.

The present invention discloses an estimation method of a quasi-stationary broadband array signal direction of arrival (DOA) based on block sparse Bayesian learning. An intra-frame correlation and aninterframe independence of a quasi-stationary broadband signal frequency spectrum are employed to set a corresponding a block sparse prior distribution model for signals, and a block sparse Bayesian model is employed to perform estimation of sparse signals, so that an estimation result with higher precision is obtained. Array receiving signals are subjected to appropriate framing processing, eachframe of the signal is subjected to Fourier transform, and each block sparse Bayesian model for each signal is established in a frequency domain; under an assumption of each frame of the signal is independent, information of all the frames is combined to establish a total Bayesian model, and hyper-parameter vectors are employed to control a sparsity of all the frames of signals to be reconstructed; and finally, the expectation maximization algorithm (EM) is employed to obtain an iterative update formula of the hyper-parameter vectors. The estimation method of a quasi-stationary broadband arraysignal direction of arrival based on block sparse Bayesian learning fully utilizes a short-time stability feature of quasi-stationary broadband array signals to establish a block sparse model, and therefore a higher DOA estimation precision can be obtained.

The invention discloses a fast sparse Bayesian learning based direction-of-arrival estimation method and mainly solves the problems of heavy computation and large location estimation error in the prior art. The method includes the implementation steps: (1) adopting antenna receivers to form a uniform linear array; (2) sampling space signals and computing an array covariance matrix R; (3) vectorizing R to obtain a sparse model vector y; (4) dividing space domain grids, and constructing an over-complete base phi(theta) according to the structure of the sparse model vector y; (5) establishing a sparse equation according to the sparse representation relation of the sparse model vector and the over-complete base; (6) defining a hyper-parameter vector alpha, and adopting a fast sparse Bayesian learning algorithm to solve the sparse equation; (7) drawing a magnitude spectrogram according to an optimal estimation value of alpha to obtain a direction-of-arrival value. By the method, estimation accuracy of target reconnaissance and passive location under the conditions of low signal to noise ratio and small snapshot number is improved, computational complexity is lowed, and the method can be applied to target reconnaissance and passive location.

The invention discloses an information recommendation method and system combining image content and keywords. The information recommendation method combining image content and keywords comprises the steps that keyword information of images in an image library and image content information containing color features and textural features are extracted, the keyword information and the image content information are expressed as a vector space model, and a corresponding keyword information matrix and an image content information matrix are obtained; the keyword information matrix and the image content information matrix are processed by utilizing a linear sparse model, and a similarity chart is obtained by calculating the similarity among the images; an image similar to a target image is inquired from the similarity chart according to the target image searched by a user, and an original recommendation list is formed; the original recommendation list is arranged to obtain a final recommendation list, and the final recommendation list is displayed.

The invention discloses a linear frequency modulation radar signal processing method based on compressed sensing. The method comprises the steps of (1) emitting a linear frequency modulation signal to a radar and preprocessing an echo signal, which means that the deramping processing of the echo signal is carried out, a difference frequency signal is outputted, and a signal model of deramping processing is established in a time domain, (2) according to the sparsity of the difference frequency signal in a frequency domain, constructing a sparse conversion matrix, and establishing a sparse representation model of the radar echo signal, (3) constructing a measurement matrix, and realizing the projection transformation of a difference frequency sparse signal to a low dimensional space, and (4) using an orthogonal matching pursuit (OMP) algorithm, reconstructing a radar difference frequency signal, and efficiently obtaining target information. Accoding to the method, the compression of radar echo signal data can be fundamentally realized, the change of a sparse model according to a radar observation distance is not needed, finally the target information is obtained, and the method is suitable for the echo signal processing of an actual radar.

The invention discloses a structure sparse representation-based remote sensing image fusion method. An adaptive weight coefficient calculation model is used for solving a luminance component of a multi-spectral image, similar image blocks are combined into a structure group, a structure group sparse model is used for solving structure group dictionaries and group sparse coefficients for the luminance component and a panchromatic image, an absolute value maximum rule is applied to partial replacement of the sparse coefficients of the panchromatic image, new sparse coefficients are generated, the group dictionary and the new sparse coefficients of the panchromatic image are used for reconstructing a high-spatial resolution luminance image, and finally, a universal component replacement model is used for fusion to acquire a high-resolution multi-spectral image. The method of the invention introduces the structure group sparse representation in the remote sensing image fusion method, overcomes the limitation that the typical sparse representation fusion method only considers a single image block, and compared with the typical sparse representation method, the method of the invention has excellent spectral preservation and spatial resolution improvement performance, and greatly shortens the dictionary training time during the remote sensing image fusion process.

The invention discloses a priori shape modeling method based on a combined sparse model, and belongs to the technical field of medical image segmentation. Through priori shape modeling, a shape library obtained by collecting clinical data is established in allusion to a specific tissue organ, wherein the shape library consists of segmented shapes of image data from different patients, so as to establish a golden standard for corresponding organs of the patients. The priori shape modeling method based on a combined sparse model comprises the following steps: 1, gridding the surfaces of the shapes in advance by sampling points on the golden standard surfaces, wherein gridded shapes in the shape library are taken as training data of a model; 2, representing the gridded shapes by a sparse shape combining model, wherein coordinates of each shape in the shape library corresponding to all vertexes of a grid are arrayed as a column vector and an array D is obtained from the whole shape library; 3, performing optimization algorithm on the sparse shape combining model to obtain a corresponding parameter; and 4, performing inverse transformation on the corresponding parameter obtained in the optimization algorithm to obtain a required priori shape.

A hearing assistance system comprises an input unit for providing electric input sound signals u i , each representing sound signals U; from a multitude n u of sound sources S i , an electroencephalography (EEG) system for recording activity of the auditory system of the user's brain and providing a multitude ny of EEG signals yj , and a source selection processing unit receiving said electric input sound signals ui and said EEG signals yj , and in dependence thereof configured to provide a source selection signal Sx indicative of the sound source Sx that the user currently pays attention to using a selective algorithm that determines a sparse model to select the most relevant EEG electrodes and time intervals based on minimizing a cost function measuring the correlation between the individual sound sources and the EEG signals, and to determine the source selection signal Sx based on the cost functions obtained for said multitude of sound sources.

The invention discloses a compressive sensing radar imaging algorithm based on subspace tracking. By using a physical optical method, corresponding echo data are calculated, radar echo data are analyzed, a sparse model of a signal is established, and echo reconstruction is realized by using an effective and steady subspace tracking algorithm, so that synthetic aperture radar imaging is realized. The signal is subjected to non-adaptive sampling by using a compressive sensing theory at the speed rate which is much lower than Nyquist sampling rate. Therefore, the compressive sensing theory is applicable to synthetic aperture radar imaging, an imaging effect can be achieved, and the aim of reducing the sampling rate of the echo data is fulfilled, so that simulation time is shortened, and the cost of a radar system can be reduced.

The invention relates to a nonlinear un-mixing method of hyperspectral images based on kernel sparse nonnegative matrix decomposition. The nonlinear un-mixing method comprises the following steps: estimating the number of end members for hyperspectral images by utilizing a dimension virtual method; then, popularizing the conventional un-mixing algorithm based on a linear mixing model to a nonlinear characteristic space by utilizing a kernel method, and solving a nonlinear spectrum un-mixing problem by using an alternative iterative optimization method. The nonlinear un-mixing method of hyperspectral images based on kernel sparse nonnegative matrix decomposition has the beneficial effects that from a mixing model of hyperspectral observation pixels, the sparsity of the hyperspectral abundance is added into a sparse model, and the linear mixing model is mapped into a nonlinear mixing model by virtue of the kernel method, so that the defects of linear un-mixing are effectively overcome, and good noise resistances are simultaneously achieved, and therefore, the nonlinear un-mixing method can be used as an effective means for solving the un-mixing of hyperspectral remote sensing images.

The invention discloses a flexible deep learning network model compression method based on channel gradient pruning, and the method comprises the steps: 1, adding a masking layer constraint to an original network, and obtaining a to-be-pruned deep convolutional neural network model; wherein the absolute value of the product of the channel gradient and the weight serves as an importance standard toupdate the masking layer constraint of the channel to obtain a mask and a sparse model, 3, carrying out pruning operation on the sparse model based on the mask, and 4, retraining a compact deep convolutional neural network model. The invention further provides an application effect of the flexible deep learning network model compression method based on channel gradient pruning on an actual objectrecognition APP, the recognition speed of the model to the object after pruning is greatly improved, and the problem that the deep neural network model cannot be applied to the actual object recognition APP due to high storage space occupation and high memory occupation, high computing resources are occupied and cannot be deployed to embedded devices, smart phones and other devices are solved, and the application range of the deep neural network is expanded.

The invention discloses a blurred image blind restoration method based on a Gaussian scale mixed type Markov expert field. The method comprises the implementation steps that (1) modeling is carried out on noise, a restored image and a restored blurred kernel through a Gaussian model, the Gaussian scale mixed type Markov expert field and a sparse model based on an l1 norm respectively in a Bayes posterior probability model; (2) a Napierian logarithm is extracted from the obtained Bayes posterior probability model to obtain a problem to be optimized; (3) the restored image and the restored blurred kernel are initialized through a blurred image and a Gaussian blurred kernel respectively, and a maximum number of iterations is set; (4) in a certain iteration, the obtained restored blurred kernel is fixedly optimized, and the restored image is optimized; (5) the obtained restored image is fixedly optimized, and the restored blurred kernel is optimized; (6) if the number of iterations is smaller than the maximum number of iterations, the step (4) and the step (5) are repeatedly executed; (7) a regularization coefficient in the step (4) is adjusted, and the known blurred image is restored through the final restored blurred kernel obtained in the step (6). According to the method, the high-quality restored image can be obtained through a single blurred image.

The invention discloses an image fusion method based on joint sparse model; the image fusion method comprises: using a learned over-complete dictionary to subject source images to joint sparse representation into a common sparse portion and respective private sparse portions; fusing the two types of sparse coefficients obtained through designed fusion rules so as to obtain a fused sparse coefficient; reconstructing an image with the fused sparse coefficient and the dictionary. Simulation experiment results show that the image fusion method provides improved fusion effect for infrared and visible light images.

The invention discloses a method and device for reconstructing a dynamic magnetic resonance image and a readable storage medium, and relates to the technical field of image processing. The method forreconstructing the dynamic magnetic resonance image comprises the steps of obtaining to-be-reconstructed sampling data in the dynamic magnetic resonance image; Decomposing the dynamic magnetic resonance image into a background low-rank tensor component and a changed foreground sparse tensor component according to the local change characteristic of the dynamic magnetic resonance image; Processing the sampling data through a preset low-rank model to obtain a background low-rank tensor component; Processing the sampling data through a preset sparse model to obtain a changed foreground sparse tensor component; Based on the background low-rank tensor component and the changed foreground sparse tensor component, the dynamic magnetic resonance image is reconstructed, and the data structure correlation of the dynamic magnetic resonance image data in the time dimension and the space dimension can be fully utilized.

The invention belongs to the technical field of airborne radar space time adaptation processing, and particularly relates to an airborne radar space time adaptation processing method based on environment dynamic perception. The airborne radar space time adaptation processing method based on the environment dynamic perception includes concrete steps: setting a work mode of an airborne radar to be an MIMO (multiple input multiple output) mode, using a receiving array to receive a time domain return signal Y, and representing a clutter scattering coefficient vector in an airborne radar observation area as gamma; marking the position of a jth clutter block of an ith distance unit in the airborne radar observation area as Aij; building a sparse model, and representing a basis matrix corresponding to observation data Y after discretization as H; obtaining vector estimation of the clutter scattering coefficient vector gamma of the airborne radar observation area by solving the sparse model, setting the work mode of the airborne radar to be a phased array mode so as to obtain the distance r'i between the clutter block on the position Aij and the airborne radar and an corresponding return signal arrival angle; obtaining a clutter covariance matrix of units with distance to be detected, and performing space time adaptation processing on a return signal which is received when the airborne radar works under the phased array mode.

The invention provides a compressed sensing synthetic aperture radar radio frequency interference suppression handling method. The method includes the following steps: (1) building a compressed sensing model of a radio frequency interference (RFI) signal based on an obvious sparse feature of a RFI component in frequency domain, (2) estimating sparseness of the RFI component by using a greedy algorithm to combine with minimum description length, (3) estimating a RFI signal component and filtering directly in time domain to an echo signal of each pulse, and (4) achieving imaging process by using an conventional compressed sensing synthetic aperture radar (SAR) reconstructing algorithm. The compressed sensing synthetic aperture radar radio frequency interference suppression handling method is based on the sparse feature of the RFI component in the frequency domain and builds a RFI sparse model based on a compressed sensing theory, restrains narrow-band and broadband RFI signals effectively by using the iterative greedy algorithm to estimate and filter the RFI component in the echo signal.