CN100510773C - Single satellite remote sensing image small target super resolution ratio reconstruction method - Google Patents
Single satellite remote sensing image small target super resolution ratio reconstruction method Download PDFInfo
- Publication number
- CN100510773C CN100510773C CNB2006100187846A CN200610018784A CN100510773C CN 100510773 C CN100510773 C CN 100510773C CN B2006100187846 A CNB2006100187846 A CN B2006100187846A CN 200610018784 A CN200610018784 A CN 200610018784A CN 100510773 C CN100510773 C CN 100510773C
- Authority
- CN
- China
- Prior art keywords
- image
- remote sensing
- sensing image
- clear
- super resolution
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Landscapes
- Image Processing (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
A method for rebuilding super resolution of small object on single frame satellite remote sense image includes setting up clear - drop model of on - track satellite remote sense influence, discriminating clear - drop parameter, estimating S - N ratio and calculating normalization factor, cutting said image to be superposed sub blocks by rebuilding algorithm, normalizing frequency domain and redundant wavelet domain, decorrelating de - noising operator, setting up fast algorithm of super resolution image rebuild, rebuilding sub pixel, using modulation transfer function as objective standard for evaluating resolution and quality of image.
Description
Technical field
The present invention relates to the satellite remote-sensing image field, particularly a kind of single satellite remote sensing image small target super resolution ratio reconstruction method that falls clear parameter identification based on satellite in orbit.
Background technology
Remote sensing image is in satellite optical imagery, transmission, gatherer process, because imaging system is subjected to atmospheric disturbance, defocuses, the influence of relative motion, image have in various degree fall clearly, be typically aliasing, fuzzy, distortion and system noise, cause the little target can not detected and identification.Super-resolution rebuilding is the algorithm that comes reconstructed image better quality, image that spatial resolution is high from the image second-rate, that resolution is lower, and the detection and the remote sensing image that are mainly used in the identification of satellite image military target, little target are measured.Successful super-resolution technique being applied on the SPOT5 satellite of France in 2002, spatial resolution is brought up to 2 times of former image, and abroad other mechanisms are just studied.China's satellite imagery spatial resolution has only about 4 meters (as No. two, resource) at present, and can reach decimeter grade (is 0.6 meter as QUICKBIRD) abroad.
Because single width image super-resolution rebuilding is the problem of inverting of a morbid state (ill-posed) equation, when carrying out inverting, not necessarily to separate, or a plurality of separating arranged, this is the most thorny issue during image recovers.Amplify owing in the process of inverting, be subjected to high frequency noise, and then flooded useful signal, and introduced fake information.Therefore in the image super-resolution process when increasing the image detail of the high frequency, be crucial and avoid amplifying high frequency random noise (being regularization).
The recovery that classical liftering (Inverse Filter) and Wiener filtering (Weiner Filter) apply to image very early, but owing to the noise amplification and the singularity problem that can not effectively solve the nature image are found the solution, and can not practicability.The beginning of the nineties, in the research of super-resolution image reconstruction method, obtained breakthrough, Andrews and Hunt in 1999 have annotated the theoretical foundation of super-resolution image reconstruction and basic skills at practical problems from the angle of mathematical physics and (have seen article " Understanding theBasis for Recovery of Spatial Frequencies Beyond the Diffraction Limit " the Digital Object IdentifierPage (s) of B.R.Hunt Super-Resolution of Imagery.: 243-248 Feb.1999 and Andrews H C, " Digital Image Restoration " .PrenticeHall that Hunt B R. writes, New York, 1977).Tsai and Huang (1984) proposes frequency domain method, image is transformed to frequency domain to be done and rebuilds high resolution image (the R.Y.Tsai and T.S.Huang that obtains spatial domain after the conversion process, Multiframe image restoration andregistration, in Advances in Computer Vision and Image Processing.), but the method is difficult to comprise priori, can not effectively suppress parasitic ripple.Cheeseman (1994) proposes MAP (Maximum Posteriori Probability) method (P.Cheeseman in the super-resolution of the national space flight planet picture project of rebuilding based on Bayesian, B.Kanefsky, R.Kruft, J.Stutz, and R.Hanson, " Super-Resolved Surface Reconstruction From Multiple Images "), in addition, the for example blind iteration deconvolution of the method method (IBD:Iterative Blind Deconvolution) that also has the spatial domain iteration, convex set projection iteration (POCS) method etc.China Harbin Institute of Technology 2003 has carried out analyzing (article " analysis of ringing and inhibition in the SUPERRESOLUTION PROCESSING FOR ACOUSTIC " of asking for an interview Li Jinzong, Huang Jianming etc.) to the analysis of ringing in the SUPERRESOLUTION PROCESSING FOR ACOUSTIC.But the research of super-resolution image reconstruction at present still exist a lot of problems as: 1. how in the amplification that suppresses high frequency noise, to restore detail of the high frequency; 2. the quality assessment of reconstructed image does not still have unified objective evaluation standard; 3. the discrimination method of efficient system parameter; 4. the order adaptability of reconstruction algorithm and robustness or the like.
Summary of the invention
Technical matters to be solved by this invention is: in view of China's satellite remote sensor level of hardware and China's real current situation, provide a kind of single satellite remote sensing image small target super resolution ratio reconstruction method that falls clear parameter identification based on China's satellite in orbit.This method can be brought up to the remote sensing image spatial resolution 1.5～2 times of former image, and signal to noise ratio (snr) improves more than the 5dB.
The present invention solves its technical matters and adopts following technical scheme:
The invention provides the single width remote sensing image small target super resolution ratio reconstruction method that falls clear Model Distinguish at the satellite in orbit remote sensor, specifically comprise:
(1) be based upon the rail satellite remote-sensing image and fall clear model, clear parameter falls in discrimination:
Clear physical model falls based on remote sensing image, to image point-like atural object, linear ground object analysis, determine that clear model and correlation parameter fall in the remote sensing image imaging system, comprise: the optic branch expands calculates function, the detector parts branch expands calculates function, the point of the relative atural object motion of remote sensor expands calculates function, and remote sensor electronic circuit point expands calculates function, sets up remote sensor imaging system transport function.
(2) estimate remote sensing image signal to noise ratio (S/N ratio) and calculate regularization factor-alpha: to the influence of reconstructed results, suppress parasitic ripple and fake information simultaneously according to the amplification of regularization factor dynamic equalization ambiguity solution process high-frequency noises signal and high frequency details.
(3) the remote sensing image recovery is the ill-conditioning problem solution procedure, increases image detail of the high frequency and image effective bandwidth when suppressing the high frequency random noise signal.
(4) decorrelation de-noising operator: image is decomposed in the mirror image symmetrical wavelet subband, on the basis of morphology Wavelet Nonlinear wavelet coding, suppress image noise, keep the image detail of the high frequency.
(5) set up single satellite remote sensing image super-resolution image reconstruction fast algorithm: deconvolution ambiguity solution in wavelet sub-band, suppress the remote sensing image aliasing, reduce the fake information of image reconstruction.
(6) reconstruction algorithm is divided into overlapping sub-piece with remote sensing images, divides and rule, and reduces algorithm complex.
(7) estimate as the remote sensing image spatial resolution with modulation transfer function and the objective standard of video quality evaluation: reconstruction algorithm is based upon on the standard that clear identification of Model Parameters and video quality evaluation fall in physics.
Method provided by the invention, in the identification of satellite image military target, or the fields such as remote sensing image measurement of the detection of little target and land resource have good application prospects.
The present invention compared with prior art has following significant effect:
One. propose the discrimination method that clear parameter falls in China satellite in orbit remote sensor, set up objective Two-Dimensional Anisotropic PSF.Reconstruction algorithm is based upon on the standard that clear identification of Model Parameters and video quality evaluation fall in physics, has improved the robustness and the adaptivity of reconstruction algorithm.
They are two years old. and according to the statistical distribution characteristic of useful signal and noise, propose, effectively suppress to understand the amplification of the high frequency noise in the blurring process and the introducing of fake information in frequency domain and redundant regularization of wavelet field method.
They are three years old. and reconstruction back remote sensing image spatial resolution is brought up to 1.5～2 times of former image, and SNR improves 5dB.
They are four years old. and reconstruction algorithm can obviously increase remote sensing image effective bandwidth, and breaks through the imaging system cutoff frequency.
They are five years old. and propose spatial resolution and picture quality that utilization MTF and SNR estimate reconstructed image, this method is more objective.
In a word, the present invention is based upon on the standard that clear identification of Model Parameters and video quality evaluation fall in the satellite in orbit remote sensor, decorrelation, ambiguity solution efficient height, and spatial resolution is brought up to 1.5～2 times of former images, and SNR can improve 5dB, and algorithm speed is fast.Can be applicable to the detection of the identification of satellite image military target, little target and the remote sensing image of land resource measures.
Description of drawings
Fig. 1: clear physical model falls in satellite remote-sensing image.
Fig. 2: single width remote sensing image anisotropy sub-pix super-resolution reconstruction algorithm FB(flow block).
Fig. 3 to Fig. 7: clear image reconstruction falls in emulation, its power spectrum and MTF curve thereof.Wherein, Fig. 3, Fig. 4, Fig. 5 be respectively raw video, fall clear image, reconstructed image, and Fig. 6 is the image power spectrum, and power spectrum has significant change before and after rebuilding.By shown in Figure 7, the loss of signal of clear image high frequency detail section is fallen, and image thickens, and reconstructed image has obviously increased effective bandwidth.The high frequency detail signal is compensated behind the image reconstruction, and ISNR (Improvement SNR) is 9.92.
Fig. 8 to Figure 11: the SPOT5 spatial resolution is that 5 meters remote sensing image reconstructed results and spatial resolution are 2.5 meters remote sensing image contrast, and MTF curve comparing result.Wherein, Fig. 8, Fig. 9, Figure 10 are respectively the reconstructed results and the SPOT5 2.5m images of SPOT5 5m image, 5m video.Figure 11 is its MTF curve, and reconstructed image is broken through the cutoff frequency of raw video SPOT5m image, and the single width reconstructed image spatial resolution has 5m to bring up to about 2.5m, and noise ratio SPOT2.5m image is few.
Figure 12 to Figure 13: No. 3 remote sensing images of china natural resources and this method reconstructed results are relatively.Figure 12 is No. 3 former images of remote sensing of resource, and spatial resolution is 3.5m, and its little target atural object is very fuzzy, and (Figure 13) little target is more clear after rebuilding, and spatial resolution is brought up to about 2 times of former image.
Figure 14 to Figure 16: the comparison of No. 3 remote sensing images of another width of cloth china natural resources and this method reconstructed results and same atural object QICKBIRD (spatial resolution is 0.6m) remote sensing image.Figure 14 is No. 3 former images of remote sensing of resource, spatial resolution is 3.5m, the little target atural object of its zigzag very fuzzy (arrow refers to the place), it is more clear to rebuild back (Figure 15) little target, Figure 16 is the high-resolution remote sensing image of same atural object QUICKBIRD, illustrate that the present invention can obtain high resolution image by the reconstruction of low resolution remote sensing image, can effectively suppress the introducing of fake information simultaneously.
Embodiment
The present invention proposes to use remote sensing image linear ground object, rim detection, graphical analysis and experimental formula to come identification China satellite in orbit remote sensor to fall clear parameter; Proposition is based on the morphology redundant wavelet transformation, increase the detail of the high frequency of remote sensing image in the image reconstruction process on the basis of inhibition HF noise signal, carry out regularization in frequency and wavelet field respectively, suppress fake information, realize small target super resolution ratio reconstruction, reconstruction algorithm is based upon on the standard that clear identification of Model Parameters and video quality evaluation fall in physics, has improved the robustness and the adaptivity of reconstruction algorithm.The detection and the remote sensing image that are mainly used in the identification of satellite image military target, little target are measured.
The present invention is the single width remote sensing image small target super resolution ratio reconstruction method that falls clear Model Distinguish, and concrete steps comprise:
(1) be based upon rail satellite remote sensing influence and fall clear model, clear parameter falls in discrimination:
As shown in Figure 1: fall clear physical model based on remote sensing image, to image point-like atural object, linear ground object analysis, determine that clear model and correlation parameter fall in the remote sensing image imaging system, comprising: the optic branch expands calculates function (PSF
Opt), the detector parts branch expands calculates function (PSF
Det), the point of the relative atural object motion of remote sensor expands calculates function (PSF
Mot), remote sensor electronic circuit point expands calculates function (PSF
Elec), and set up remote sensor imaging system transport function (PSF
Total).For the little target detection of remote sensing image, it is the key to the issue of restriction Target Recognition that remote sensing image falls clear.The identification of falling clear parameter has improved the robustness and the adaptivity of reconstruction algorithm.
(2) estimate remote sensing image signal to noise ratio (snr) and calculate regularization factor-alpha: in the influence of ambiguity solution process high-frequency noises signal and detail signal, suppress the amplification and the parasitic ripple of HF noise signal to image according to regularization factor dynamic equalization.Because the classical liftering (inverse filter) and the limitation of Wiener filtering (Wiener filter), the present invention has adopted respectively in frequency domain, wavelet field because the image restoration problem of inverting is carried out the method for regularization (Regularity), can effectively increase the effective bandwidth of image detail of the high frequency and image when suppressing the high frequency random noise signal.
(3) the remote sensing image recovery is the ill-conditioning problem solution procedure, increases image detail of the high frequency and image effective bandwidth when suppressing the high frequency random noise signal.
(4) decorrelation de-noising: significantly remote sensing image is divided into overlapping sub-piece, being about to image decomposes in the mirror image symmetrical wavelet subband, on the basis of morphology Wavelet Nonlinear wavelet coding, according to noise and the different statistical distribution characteristic of useful signal in frequency domain and wavelet field, in redundant wavelet field, suppress the image random noise, keep image detail of the high frequency (edge, texture) simultaneously.
(5) set up single satellite remote sensing image super-resolution image reconstruction fast algorithm: based on the two dimension of identified parameters PSF, in redundant wavelet field, carry out the deconvolution ambiguity solution, suppress the amplification of high frequency noise and the introducing of fake information simultaneously.
(6) reconstruction algorithm is divided into overlapping sub-piece with remote sensing images, divides and rule, and reduces algorithm complex.
(7) estimate as the remote sensing image spatial resolution with modulation transfer function (MTF) and the objective standard of video quality evaluation: reconstruction algorithm is based upon on the standard that clear identification of Model Parameters and video quality evaluation fall in physics.
In the above-mentioned steps (4-7), big image can be divided into overlapping 512 * 512 sub-piece is arranged, to reduce algorithm complex, complexity is O (NlogN).
This method has realized single frames remote sensing image super-resolution image reconstruction, spatial resolution can be brought up to 1.5～2 times of former image, and possessed the requirement that engineering is used, for example possessed in the identification of satellite image military target, or the requirement of the fields such as remote sensing image measurement of the detection of little target and land resource use.
The invention will be further described below in conjunction with accompanying drawing.
1. theoretical foundation: what the satellite remote sensing imaging system was subjected to inevitably that atmospheric disturbance, the relative atural object scene motion of remote sensor imaging system, geometry deformation etc. cause defocuses, and owes sampling, system's random noise, cause falling of remote sensing image clear, degenerate.Super-resolution rebuilding is the algorithm that comes reconstructed image better quality, image that spatial resolution is higher from the image second-rate, that resolution is lower, and actual is the problem of inverting.When carrying out complementary operation, inverting is not necessarily separated, i.e. singular problem; Inversion equation has a plurality of separating in addition.Both of these case all is called the pathosis of image restoration.
The support region of observing image in theory is a limited area on the two dimensional surface, and its imaging model is:
Y＝PSF
total*X+N (1)
In the formula (1):
Y-observation image
The picture of X-actual atural object
PSF
TotalThe optical transfer function of-system
N-random noise
Wherein:
PSF
total(x，y)＝PSF
opt*PSF
mot*PSF
det*PSF
elec
Fall clear physical model based on as shown in Figure 1 satellite remote-sensing image, use for reference the optic branch and expand and calculate function (PSF
Opt), the detector parts branch expands calculates function (PSF
Det), the point of the relative atural object motion of remote sensor expands calculates function (PSF
Mot) experimental formula, the method that proposes according to the present invention according to point-like atural object and linear ground object image measurement and analysis, identification is also set up remote sensor Two-Dimensional Anisotropic PSF
TotalRemote sensing image restores and to fall on the basis of clear parameter and observation image inverting at known remote sensor and invert and estimate actual atural object scene.Decorrelation, ambiguity solution obtain the optimum estimate of atural object image.
2. algorithm flow block diagram:
See Fig. 2: at first clear parameter falls in identification according to typical feature, remote sensing images significantly are divided into overlapping sub-piece, divide and rule, reduce algorithm complex, wavelet transformation has the characteristic of decorrelation, and edge, texture to non-stationary signal-image have good approximation capability, and this point obviously is better than the filtering algorithm of frequency domain, therefore carry out regularization of frequency and wavelet field respectively.The present invention adopts the orthogonal property of mirror image wavelet basis, and redundant wavelet transformation decomposes useful signal and noise in the different subbands, carries out the morphological wavelet coding in wavelet sub-band, de-noising and decorrelation; The deconvolution ambiguity solution.The reconstruct of finishing sub-pix at last obtains the remote sensing images of high spatial resolution.The view picture remote sensing image is formed in sub-piece registration assembly.
3. step:
(1) measure and analyze based on point-like atural object (linear ground object), rim detection, imaged image, identification China satellite in orbit remote sensor is relevant to fall clear parameter, comprising: PSF
Opt, PSF
Det, PSF
Mot, PSF
Elec, set up Two-Dimensional Anisotropic PSF
Total
(2) will be significantly image be divided into overlapping 512 * 512 sub-piece arranged.
(3) according to signal, regularization of the adaptive adjustment of noise power spectrum factor-alpha, in wavelet field and frequency domain, carry out regularization respectively.
(4) the redundant sub-band division of mirror image small echo is utilized the orthogonality decorrelation of mirror image small echo.
(5) in wavelet field, carry out the morphology wavelet coding, suppress process of reconstruction medium-high frequency random noise.
(6) ambiguity solution on Two-Dimensional Anisotropic PSF basis carries out deconvolution ambiguity solution and the parasitic ripple that suppresses image in wavelet field, reduce fake information.
(7) the redundant small echo remote sensing image of sub-pix reconstruct.
(8) repeat 3-7 steps.
(9) the view picture high spatial resolution image is formed in overlapping sub-piece geometrical registration assembly.
4. use:
This patent can improve the satellite remote-sensing image spatial resolution, is mainly used in high-resolution remote sensing image and (comprises high spectrum, SAR) obtain, the remote sensing image location and the measurement of the detection of military target identification, little target and land resource.
Claims (5)
1. fall the single satellite remote sensing image small target super resolution ratio reconstruction method of clear parameter identification based on satellite in orbit, it is characterized in that comprising to fall the single width remote sensing image small target super resolution ratio reconstruction method of clear parameter identification:
(1) be based upon the rail satellite remote-sensing image and fall clear model, clear parameter falls in discrimination:
Clear physical model falls based on remote sensing image, to image point-like atural object, linear ground object analysis, determine that clear model and correlation parameter fall in the remote sensing image imaging system, comprise: the optic branch expands calculates function, the detector parts branch expands calculates function, and the point of the relative atural object motion of remote sensor expands calculates function, and remote sensor electronic circuit point expands calculates function, set up remote sensor imaging system transport function
(2) estimate remote sensing image signal to noise ratio (S/N ratio) and calculate regularization factor-alpha: to the influence of reconstructed results, suppress parasitic ripple and fake information according to the amplification of regularization factor dynamic equalization ambiguity solution process high-frequency noises signal and high frequency details simultaneously,
(3) remote sensing image recovers: be the ill-conditioning problem solution procedure, when suppressing the high frequency random noise signal, increase image detail of the high frequency and image effective bandwidth,
(4) decorrelation de-noising operator: image is decomposed in the mirror image symmetrical wavelet subband, on the basis of morphology Wavelet Nonlinear wavelet coding, suppress image noise, keep the image detail of the high frequency,
(5) set up single satellite remote sensing image super-resolution image reconstruction fast algorithm: deconvolution ambiguity solution in wavelet sub-band, suppress the remote sensing image aliasing, reduce the fake information of image reconstruction,
(6) reconstruction algorithm is divided into overlapping sub-piece with remote sensing images, divides and rule, and reduces algorithm complex,
(7) estimate as the remote sensing image spatial resolution with modulation transfer function and the objective standard of video quality evaluation: reconstruction algorithm is based upon on the standard that clear identification of Model Parameters and video quality evaluation fall in physics.
2. single satellite remote sensing image small target super resolution ratio reconstruction method according to claim 1 is characterized in that in frequency domain, redundant wavelet field remote sensing image being restored respectively carry out regularization.
3. single satellite remote sensing image small target super resolution ratio reconstruction method according to claim 1, it is characterized in that: single satellite remote sensing image small target super resolution ratio reconstruction method is to be based upon on the standard of falling clear identification of Model Parameters and video quality evaluation, and clear model parameter falls in identification according to image measurement, with the evaluation criterion of modulation transfer function and signal to noise ratio (S/N ratio) compensation as remote sensing image spatial resolution and reconstruction algorithm quality.
4. single satellite remote sensing image small target super resolution ratio reconstruction method according to claim 1 is characterized in that for the single frames remote sensing image, and its spatial resolution is 1.5～2 times of former image, and signal to noise ratio (S/N ratio) improves 5dB.
5. single satellite remote sensing image small target super resolution ratio reconstruction method according to claim 1 is characterized in that big image is divided into overlapping 512 * 512 sub-piece is arranged, and algorithm complex is O (NlogN).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100187846A CN100510773C (en) | 2006-04-14 | 2006-04-14 | Single satellite remote sensing image small target super resolution ratio reconstruction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100187846A CN100510773C (en) | 2006-04-14 | 2006-04-14 | Single satellite remote sensing image small target super resolution ratio reconstruction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1831556A CN1831556A (en) | 2006-09-13 |
| CN100510773C true CN100510773C (en) | 2009-07-08 |
Family
ID=36993968
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2006100187846A Expired - Fee Related CN100510773C (en) | 2006-04-14 | 2006-04-14 | Single satellite remote sensing image small target super resolution ratio reconstruction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100510773C (en) |
Families Citing this family (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101578632B (en) * | 2007-10-10 | 2013-03-27 | 美国日本电气实验室公司 | Soft edge smoothness prior and application on alpha channel super resolution |
| WO2009087725A1 (en) * | 2008-01-09 | 2009-07-16 | Panasonic Corporation | Image data generation device, image data generation method, and image data generation program |
| CN101540039B (en) * | 2008-03-21 | 2011-09-07 | 李金宗 | Method for super resolution of single-frame images |
| CN101315424B (en) * | 2008-07-29 | 2011-06-08 | 中国科学院对地观测与数字地球科学中心 | Multi-satellite remote sensing data integrated parallel ground pretreatment system |
| CN101408621B (en) * | 2008-11-28 | 2011-12-28 | 哈尔滨工业大学 | Ultra-resolution method based on polarization synthetic aperture radar image |
| BR112013016887B1 (en) * | 2011-01-07 | 2021-12-14 | Sony Computer Entertaiment America Llc | METHOD AND APPLIANCE FOR ANTI-INTRODUCTION OF MORPHOLOGICAL ERROR AND, NON-TRANSITORY COMPUTER READable STORAGE MEDIUM |
| CN102819829A (en) * | 2012-07-17 | 2012-12-12 | 中国科学院地理科学与资源研究所 | Rebuilding algorithm for super-resolution remote sensing image based on fractal theory |
| CN103020905B (en) * | 2012-10-09 | 2016-05-25 | 武汉大学 | For the sparse constraint adaptive N LM super resolution ratio reconstruction method of character image |
| CN104766272B (en) * | 2015-04-17 | 2016-03-16 | 河海大学 | A kind of image super-resolution rebuilding method based on sub-pix skew model |
| CN106570862B (en) * | 2016-10-25 | 2020-02-21 | 中国人民解放军信息工程大学 | Method and device for evaluating super-resolution reconstruction quality |
| CN109270516B (en) * | 2018-09-01 | 2022-05-17 | 哈尔滨工程大学 | Beam forming method suitable for unmanned mobile platform to detect naval vessel line spectrum |
| CN109636727B (en) * | 2018-12-17 | 2022-11-15 | 辽宁工程技术大学 | Super-resolution reconstruction image spatial resolution evaluation method |
| CN110728658A (en) * | 2019-09-16 | 2020-01-24 | 武汉大学 | High-resolution remote sensing image weak target detection method based on deep learning |
| CN111080532A (en) * | 2019-10-16 | 2020-04-28 | 北京理工大学深圳研究院 | Remote sensing image super-resolution restoration method based on ideal edge extrapolation |
| CN111325184B (en) * | 2020-03-20 | 2023-04-18 | 宁夏回族自治区自然资源勘测调查院 | Intelligent interpretation and change information detection method for remote sensing image |
| CN111399477B (en) * | 2020-03-20 | 2021-05-04 | 北华航天工业学院 | Parameter separation-based Wiener nonlinear system identification method |
| CN112131994B (en) * | 2020-09-16 | 2024-01-23 | 广东海洋大学 | Anti-noise satellite image small target detection method |
| CN113962897B (en) * | 2021-11-02 | 2022-09-02 | 中国空间技术研究院 | Modulation transfer function compensation method and device based on sequence remote sensing image |
| CN116612048B (en) * | 2023-07-17 | 2023-09-26 | 山东产研卫星信息技术产业研究院有限公司 | Method and system for deblurring optical satellite remote sensing image |
-
2006
- 2006-04-14 CN CNB2006100187846A patent/CN100510773C/en not_active Expired - Fee Related
Non-Patent Citations (6)
| Title |
|---|
| Super-resolution of imagery: understanding the basis forrecovery of spatial frequencies beyond the diffraction limit. Hunt, B.R.Information, Decision and Control, 1999,Vol.IDC 99. Proceedings. 1999 . 1999 |
| Super-resolution of imagery: understanding the basis forrecovery of spatial frequencies beyond the diffraction limit. Hunt, B.R.Information, Decision and Control, 1999,Vol.IDC 99. Proceedings. 1999 . 1999 * |
| 基于形态插值的图像重建方法. 段汕，秦前清.计算机工程与应用，第2004年卷第18期. 2004 |
| 基于形态插值的图像重建方法. 段汕，秦前清.计算机工程与应用，第2004年卷第18期. 2004 * |
| 超分辨率处理中振铃现象的分析与抑制. 李金宗，黄建明，陈凤，李冬冬.系统工程与电子技术，第25卷第6期. 2003 |
| 超分辨率处理中振铃现象的分析与抑制. 李金宗，黄建明，陈凤，李冬冬.系统工程与电子技术，第25卷第6期. 2003 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1831556A (en) | 2006-09-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100510773C (en) | Single satellite remote sensing image small target super resolution ratio reconstruction method | |
| CN100510772C (en) | Small target super resolution reconstruction method for remote sensing image | |
| CN102800071B (en) | Method for reconstructing super resolution of sequence image POCS | |
| CN101847257B (en) | Image denoising method based on non-local means and multi-level directional images | |
| Zheng et al. | Wavelet based nonlocal-means super-resolution for video sequences | |
| CN102945548A (en) | Directional pyramid filtering-based image processing method and device | |
| CN109447903B (en) | Method for establishing semi-reference super-resolution reconstruction image quality evaluation model | |
| CN101540039B (en) | Method for super resolution of single-frame images | |
| CN102722879A (en) | SAR (synthetic aperture radar) image despeckle method based on target extraction and three-dimensional block matching denoising | |
| Khattab et al. | A Hybrid Regularization-Based Multi-Frame Super-Resolution Using Bayesian Framework. | |
| CN109615584B (en) | SAR image sequence MAP super-resolution reconstruction method based on homography constraint | |
| Islam et al. | Super resolution of 3d MRI images using a Gaussian scale mixture model constraint | |
| CN107767342B (en) | Wavelet transform super-resolution image reconstruction method based on integral adjustment model | |
| Shen et al. | Universal reconstruction method for radiometric quality improvement of remote sensing images | |
| CN107154019B (en) | Electromagnetic imaging super-resolution method based on LR-geometric interpolation | |
| CN117274094B (en) | Deconvolution reconstruction method for improving wearable ultrasonic imaging quality | |
| Vijayalakshmi et al. | A Contrast Adjusted Wavelet and Curvelet Transform Based Fusion of Breast Images | |
| Aslan et al. | A comparison study for image denoising | |
| Fazel et al. | Speckle reduction of SAR images using curvelet and wavelet transforms based on spatial features characteristics | |
| Wang et al. | An Improved Hyperspectral Image Super Resolution Restoration Algorithm Based on POCS | |
| Chavez-Roman et al. | Image super resolution using interpolation and edge extraction in wavelet transform space | |
| Sharma et al. | Redundant wavelet transform based image super resolution | |
| Suryanarayana et al. | Sparse representation based super-resolution algorithm using wavelet domain interpolation and nonlocal means | |
| Algamudi et al. | A New Technique to Enhance SODISM Images based on the Modified Undecimated Wavelet Transform | |
| Hashim | Image Denoising Based Improved Curvelet Threshold |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| ASS | Succession or assignment of patent right |
Owner name: BAINIAN JINHAI SECURITY TECHNOLOGY CO., LTD. Free format text: FORMER OWNER: WUHAN UNIVERSITY Effective date: 20121214 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| COR | Change of bibliographic data |
Free format text: CORRECT: ADDRESS; FROM: 430072 WUHAN, HUBEI PROVINCE TO: 450000 ZHENGZHOU, HENAN PROVINCE |
|
| TR01 | Transfer of patent right |
Effective date of registration: 20121214 Address after: 2210, room 12, century green summit, 450000 business circle road, Zheng Dong New District, Zhengzhou, Henan Patentee after: Bainianjinhai Security Technology Co., Ltd. Address before: 430072 Hubei city of Wuhan province Wuchang Luojiashan Patentee before: Wuhan University |
|
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20090708 Termination date: 20190414 |