30 results about "Projections onto convex sets" patented technology

SENSitivity Encoding (SENSE) has demonstrated potential for significant scan time reduction using multiple receiver channels. SENSE reconstruction algorithms for non-uniformly sampled data proposed to date require relatively high computational demands. A Projection Onto Convex Sets (POCS)-based SENSE reconstruction method (POCSENSE) has been recently proposed as an efficient reconstruction technique in rectilinear sampling schemes. POCSENSE is an iterative algorithm with a few constraints imposed on the acquired data sets at each iteration. Although POCSENSE can be readily performed on rectilinearly acquired k-space data, it is difficult to apply to non-uniformly acquired k-space data. Iterative Next Neighbor re-Gridding (INNG) algorithm is a recently proposed new reconstruction method for non-uniformly sampled k-space data. The POCSENSE algorithm can be extended to non-rectilinear sampling schemes by using the INNG algorithm. The resulting algorithm (POCSENSINNG) is an efficient SENSE reconstruction algorithm for non-uniformly sampled k-space data, taking into account coil sensitivities.

The invention discloses a concentration detection precision correction method based on projection onto a convex set in an electron nose system. The method comprises the following steps: arranging a high precision standard gas sensor in an electron nose end sensor array, transmitting a sensor array detection signal to a server, carrying out class judgment and concentration calculation to obtain a concentration detection result, adjusting the response signals of all gas sensors to be corrected by using a method of projection onto a convex set with the concentration detection result, the temperature and the humidity of the standard gas by the high precision standard gas as references, and correcting the input value of a gas concentration calculation network to make the standard gas concentration value detected by the gas sensors to be corrected is close to the standard gas concentration value detected by the high precision standard gas sensor in order to determine the sensor correction coefficient and correct the concentration detection results of other gases to be detected. The method improves the concentration detection precision of an electron nose to a gas to be detected, and effectively solves the problems of difference and long-term drift of electron nose sensors.

The invention provides a method of simultaneously carrying out five-dimensional seismic data reconstruction and noise suppression. Firstly, in view of the defect of insufficient low-dimensional data reconstruction precision, Fourier transform is adopted as a sparse basis, five-dimensional seismic data information is used for data reconstruction, a projection onto convex sets algorithm is introduced during the process, a threshold parameter for an exponential square root decay law is provided, and a hard threshold operator is adopted to carry out independent reconstruction on each time slice; then, a weighting factor is introduced in the traditional projection onto convex sets algorithm, and thus, influences on the reconstruction result by noise during the reconstruction process are reduced; and finally, the method of simultaneously carrying out five-dimensional seismic data reconstruction and noise suppression can be realized. According to the method of simultaneously carrying out five-dimensional seismic data reconstruction and noise suppression provided by the invention, defects that the traditional two-dimensional or three-dimensional seismic data reconstruction precision is insufficient and the traditional method cannot simultaneously carry out data reconstruction and noise suppression are overcome, the reconstruction precision is greatly improved and requirements on memory of a computer are reduced.

According to a preferred aspect of the instant invention, there is provided herein a system and method for Interpolation of seismic data with a POCS (projection onto convex sets) algorithm that can produce high quality interpolation results, at a reduced computational cost. In particular, optimizing the threshold calculations, allowing for aliased data, allowing the algorithm to skip unneeded Fourier transforms, and parallelization makes the method more practical and robust. Robust interpolation of aliased data may be done by limiting the calculations of the beginning iterations to the lowest temporal and spatial frequencies included in the data, then gradually allowing higher temporal and spatial frequencies into the following iterations.

The invention provides a magnetic susceptibility inversion method and apparatus. The method comprises: all tensor magnetic gradient data are obtained; according to all tensor magnetic gradient data, a Tikhonov regularization model is established; with a CPU and GPU coordinated parallel way, iterative solving is carried out on the regularization model by using a conjugate gradient algorithm, and magnetic susceptibility is obtained by inversion. According to the invention, priori constraining is carried out on a model and data; a precondition type projection-onto-convex-set mixed conjugate gradient algorithm with rapid convergence is developed during the solving process; and the large-scale matrix calculation efficiency during the algorithm process is improved based on GPU and GPU combined calculation. The two-dimensional and three-dimensional fitting data examples demonstrate that the magnetic susceptibility result obtained by inversion of the all tensor magnetic gradient data is more accurate than the result obtained by TMI data calculation. With the method and apparatus, the high-precision magnetic susceptibility can be obtained by inversion; and thus a problem that the magnetic susceptibility obtained by inversion is not accurate in the prior art can be solved.

Disclosed herein is a video signal processing method and apparatus. In an embodiment of the present invention, video having a second screen size is generated from input video having a first screen size through Wavelet / Bicubic video interpolation, and a parallel Projection Onto Convex Sets (POCS) method is applied to the generated video having the second screen size. In this embodiment, the video having the second screen size, which is generated through Wavelet / Bicubic video interpolation, is used as an initial value for the POCS method, and a motion component estimated based on the video having the first screen size and the video having the second screen size obtained through the use of the POCS method is used as a correction value for the POCS method. Accordingly, HD-class video can be rapidly produced from SD-class video without image degradation.

The invention discloses a sampling method based on remote sensing images. The method comprises the steps of adopting projections onto convex sets POCS to perform up-sampling on remote sensing images, obtaining an intermediate result which can properly reconstruct a high-frequency detail part through superiorities of POCS, adopting a simple and effective prior art suitable for characteristics of the remote sensing images, setting Gaussian functions of a modulation transfer function MTF which presents anisotropy between the intermediate result and a real result, utilizing a Bayesian method to perform adaptive estimation on parameters, and utilizing the decomposition technology of a complex wavelet packet field for further deblurring and denoising, so as to obtain more accurate high-definition images. The method has good effects in terms of recovery and reconstruction of the high-frequency detail part and deblurring and denoising.

A similar image retrieving device (1) comprises: an image database (21) for storage of sets of image data, and sets of keywords each associated with a corresponding image data; a cluster classification section (11) to read the sets of image data, provide a respective one of the sets of image data with a compatibility value as an index representative of a set of compatibilities of a corresponding one of the sets of keywords, and classify the sets of image data into clusters thereof in accordance with the compatibility value; an optimum cluster extracting section (12) to provide the set of query image data with a compatibility value, and select one of clusters to which the query image data is to belong to minimize an error caused in a Projection onto Convex Sets using the clusters; and a similar image extracting section (13) to output, as a similar image, a set of image data provided a close compatibility value, among the sets of image data belonging to the cluster selected by the optimum cluster extractor.

Method for stabilizing the updated model (13) in iterative seismic data inversion so that the model-simulated data for the next iteration does not “blow up.” A Projection Onto Convex Sets (“POCS”) operator is defined that converts the matrix corresponding to the model to a positive semi-definite matrix. The stability projection operator may be looped with physical constraint projection operators until the model converges (15). The resulting stable and constrained model is then used to simulate seismic data in the next cycle of the outer iteration loop (16).

The present invention proposes a method for reconstructing three-dimensional irregular seismic data in the frequency domain, which is characterized in that firstly, Fourier transform is used to transform the three-dimensional seismic data in the time domain into the frequency domain, and then the idea of ​​convex set projection algorithm is adopted and the localization of seismic data can be described Curvelet transform of features; in the iterative process, a new threshold parameter that decays according to the exponential law is proposed, and a soft threshold operator is used to reconstruct each frequency slice separately, thereby reducing the number of iterations and improving the reconstruction accuracy, achieving three-dimensional reconstruction Purpose of Seismic Data. The present invention proposes a new threshold parameter that decays according to an exponential law, uses a soft threshold operator to reconstruct each frequency slice separately, overcomes the disadvantage of slow convergence speed of the traditional threshold parameter, reduces the computational complexity of the algorithm, and greatly The calculation efficiency is greatly improved and the calculation time is saved.

The present invention proposes a method capable of performing five-dimensional seismic data reconstruction and noise suppression at the same time. First, aiming at the lack of precision of low-dimensional data reconstruction, Fourier transform is used as a sparse basis, and data reconstruction is carried out by using five-dimensional seismic data information. In this process, the convex set projection algorithm is introduced, the threshold parameter of the exponential square root decay law is proposed, and the hard threshold operator is used to reconstruct each time slice separately, and then the weighting factor is introduced into the traditional convex set projection algorithm, so that in the reconstruction process The impact of noise on the reconstruction results was reduced, and finally a method capable of simultaneous 5D seismic data reconstruction and noise suppression was realized. The method for simultaneously performing five-dimensional seismic data reconstruction and noise suppression proposed by the present invention overcomes the shortcomings of traditional two-dimensional or three-dimensional seismic data reconstruction accuracy and the shortcomings of traditional methods that cannot simultaneously perform data reconstruction and noise suppression, and greatly improves reconstruction accuracy , reducing the requirements for computer memory.

The invention discloses a partial echo compressed sensing-based quick magnetic resonance imaging (MRI) method. The conventional imaging method has low speed and high hardware cost. The method comprises the following steps of: acquiring echo data of a random variable density part, namely intensively acquiring data in a central area of a k-space and acquiring the data around the k-space randomly and sparsely to generate a two-dimensional random mask, adding the two-dimensional random mask into every data point which needs to be acquired on a frequency coding shaft to form a three-dimensional random mask, and acquiring the data of the k-space according to the generated three-dimensional random mask; re-establishing by projection onto convex sets based on a wavelet domain which is de-noised by soft thresholding; and nonlinearly re-establishing a minimum L1 normal number based on finite difference transformation, namely sparsely transforming an image space signal x, determining an optimization objective and solving the optimization objective. By the method of the invention, partial echo technology and compressed sensing technology are combined and applied to data acquisition of MRI, sothat echo time is shortened, and data acquisition time is shortened at the same time.

The invention discloses a restraining method on edge Halo effects during the process of resetting a projections onto convex sets (POCS) super-resolution image, and belongs to the technical field of image processing techniques. The restraining method on the edge Halo effects during the process of resetting the POCS super-resolution image comprises a step of reading the sequence of a low-resolution image, a step of acquiring a high-resolution initial estimated value of convex sets by applying of a wavelet bicubic interpolation, a step of carrying out motion estimation to achieve low-resolution image registration by applying of a pre-filtering sub pixel iteration method, a step of determining a point spread function with edge-preserving characteristics by comprehensively considering space position information and grey information restraint of high-resolution estimation image pixels, a step of determining a length-variable loose projection parameter value according to relevance between all low-resolution observation frames and reference frames, and a step of resetting super-resolution by applying of a POCS method and combining of the high-resolution initial estimated value, the point spread function with the edge-preserving characteristics and the length-variable loose projection parameter value. Therefore, the image space resolution is improved and the ideal high-resolution image is obtained. By means of the restraining method, the edge Halo effects in the reset image can be reduced to a very large extent.

The invention discloses a remote sensing image upsampling method based on POCS (projections onto convex sets) and complex wavelet packet field, which comprises the following steps: S1) upsampling a remote sensing image by using the POCS, thereby acquiring an initial intermediate result image; S2) using a Bayes method to estimate a modulation transfer function (MTF) and a noise parameter between the intermediate result image acquired by using the POCS and a real image; and S3) according to the estimated MTF and noise parameter values, performing the de-blurring and noise-reducing operations onthe intermediate result image by adopting the complex wavelet packet transform, thereby acquiring a final upsampling result. By using the method provided by the invention, a clear high-resolution remote sensing image can be efficiently recovered from a single blur, noisy and low-resolution remote sensing image.

The invention relates to a regularized decoding method for image block compressed sensing, comprising the following steps: an image X is divided into small images, and the sub images obtained are spread into a column vector xj; a measurement matrix is selected to measure data to get a measured value yj; and an original signal is reconstructed with the measured value and the measurement matrix, and a reconstruction end introduces the total variation regularization item of the image to a block compressed sensing smooth projection decoding algorithm to carry out total variation gradient descent, threshold processing, and projection onto convex sets. The invention designs a regularized decoding method for block compressed sensing, which enables the decoded image to be robust and obtain a better reconstruction effect.

The invention provides a magnetic susceptibility inversion method and apparatus. The method comprises: all tensor magnetic gradient data are obtained; according to all tensor magnetic gradient data, a Tikhonov regularization model is established; with a CPU and GPU coordinated parallel way, iterative solving is carried out on the regularization model by using a conjugate gradient algorithm, and magnetic susceptibility is obtained by inversion. According to the invention, priori constraining is carried out on a model and data; a precondition type projection-onto-convex-set mixed conjugate gradient algorithm with rapid convergence is developed during the solving process; and the large-scale matrix calculation efficiency during the algorithm process is improved based on GPU and GPU combined calculation. The two-dimensional and three-dimensional fitting data examples demonstrate that the magnetic susceptibility result obtained by inversion of the all tensor magnetic gradient data is more accurate than the result obtained by TMI data calculation. With the method and apparatus, the high-precision magnetic susceptibility can be obtained by inversion; and thus a problem that the magnetic susceptibility obtained by inversion is not accurate in the prior art can be solved.