709results about How to "Improve reconstruction accuracy" patented technology

A system and method for processing of audio and speech signals is disclosed, which provide compatibility over a range of communication devices operating at different sampling frequencies and / or bit rates. The analyzer of the system divides the input signal in different portions, at least one of which carries information sufficient to provide intelligible reconstruction of the input signal. The analyzer also encodes separate information about other portions of the signal in an embedded manner, so that a smooth transition can be achieved from low bit-rate to high bit-rate applications. Accordingly, communication devices operating at different sampling rates and / or bit-rates can extract corresponding information from the output bit stream of the analyzer. In the present invention embedded information generally relates to separate parameters of the input signal, or to additional resolution in the transmission of original signal parameters. Non-linear techniques for enhancing the overall performance of the system are also disclosed. Also disclosed is a novel method of improving the quantization of signal parameters. In a specific embodiment the input signal is processed in two or more modes dependent on the state of the signal in a frame. When the signal is determined to be in a transition state, the encoder provides phase information about N sinusoids, which the decoder end uses to improve the quality of the output signal at low bit rates.

A system and method for processing of audio and speech signals is disclosed, which provide compatibility over a range of communication devices operating at different sampling frequencies and / or bit rates. The analyzer of the system divides the input signal in different portions, at least one of which carries information sufficient to provide intelligible reconstruction of the input signal. The analyzer also encodes separate information about other portions of the signal in an embedded manner, so that a smooth transition can be achieved from low bit-rate to high bit-rate applications. Accordingly, communication devices operating at different sampling rates and / or bit-rates can extract corresponding information from the output bit stream of the analyzer. In the present invention embedded information generally relates to separate parameters of the input signal, or to additional resolution in the transmission of original signal parameters. Non-linear techniques for enhancing the overall performance of the system are also disclosed. Also disclosed is a novel method of improving the quantization of signal parameters. In a specific embodiment the input signal is processed in two or more modes dependent on the state of the signal in a frame. When the signal is determined to be in a transition state, the encoder provides phase information about N sinusoids, which the decoder end uses to improve the quality of the output signal at low bit rates.

The invention belongs to crossing field of computer vision and intelligent robots and discloses a variational mechanism-based large-area indoor scene reconstruction method. The method comprises the following steps: step 1, acquiring calibration parameters of a camera, and building an aberration correcting model; step 2, building a camera position and gesture depiction and camera projection model; step 3, utilizing an SFM-based monocular SFM (Space Frequency Modulation) algorithm to realize camera position and gesture estimation; step 4, building a variational mechanism-based depth map estimation model, and performing solving on the model; and step 5, building a key frame selection mechanism to realize three-dimensional scene renewal. According to the invention, an RGB (Red Green Blue) camera is adopted to acquire environmental data, and a variational mechanism-based depth map generation method is proposed through utilizing a high-precision monocular positioning algorithm, so that quick large-area indoor three-dimensional scene reconstruction is realized, and problems of three-dimensional reconstruction algorithm cost and real-time performance are effectively solved.

The invention discloses a three-dimensional reconstruction method based on coding structured light, comprising the following steps: 1) projecting structured light to an object to be measured, and capturing an image modulated by the object to be measured by a camera; 2) matching an optical template, comprising: (2.1) positioning the optical strip boundary, scanning along each row of the image, determining a pixel point with strong gray variation as a candidate marginal point, and searching a local domain; and (2.2) matching the optical strip: adopting a color cluster method to build a color matching proper vector, comparing image color with a projected color, and defining Euclidean distance between the color proper vector and the cluster center to distribute the colors of red, green, blue and white of the candidate optical strip; and 3) using a calibrated system parameter for three-dimensional reconstruction of the object to be measured, determining the relation between a space point coordinate and the image coordinate point thereof by the calibrated conversion matrix parameter; and restoring three-dimensional spatial coordinate from the image coordinate of a feature point. The invention can simplify calculation process and has high matching precision and high reconstruction precision.

A computed tomography imaging scanner (10) acquires helical cone beam projection data for a volume of interest (46) using at least two source trajectory helices. A reconstruction processor (62) reconstructs the projection data for each helix to generate a corresponding time skewed image representation. A voxel time processor (66) computes an acquisition time for each voxel in each time skewed image representation. A voxel interpolator (68) computes an interpolated voxel value for each voxel based on values of the voxel in the time skewed image representations and corresponding voxel acquisition times. In an electronic embodiment, the computed tomography scanner (10) includes an x-ray source (12) with an axially oriented cylindrical anode (92), an electron source (961, 962) irradiating the cylindrical anode (92) to produce an x-ray beam (120, 122, 124, 126), and an electron beam deflector (98, 100) that deflects the electron beam along the anode (92) to axially sweep the x-ray beam.

A method of reconstructing a 3D model includes reconstructing a 3D voxel-based visual hull model using input images of an object captured by a multi view camera; converting the 3D voxel-based visual hull model into a mesh model; and generating a result of view-dependent rendering of a 3D model by performing the view-dependent texture mapping on the mesh model obtained through the conversion. Further, the reconstructing includes defining a 3D voxel space to be reconstructed; and excluding voxels not belonging to the object from the defined 3D voxel space.

The invention discloses a method for three-dimensionally reestablishing a surface contour of a human body, and the method is characterized by comprising the following steps: (1) the projection region of a protector, the shooting region of a camera and the measured part of the human body are corresponding; (2) the projector projects one image once, the camera collects one image, and the projected image comprises black and white images, Gray code images and phase shift grating images; and (3) the collected pictures are processed to obtain the three-dimensional information of the surface contour of the human body. In the invention, a phase error compensation curve is designed and the reestablishment accuracy of the surface contour of the human body is enhanced according to the characteristics of the projector and the time characteristics in the projecting and shooting courses; a common period calibration method is improved and used for calibrating the absolute phase of the surface contour of the human body, thereby enhancing the reestablishment accuracy of the surface contour of the human body; and an independent projected image generating module is adopted and used for generating the projected images and controlling the projected images of the projector, thereby facilitating the control on the time interval between the projected images of the projector and the collected images of the camera and enhancing the stability of phase error fitted functions.

The present invention discloses a three-dimensional reconstruction method and system capable of maintaining sharp features. The method includes the following steps that: 1) a two-dimensional image filtering method extended to a three-dimensional space is adopted to perform smoothing de-noising on inputted roughly-registered point cloud; 2) an improved region growth method is adopted to perform outlier removal on the smoothed roughly-registered point cloud; 3) a kd-tree (k-dimensional tree) acceleration-based ICP (iterative closest point) algorithm is adopted to perform precise registration on the outlier-removed roughly-registered point cloud; 4) a neighborhood search and boundary point detection-based fusion method is adopted to fuse the precisely-registered point cloud; and 5) a feature point detection and adaptive step size update-based method is adopted to perform surface reconstruction on the fused precisely-registered point cloud. The three-dimensional reconstruction system is composed of a point cloud preprocessing module, a point cloud combining module and a surface reconstruction module. The three-dimensional reconstruction system which is realized based on the method of the invention can maintain the sharp features of the edge of a reconstructed model, and therefore, reconstruction speed is considered with accuracy ensured.

The invention discloses a power line three-dimensional reconstructing method based on airborne laser radar point cloud. Data of the airborne laser radar point cloud have been correctly classified. The method includes the following steps: 1, loading the correctly-classified data of the airborne laser radar point cloud and initial line track data of an overhead transmission line; 2, extracting accurate information of the positions and the number of electric towers and the track of the power transmission line, and determining the total number of spans of the power transmission line; 3, determining the two-dimensional space range and power line laser radar points of each span of the power transmission line; 4, clustering the determined power line laser radar points of the spans; 5, carrying out power line three-dimensional reconstruction. According to the power line three-dimensional reconstructing method, the number of required parameters is small, the automation degree is high, robustness and universality are better, and the power line three-dimensional reconstructing method has the advantage of being insensitive to the factors such as the number of power lines, the types of the power lines, the power line space arrangement structure, gross error points, point-cloud irregular breaking and the line length; in addition, a reconstruction model has the high reconstruction accuracy.

The present invention proposes an evaluation method of solar energy utilization potential in urban high-density areas based on low-altitude photogrammetry, comprising the following steps: low-altitude data acquisition using drones, acquisition of three-dimensional point cloud data, screening of building subject points, selection of seed region, region growth and point cloud patch segmentation, three-dimensional reconstruction, building geometry model, solar radiation simulation and solar energy utilization potential assessment steps. The invention utilizes low-altitude photogrammetry to obtain point cloud data in a high-density area of the city, and performs semantic division and three-dimensional reconstruction on the point cloud data through a parameterization tool to obtain a singular geometric model. Therefore, the solar radiation and dynamic change obtained from the roof and facade of the building are simulated and analyzed in the same platform.

The invention discloses a photoacoustic and fluorescence dual-mode integrated tomography imaging system and a photoacoustic and fluorescence dual-mode integrated tomography imaging method. The system comprises a modulation laser light source (1), a vibrating mirror (2), a sample rotary table (3), a filtering wheel (4), a gating intensified charge-coupled device (ICCD) camera (5), a photoacoustic detection acquisition module (6), a system control module (7), a computer (8), a data bus (9) and a control bus (10), wherein the vibrating mirror, the sample rotary table, the filtering wheel and the gating ICCD camera are positioned on the same straight line with the direction of lasers; the ultrasound transduction array normal direction of the photoacoustic detection acquisition module is perpendicular to the straight line to form a crossed vertical detection mode. According to the imaging method, light absorption distribution formed by photoacoustic tomography imaging is used as prior information for the reestablishment of fluorescence tomography imaging, and the positions of fluorophores, fluorescent yield and fluorescence lifetime parameters are reestablished by an iterative method. By the photoacoustic and fluorescence dual-mode integrated tomography imaging system and the photoacoustic and fluorescence dual-mode integrated tomography imaging method, the imaging complexity can be reduced, the cost of hardware can be saved, and the flexibility of the system and the reestablishing accuracy of the fluorescence tomography imaging can be improved.

The invention provides a three-dimensional framework fast extraction method based on branch feathers, which comprises the following steps: firstly, automatically extracting root points and tip end points of a tree type voxel model; then, carrying out seed point distance conversion on the root points and the tip end points of the voxel model; automatically judging the branching feathers and optimizing cutting surfaces according to the growth strategy of regions from the root points or the tip end points; decomposing an object into meaningful components with reasonable delimitation between components; and finally, extracting voxel component frameworks and connecting the voxel component frameworks into a structural framework of the object. The framework fast extraction method based on branch feathers is fast and effectively, and the structural integral frameworks maintain topological structures of the original voxel model, and can not generate fracture and redundant complicated branches. The actual plant model construction is carried out on the basis of the framework, the actual scanning data can be processed, and the invention has the anti-noise capability and has high reconstruction accuracy. The test on a plurality of data sets proves that the invention is applicable to bodies with annular structures and can process surface voxel models and solid voxel models.

The invention provides a method for realizing a semantic SLAM system based on a mobile terminal and relates to fusion of construction of an SLAM system and point cloud semantic analysis. The method comprises the following steps: 1) solving a camera pose based on an improved FAST and LDB characteristic point method; 2) carrying out back-end optimization to obtain an accurate point cloud position and an accurate camera pose; 3) carrying out loopback detection; 4) constructing a global point cloud map; 5) realizing an augmented reality system by utilizing the SLAM system; 6) realizing 3D point cloud semantic segmentation by utilizing semantic segmentation; and 7) optimizing the semantic SLAM system based on the mobile terminal. Aiming at environmental cognition in flexible indoor and outdoorscenes and augmented reality demands, especially in the fields of unmanned driving, path planning and the like, a solution low in cost and good in performance is obtained, and the method has extensiveapplication value and expansion capability.

The invention discloses a method for iterating and reconstructing a double-energy-spectrum CT image. The method is used for reconstructing a base material density image of a measured object and comprises the following steps of (1) selecting a calibration model body containing tow types of different base materials according to the size of a material of the measured object; measuring the model body through a double-energy-spectrum CT system, and calibrating a double-energy-spectrum multi-color orthographic projection equation presenting a curvilinear relationship between a multi-color projection value and base material thickness combination; (2) collecting double-energy-spectrum multi-color projection data of the measured object by utilizing the double-energy-spectrum CT system; (3) constructing a iterative scheme according to the calibrated multi-color orthographic projection equation, and iterating and reconstructing the double-base-material density image of the measured object through comparison with the practically measured multi-color projection data and addition of a physical constraint condition. Compared with the existing method, the method for iterating and reconstructing the double-energy-spectrum CT image has the advantages that the high energy spectrum and the low energy spectrum of the double-energy-spectrum CT system and an attenuation coefficient of the base materials do not need to be detected in advance, and meanwhile the method is applied to the situation that the high energy spectrum and the low energy spectrum are matched with each other or not in geometry.

The invention discloses a distributed microphon array sound source positioning method based on space sparsity, comprising steps of performing signal characteristic extraction through two steps of discrete cosine transform (discrete cosine transform DCT), utilizing the characteristics to construct a sparsity positioning model in order to comprehensively utilize the short time or long time characteristics of the voice signal, reducing the dimension of the model, utilizing a dictionary learning method to perform correction on the model, adopting an approximating I0 norm reconstruction algorithm to perform reconstruction on a sparse signal to find a sound source. The invention has no special requirement on the array structure, requires less calculation, effectively improves the capability of resisting noise and reverberation and positioning accuracy, and is suitable for indoor sound positioning and widely applicable to the video conference system, the voice recognition system and the intelligent robots.

Provided is a magnetic resonance imaging (MRI) apparatus including an acquisition unit configured to acquire an undersampled spectrum in a k-space and a reconstruction unit configured to generate a target image based on the undersampled spectrum, wherein the reconstruction unit includes: a first sub-reconstruction unit configured to perform initial reconstruction on data corresponding to unsampled positions in the k-space by using a Split Bregman algorithm or approximate sparse coding; a second sub-reconstruction unit configured to decompose the initially reconstructed spectrum in the k-space into multiple frequency bands to thereby generate a plurality of individual spectra and perform dictionary learning reconstruction on images respectively corresponding to the decomposed multiple frequency bands by alternating sparse approximation and reconstructing of measured frequencies; and an image generator configured to generate a target image by merging together the reconstructed images respectively corresponding to the multiple frequency bands.

The invention discloses a large-scale MIMO time-varying channel state information compression feedback and reconstruction method. The method comprises the steps that a channel matrix sequence is acquired, T channel matrixes are subjected to DFT separately, and a channel matrix sequence which is sparse in the angle delay domain is obtained; a channel feedback and reconstruction model CsiNet-LSTM isconstructed, the channel matrix sequence is input to a coder, and codewords are output; the codewords are sequentially input to a decoder, and a reconstructed channel matrix sequence is output; the channel feedback and reconstruction model is trained to gradually approximate to the channel matrix sequence to obtain model parameters; each channel matrix in the output reconstructed channel matrix sequence is subjected to two-dimensional inverse DFT, and a reconstruction value of an original space-frequency domain matrix sequence is obtained through recovery; and channel state information to befed back and reconstructed is input to the model, and a reconstruction value is output. According to the method, the feedback overhead of large-scale MIMO channel information can be reduced, the reconstruction precision is improved, and the excellent robustness is particularly achieved on decrease of the compression ratio.

The invention provides a high-precision three-dimensional reconstruction method and system based on binary surface structured light. The method comprises the steps of calibrating a CCD high-definitioncamera and a projector, projecting directional light source emitted by an LED lamp onto a scene where a target object is located, acquiring an image through the CCD high-definition camera, matching acontour approximate to the target object through a self-adaptive identification module, carrying out structured light information coding according to the result recognized by the self-adaptive identification module, projecting the structured light to the approximate position where the plane of the target contour is located by the projector, the CCD high-definition camera acquiring a modulated image projected on a target object, combining the structural light characteristic points with the gradient information obtained by the LED light contour to obtain a final three-dimensional image. The three-dimensional reconstruction system has the beneficial effects that the adaptability of the three-dimensional reconstruction system to environmental changes can be improved, the reconstruction precision and accuracy of the target object are effectively improved, and the problem that the operation cost is increased due to the multi-objective environment is solved.

A 4D reconstruction method of the vascular axis of coronary artery, which belongs to the field of medical detection technology, comprises following steps: constructing the projection models of an X-ray angiography system at two angles according to X-ray coronary artery angiogram sequences at two angles, deducing the geometric transformation relationship between the two angle images, 3D reconstructing the sampled points of interested blood vessels selected by hand in the image of the first moment, connecting the reconstructed points to obtain a broken line, acquiring the 3D axis of the blood vessel at the first moment by snake transformation with respect to the broken line as the initial position, and acquiring 3D vascular axis in each subsequent moment in the image sequence by snake transformation with respect to the 3D vascular axis of the previous moment as the initial position, thus completing the 4D reconstruction of the vascular axis in the entire sequence. The method can greatly reduce the uncertainty and the error due to the operation of operators, thus improving the repeatability of the result and achieving convenient operation and high efficiency.

The invention discloses a multiple observed value vector sparsity self-adaptive compressed sampling matching pursuit method, which relates to the technical field of information and communication. The multiple observed value vector sparsity self-adaptive compressed sampling matching pursuit method is provided for solving the problem of recovering an original multiband signal from multiple observed value vectors with unknown sparsity after continuous-limited module conversion through sampling by a modulated broadband converter under an Xampling framework. The multiple observed value vector sparsity self-adaptive compressed sampling matching pursuit method comprises the steps of: conducting self-adaptive estimation on sparsity of a signal; updating the sparsity with a given step length factor through repeated iteration so that the sparsity gradually approaches the actual sparsity of the signal; correcting a support set through a backtracking thought and a minimum mean square criterion; stopping iteration until an residual error is less than a set threshold value; and finally reconstructing an original multiband signal through pseudo inverse operation by utilizing the obtained complete support set. The multiple observed value vector sparsity self-adaptive compressed sampling matching pursuit method can achieve the analog reconstruction of the multiband signal based on compressed sensing.

The invention discloses an indoor three-dimensional model reconstruction method and system with structural semantics, and the method comprises the steps: collecting indoor three-dimensional point cloud data through a mobile depth scanning device, and carrying out the preprocessing of the three-dimensional point cloud data; extracting a wall surface component of the functional space; obtaining a function space boundary; door and window parts are extracted and parameterized; carrying out plane segmentation on the global point cloud data to obtain a global plane, and optimizing the geometric andincidence relation of the wall surface part and the door and window part according to a plane structure global optimization algorithm; and outputting the parameterized data to an indoor three-dimensional model with structural semantics. According to the method, based on the mobile depth scanning equipment, the low-precision point cloud and the camera track subjected to semantic marking are fused for rapid three-dimensional reconstruction of the indoor model with structural semantics, and better robustness and reconstruction precision are achieved.

A method and apparatus for spatio-temporal compressive sensing, which allows accurate reconstruction of missing values in any digital information represented in matrix or tensor form, is disclosed. The method of embodiments comprises three main components: (i) a method for finding sparse, low-rank approximations of the data of interest that account for spatial and temporal properties of the data, (ii) a method for finding a refined approximation that better satisfies the measurement constraints while staying close to the low-rank approximations obtained by SRMF, and (iii) a method for combining global and local interpolation. The approach of embodiments also provides methods to perform common data analysis tasks, such as tomography, prediction, and anomaly detection, in a unified fashion.

The invention brings forward a frequency domain three-dimensional irregular earthquake data reconstruction method. The method is characterized in that first of all, three-dimensional earthquake data in a time domain is converted to a frequency domain by use of Fourier transform, and then, a projection onto convex set (POCS) algorithm is employed and curvelet transform capable of describing localized features of the earthquake data is introduced; and in an iteration process, a new threshold parameter attenuating according to an index rule is brought forward, and each frequency slice is individually reconstructed by use of a soft threshold operator, such that the iteration frequency is reduced, the reconstruction precision is improved, and the purpose of reconstructing the three-dimensional earthquake data is realized. According to the invention, the new threshold parameter attenuating according to the index rule is brought forward and each frequency slice is reconstructed in individually by use of the soft threshold operator, such that the disadvantage of quite slow convergence speed of a conventional threshold parameter is overcome, the calculation complexity of an algorithm is reduced, the calculation efficiency is substantially improved, and the operation time is reduced.

The invention discloses a quantization and inverse quantization method in large-scale MIMO channel state information feedback. The method comprises the following steps: firstly, acquiring a channel matrix at a user side, and performing two-dimensional DFT on the channel matrix, so that the channel matrix in a space-frequency domain is transformed into a channel matrix sparse in an angle-time delaydomain; secondly, constructing a model Quantized-CsiNet which is subjected to quantized channel feedback and reconstruction; thirdly, training the quantized channel feedback and reconstruction model;secondly, performing two-dimensional inverse DFT on the output reconstructed channel matrix, and recovering a channel matrix reconstruction value of an original space-frequency domain; and finally, applying the trained Quantized-CsiNet model to channel state information feedback in each actual scene, and reconstructing an original channel matrix. According to the method, the quantization module and the inverse quantization module are added into the CsiNet, so that the actually transmissible channel state information bit stream can be obtained, the feedback overhead of large-scale MIMO channelinformation is reduced, the reconstruction precision is improved, and particularly, the method has excellent robustness for quantization errors.

The invention discloses a binocular visual sense reconstruction method considering three-dimensional distortion, belonging to the computer visual sense measurement technology field. The binocular visual sense reconstruction method considering the three-dimensional distortion introduces a three-dimensional distortion correction function to perform compensation optimization on a visual sense reconstruction result and realizes three-dimensional high accuracy remodeling. The binocular visual sense reconstruction method considering the three-dimensional distortion adopts a traditional binocular visual sense reconstruction method to obtain a three-dimensional reconstruction initial values of measurement points, solves a three-dimensional distortion compensation value on the basis of a relative relation between a standard object and a binocular camera and the reconstruction initial values, thus performs distortion compensation on coordinates of the camera of a tested point which are respectively extracted from a left image and a right image, combines with binocular visual sense theory to reconstruct, and realizes the binocular three-dimensional high accuracy reconstruction. The binocular visual sense reconstruction method considering three-dimensional distortion is high in reconstruction accuracy, can compensate the three-dimensional distortion of the binocular visual sense, realizes form and surface reconstruction of a part having a big curvature and a big size and improves three-dimensional reconstruction accuracy of the binocular visual sense.

The invention discloses a wavefront quality detection method for a collimation system and a wavefront quality detection device suitable for a large-aperture collimation system. The method comprises the following steps of: acquiring the wavefront slopes Stest of all sub apertures of a micro lens array according to data acquired by a photoelectric detector; substituting Stest into a model method reconstruction formula by a model method of a wavefront reconstruction algorithm to solve the expansion coefficient of a Zernike polynomial and discrete point wavefront phases Wmodal of all apertures; substituting the expansion coefficient into the model method reconstruction formula to solve a group of wavefront slopes Smodal of all sub apertures; calculating the deference of Stest and Smodal of all apertures to obtain the slop Sleft; substituting Sleft into an area method reconstruction formula by an area method in the wavefront reconstruction algorithm to solve the discrete point wavefront phases Wleft of all apertures; and superposing the Wmodal and Wleft to obtain wavefront phases so as to determine the beam quality of the collimation system to be detected. By the method and the device, reconstruction accuracy can be improved.

The invention discloses a bioluminescence tomography reconstruction method. The method includes the steps: acquiring signal distribution conditions of a bioluminescent light source in a small animal body on a body surface by the aid of a bioluminescence tomography data acquisition platform and acquiring anatomical structure information in the small animal body by the aid of a Micro-CT (computed tomography) system; and constructing a system equation based on a three-order simplified spherical harmonic approximation model of a radiative transport equation and a finite-element mesh, and solving the system equation by a compressive sensing sparse regularization method to obtain accurate positioning information of the light source in the small animal body. By the aid of the bioluminescence tomography reconstruction method, three-dimensional light source distribution and the positioning information in the small animal body can be inversed through the two-dimensional light source distribution and the anatomical structure information of the small animal body surface.

The invention discloses a large-scale MIMO channel joint estimation and feedback method based on deep learning. The method comprises the steps: performing initial channel estimation at a user side; constructing a channel estimation subnet CEnet, and minimizing the estimation error through training; constructing a channel feedback subnet; at the user side, inputting the optimized channel estimation value, and outputting compressed code words; at the base station end, inputting the code words, and outputting the reconstructed channel matrix. And the two subnets jointly form a channel estimation and feedback joint network CEFnet. A previous CSI feedback network assumes that perfect channel state information is obtained, does not consider that a channel in practice is obtained by estimation, and has errors and noise. According to the invention, a complete downlink channel estimation and feedback process is realized by constructing a channel estimation and feedback joint network CEFnet, the purpose of eliminating errors and noise is achieved by using a brand new network architecture, and the reconstruction precision is improved while the feedback overhead is reduced.

The invention discloses a self-adaption online dictionary learning super-resolution method. The method firstly selects one group of high-resolution training image set and one group of low-resolution training image set,then establishes a relationship on the high-resolution training image set and the low-resolution training image set, and provides more prior information for image reconstructing through high-resolution / low-resolution dictionary to obtain a better reconstructing effect under the condition of high amplification factor, at the same time, the method uses SP (Self Propelled) sparse coding algorithm, overcomes the defect of low reconstructing accuracy due to traditional greedy algorithm and keeps a lower computation complexity. The method has the advantages of overcoming the defects of complex computation and low training speed in the current mainstream dictionary learning algorithms such as MOD (Multimedia On Demand) and K-SVD(k-singular value decomposition) and shortening image reconstructing time in general. Compared with the prior art based on learning image super-resolution, the method of the invention has a higher reconstructing accuracy and a shorter algorithm time.