CN102044068A - Wavelet MTF compensation method based on optimal core shape - Google Patents
Wavelet MTF compensation method based on optimal core shape Download PDFInfo
- Publication number
- CN102044068A CN102044068A CN 201010543517 CN201010543517A CN102044068A CN 102044068 A CN102044068 A CN 102044068A CN 201010543517 CN201010543517 CN 201010543517 CN 201010543517 A CN201010543517 A CN 201010543517A CN 102044068 A CN102044068 A CN 102044068A
- Authority
- CN
- China
- Prior art keywords
- coefficient
- wavelet
- image
- mtf
- obtains
- 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.)
- Granted
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a wavelet modulation transfer function (MTF) compensation method based on the optimal core shape, and can be applied to a ground processing system for satellite remote sensing images. The method comprises the following steps of: according to priori knowledge of noise of an imaging system, inputting a threshold value, and partially separating input digital image signals and noise in a wavelet domain; inputting parameters a and b for initial core design, performing wavelet domain modulation transfer function compensation (MTFC) processing on the signal part, adding with the noise part, performing wavelet domain normalizatioin processing and outputting process images; continuously regulating the parameters a and b, generating different cores and corresponding different process images, and measuring corresponding just noticeable difference (JND) and MTF through the process images; and recording the parameters a and b which meet the requirement, and using the parameters a and b as the optimal core parameters of the imaging system. The invention can improve the MTF for the imaging system with low ontrack dynamic MTF, improve the quality grade of the images, and improve system performance.
Description
Technical field
The present invention relates to aerospace optical remote sensing imaging and Ground Processing System field, particularly relate to a kind of small echo MTF compensation method, realize the lifting of picture quality based on best nuclear shape.
Background technology
Following table is the used English abbreviation explanation of the present invention:
| Abbreviation | English name | Chinese |
| DN | Digital?Number | Digital quantity |
| JND | Just?noticeable?difference | Just noticeable difference |
| MTF | Modulation?Transfer?Function | Modulation transfer function |
| MTFC | Modulation?Transfer?Function?Compensation | The MTF compensation |
| NASA | National?Aeronautics?and?Space?Administration | NASA |
In the remote optical sensing development of imaging system, the user often retrains the imaging system parameter, the index request that the most frequently used is at the dynamic MTF of rail, and the dynamic MTF that requires such as NASA is greater than 0.1.At the dynamic MTF of rail is to be determined by the many influence factors in the imaging process, comprise the static MTF of optical sensor, atmosphere, motion, integration synchronously, the attitude of satellite, drift correction, flutter, space environment, veiling glare etc., cause the fuzzy of image at the dynamic MTF of rail.Because it is numerous in the dynamic MTF influence factor of rail, it is often relatively more difficult to satisfy index request, external numerous satellite company often adopts the method for MTF compensation to improve at the dynamic MTF of rail, as IKONOS, Orbview-3, Geoeye-1 (high-resolution satellite of U.S.'s emission) etc., even the Pleiades satellite that France is about to launch has proposed the static MTF of reduction camera, reduce aliasing,, reduced development cost greatly by the optimal design idea of MTF compensation raising at the dynamic MTF of rail.This shows the significance of MTF compensation method in remotely sensed image system and remote sensor development, but external disclosed more detailed MTF compensation method only has the Wiener filtering method based on frequency domain of Korea S Kompsat-1, and other MTF of remotely sensed image system compensation method is not open.The method that domestic colleges and universities and research unit propose mostly is based on the MTF compensation method in frequency domain or spatial domain, does not appear in the newspapers based on the MTF compensation method of small echo.
Generally speaking, MTF compensation method based on frequency domain or spatial domain has following shortcoming: 1. generally only be applicable to the processing of stationary signal based on the method for frequency domain, distinguish signal and noise by the statistical distribution of noise, and image often comprises edge and non-stationary signal, fourier spectrum is widely distributed, especially at high frequency treatment, frequency domain method is difficult to distinguish noise and signal, and edge causes ring easily; 2. the method based on the spatial domain often adopts iteration, iterating to certain circulation time stops, picture quality is just better, having identical shortcoming with method based on frequency domain is exactly ringing effect, method calculated amount based on iteration is often bigger in addition, need the research fast algorithm, handle otherwise be unsuitable for the high speed big data quantity; Though 3. calculate simply based on the method in frequency domain or spatial domain, the divided degree of picture signal and noise is relatively poor, picture signal and noise amplify substantially together, and this is based on frequency domain or the topmost shortcoming of spatial domain method.
Summary of the invention
Technology of the present invention is dealt with problems and is: overcome the deficiencies in the prior art, a kind of small echo MTF compensation method based on best nuclear shape is provided, this method distinguishes the priori noise by the wavelet field threshold value and wavelet field MTFC handles, and has solved the problem that signal and noise amplify simultaneously.Adopt the hard-threshold of revising to suppress to noise signal simultaneously, rather than simple the removal, realized the processing procedure information lossless.
Technical solution of the present invention is:
A kind of small echo MTF compensation method based on best nuclear shape, step is as follows:
(1) digital picture is carried out wavelet transformation and decomposes, obtain the original decomposition coefficient, the original decomposition coefficient is carried out following two-part processing:
First: judge that whether each coefficient of dissociation is greater than threshold value σ in the original decomposition coefficient, will be less than the coefficient of dissociation zero setting of threshold value σ, coefficient of dissociation greater than threshold value σ remains unchanged, and afterwards the coefficient of dissociation that obtains is carried out inverse wavelet transform and obtains signal pattern; Utilize formula H (u)=1+a (u-1) sin
2(bu) calculate the value of examining H (u), H (u) is generated two-dimensional convolution by the symmetry rotation examine P; Wherein, u is the yardstick of wavelet decomposition and u for smaller or equal to 6 natural number, and a, b are input quantity, a 〉=0,0≤b≤π/u;
Next signal pattern that inverse wavelet transform is obtained and two-dimensional convolution nuclear P carries out the image after wavelet field MTFC processing obtains handling; Described wavelet field MTFC handles and is meant that signal pattern is at first carried out the spatial domain deconvolution to be handled, image after then deconvolution being handled carries out wavelet decomposition and obtains coefficient of wavelet decomposition, again coefficient of wavelet decomposition and two-dimensional convolution nuclear P are carried out convolution algorithm, at last the data after the convolution algorithm are carried out the signal pattern after wavelet inverse transformation obtains handling, enter step (2) afterwards; Described threshold value σ is meant the integer between 0～0.1.
Second portion: judge that whether each coefficient of dissociation is greater than threshold value σ in the original decomposition coefficient, will be greater than the coefficient of dissociation zero setting of threshold value σ, coefficient of dissociation less than threshold value σ remains unchanged, and afterwards the coefficient of dissociation that obtains is carried out inverse wavelet transform and obtains noise image, enters step (2) afterwards;
(2) noise image that obtains of signal pattern after the processing that first in the step (1) is obtained and second portion carries out sum operation, enters step (3) afterwards;
(3) result with step (2) sum operation carries out wavelet field regularization processing, output procedure image;
(4) the J ND value and the mtf value of the procedural image of output in the calculation procedure (3), judge whether to satisfy J ND value greater than 0.4 and mtf value greater than 0.1, if satisfy, input parameter a when then calculating nuclear H (u) in the first in the step (1) and b are one group of best nuclear parameter, the procedural image of output is an optimized image in the step (3) simultaneously, with best nuclear parameter and optimized image output; If do not satisfy, then reselect the value of input parameter a and b, by formula H (u)=1+a (u-1) sin
2(bu) the symmetry rotation generates two-dimensional convolution nuclear P.Turn back to afterwards and carry out wavelet field MTFC in the first of step (1) and handle, carry out wavelet field MTFC and enter step (2) after handling.
Described wavelet field regularization is handled and is meant that image is carried out wavelet decomposition earlier obtains coefficient of wavelet decomposition, and the coefficient of wavelet decomposition that obtains and the hard-threshold factor of correction are multiplied each other, and at last multiplied result is carried out wavelet inverse transformation, and then the output procedure image; The hard-threshold factor of described correction is
Wherein, the hard-threshold multiplier factor of δ for revising, w is a coefficient of wavelet decomposition, and u is the yardstick of wavelet decomposition, and σ is meant the integer between 0～0.1.
The present invention's beneficial effect compared with prior art is:
(1) the present invention is directed to picture quality in the aerospace optical remote sensing imaging system (at the dynamic MTF of rail) problem,, proposed a kind of small echo MTF compensation method based on best nuclear shape based on the MTFC processing module of ground image disposal system.Distinguish priori noise and wavelet field MTFC processing by the wavelet field threshold value, solved the problem that signal and noise amplify simultaneously.Adopt the hard-threshold of revising to suppress to noise signal simultaneously, rather than simple the removal, realized the processing procedure information lossless.Effectively promoted at the dynamic MTF of rail, compared to existing technology, had information lossless, be applicable to high spatial frequency low-key system target, MTF promotes the high advantage of multiple, this method also can be used for carrying out the camera overall design, carries out design parameter optimization.
(2) the present invention is by the threshold zone sub-signal and the noise of noise priori, only signal section being carried out wavelet field MTFC handles, the shortcoming of effectively having avoided noise and signal to amplify together, various complex situations have been considered in the design of residual noise influence having carried out nuclear after the signal pattern deconvolution.According to H (u)=1+a (u-1) sin
2(bu) design nuclear generates two-dimensional convolution nuclear by the symmetry rotation, and (u-1) dull rising factor of nuclear design can promote high frequency to greatest extent, and u=1 represents low frequency, does not amplify, and guarantees that gross energy is constant, design sin
2() function, the amplification of residual noise when suitably reducing high frequency, the residual noise size has determined the shape of nuclear, noise is big more, and it is more little to promote multiple, and nuclear has the characteristic more than or equal to 1 simultaneously, a, b control nuclear shape, when a, b within span, can guarantee that spatial frequency promotes to greatest extent, can reach again simultaneously and promote compromise that multiple and residual noise amplify.Measure the MTF curve at rail, high frequency attenuation is big, and the picture modulation percentage of the nyquist frequency that the imaging system that has obtains only has about 0.03, human eye is difficult to observe image information, by the value of choose reasonable a, b, picture modulation percentage can be brought up to more than 0.1, satisfies request for utilization.
(3) the present invention is directed to aerospace optical remote sensing imaging system picture quality and image applications, by the evaluation method that subjective JND and objective MTF combine, as the criterion of optimizing nuclear parameter, the optimizing process of nuclear parameter is artificially controlled, and stability is high, and is with strong points.Based on the method for Bayesian Estimation and other automatic optimals, all have more hypothesis prerequisite in the document, suitable image range is narrower, often adopts and the relatively poor Y-PSNR objective evaluations such as (PSNR) of image applications correlativity.And the process of optimal estimation and automatic optimal, processing speed is slower.
(4) the present invention is directed to aerospace optical remote sensing imaging system picture quality and image applications, require disposal route can not cause information dropout, in processing procedure, keep noise image, and doing the after-applied correction hard-threshold of sum operation, it is relevant with the input threshold value to revise hard-threshold, can be by the input threshold value control, and the input threshold value is determined according to the priori of camera noise.Whole process does not have the process of denoising, therefore can not cause information dropout, meets the requirement of using in the aerospace optical remote sensing imaging Ground Processing System MTFC module.
Description of drawings
Fig. 1 is the inventive method process flow diagram;
Fig. 2 is a nuclear shape design drawing of the present invention;
Fig. 3 is the hard-threshold figure of correction of the present invention.
Embodiment
Below in conjunction with accompanying drawing the specific embodiment of the present invention is further described in detail:
As shown in Figure 1, a kind of small echo MTF compensation method based on best nuclear shape can be applicable to the satellite remote sensing images Ground Processing System.According to the priori of imaging system noise, the input threshold value realizes that in wavelet field the part of input digital image signal and noise is separated.Input a, b parameter are carried out the incipient nucleus design, signal section is carried out wavelet field MTFC handle, and after the noise section addition, carry out the wavelet field regularization and handle the output procedure image.By continuous adjustment a, b parameter, generate different nuclear and corresponding various process image, measure corresponding J ND, MTF by procedural image.A, b parameter that record meets the demands are as the best nuclear parameter of this imaging system.The present invention can promote for MTF is provided in the lower imaging system of the dynamic MTF of rail, improves the quality grade of image, improves system performance.Step is as follows:
(1) in aerospace optical remote sensing imaging Ground Processing System, particularly high resolving power camera imaging system often has the MTF compensating module, and this module is accepted the camera shot digital images; Digital picture is carried out wavelet transformation decompose, the small echo form of employing is not done qualification, and with the wlet1 symbolic representation, wavelet decomposition yardstick u is the natural number smaller or equal to 6, obtains the original decomposition coefficient, and the original decomposition coefficient is carried out following two-part processing:
First: judge that whether each coefficient of dissociation is greater than threshold value σ in the original decomposition coefficient, described threshold value σ is meant the number between 0 and 0.1, will be less than the coefficient of dissociation zero setting of threshold value σ, coefficient of dissociation greater than threshold value σ remains unchanged, and afterwards the coefficient of dissociation that obtains is carried out inverse wavelet transform and obtains signal pattern; Utilize formula H (u)=1+a (u-1) sin
2(bu) calculate nuclear H (u), then H (u) is carried out the symmetry rotation and generate two-dimensional convolution nuclear P, the physical significance of described two-dimensional convolution nuclear P can be understood as the two-dimensional points spread function, u is the yardstick of wavelet decomposition, and a, b are input quantity, a 〉=0,0≤b≤π/u, the shape of a, b control nuclear H (u); Next signal pattern that inverse wavelet transform is obtained and two-dimensional convolution nuclear P carries out the image after wavelet field MTFC processing obtains handling, described wavelet field MTFC handles and is meant that signal pattern is at first carried out the spatial domain deconvolution to be handled, described spatial domain deconvolution is handled, do not do qualification, can be referring to relevant document.Obtain the image after deconvolution is handled, secondly the image after deconvolution being handled adopts the small echo that is different from wlet1, wlet2 represents with symbol, carry out wavelet decomposition and obtain coefficient of wavelet decomposition, wavelet decomposition yardstick u is the natural number smaller or equal to 6, again coefficient of wavelet decomposition and two-dimensional convolution nuclear P are carried out convolution algorithm, at last the data after the convolution algorithm are carried out the signal pattern after wavelet inverse transformation obtains handling.
Second portion: judge that whether each coefficient of dissociation is greater than threshold value σ in the original decomposition coefficient, will be greater than the coefficient of dissociation zero setting of threshold value σ, coefficient of dissociation less than threshold value σ remains unchanged, and afterwards the coefficient of dissociation that obtains is carried out inverse wavelet transform and obtains noise image.
(2) noise image that obtains of signal pattern after the processing that first in the step (1) is obtained and second portion carries out sum operation;
(3) result with step (2) sum operation carries out wavelet field regularization processing, the output procedure image, described wavelet field regularization is handled and is meant the small echo that the employing of image elder generation is different from wlet1 and wlet2, wlet3 represents with symbol, carries out wavelet decomposition, obtains coefficient of wavelet decomposition, wavelet decomposition yardstick u is the natural number smaller or equal to 6, the coefficient of wavelet decomposition that obtains and the hard-threshold factor of correction are multiplied each other, at last multiplied result is carried out wavelet inverse transformation, and then the output procedure image; The hard-threshold factor of described correction is
Wherein, the hard-threshold factor of δ for revising, w is a coefficient of wavelet decomposition, and u is the yardstick of wavelet decomposition, and σ is meant the number between 0～0.1;
(4) the JND value and the mtf value of the procedural image of output in the calculation procedure (3), described JN D value is a just noticeable difference, be the subjectivity amount, its measuring method can be with reference to international standard ISO12233, described MTF is a modulation transfer function, mould value for the point spread function Fourier transform, its value can be measured from image by sword Bian Fa etc., judge whether to satisfy JN D value greater than 0.4 and mtf value greater than 0.1, if satisfy, input parameter a when then calculating nuclear H (u) in the first in the step (1) and b are one group of best nuclear parameter, and the procedural image of output is an optimized image in the step (3) simultaneously, with best nuclear parameter and optimized image output; If do not satisfy, then reselect the value of input parameter a and b, by formula H (u)=1+a (u-1) sin
2(bu) calculate the value of examining H (u), turn back to afterwards in the first of step (1), carry out wavelet field MTFC and handle, carry out entering step (2) after the wavelet field MTFC processing.
Input threshold value σ in the described step (1) is according to the camera noise characteristics, consider the camera space of 8 quantifications, camera electronics noise is the about 1DN of standard deviation, and average is 0 white Gaussian noise, corresponding coefficient of wavelet decomposition also is the white noise of 0 average, and maximal value is 0.004.If camera electronics noise criteria difference changes k doubly, input threshold value σ is set to 0.004 * k.Noise can not surpass 25 DN generally speaking, and therefore importing threshold value σ maximum occurrences is 0.004 * 25=0.1, and minimum value is 0.
As shown in Figure 2, one dimension nuclear has been designed to peaked symmetrical structure, and when a, b are in span, H (u) is more than or equal to 1, such design, can guarantee the system capacity unchangeability, take into account the two kinds of situations that influence that promote high frequency to greatest extent and avoid residual noise again, when the maximum occurrences of bu (
] when interval, each yardstick is monotone increasing successively, high fdrequency component can access to greatest extent and improve, and is applicable to the system that noise is less, when the maximum occurrences of bu (
] when interval, carry out the high frequency attenuation restriction, it is maximum that characteristics are that intermediate frequency amplifies, and high frequency also amplifies, but degree is less than intermediate frequency, considers trading off that residual noise and high fdrequency component improve, and is applicable to the imaging system that noise ratio is more serious.
As shown in Figure 3, considered to adopt the hard-threshold of revising from the fidelity angle.When coefficient of wavelet decomposition be positioned at (u σ, u σ] when interval, coefficient of wavelet decomposition major part is noise, with coefficient of wavelet decomposition * 0.5, in the time of outside coefficient of wavelet decomposition is positioned at the interval, coefficient of wavelet decomposition remains unchanged.The hard-threshold factor of revising is defined as:
Wherein δ is the hard-threshold multiplier factor of correction, and w is a coefficient of wavelet decomposition, and u and σ definition interval are same as described above.
The content that is not described in detail in the instructions of the present invention belongs to those skilled in the art's known technology.
Claims (2)
1. small echo MTF compensation method based on best nuclear shape is characterized in that step is as follows:
(1) digital picture is carried out wavelet transformation and decomposes, obtain the original decomposition coefficient, the original decomposition coefficient is carried out following two-part processing:
First: judge that whether each coefficient of dissociation is greater than threshold value σ in the original decomposition coefficient, will be less than the coefficient of dissociation zero setting of threshold value σ, coefficient of dissociation greater than threshold value σ remains unchanged, and afterwards the coefficient of dissociation that obtains is carried out inverse wavelet transform and obtains signal pattern; Utilize formula H (u)=1+a (u-1) sin
2(bu) calculate the value of examining H (u), H (u) is generated two-dimensional convolution by the symmetry rotation examine P; Wherein, u is the yardstick of wavelet decomposition and u for smaller or equal to 6 natural number, and a, b are input quantity, a 〉=0,0≤b≤π/u;
Next signal pattern that inverse wavelet transform is obtained and two-dimensional convolution nuclear P carries out the image after wavelet field MTFC processing obtains handling; Described wavelet field MTFC handles and is meant that signal pattern is at first carried out the spatial domain deconvolution to be handled, image after then deconvolution being handled carries out wavelet decomposition and obtains coefficient of wavelet decomposition, again coefficient of wavelet decomposition and two-dimensional convolution nuclear P are carried out convolution algorithm, at last the data after the convolution algorithm are carried out the signal pattern after wavelet inverse transformation obtains handling, enter step (2) afterwards; Described threshold value σ is meant the integer between 0～0.1.
Second portion: judge that whether each coefficient of dissociation is greater than threshold value σ in the original decomposition coefficient, will be greater than the coefficient of dissociation zero setting of threshold value σ, coefficient of dissociation less than threshold value σ remains unchanged, and afterwards the coefficient of dissociation that obtains is carried out inverse wavelet transform and obtains noise image, enters step (2) afterwards;
(2) noise image that obtains of signal pattern after the processing that first in the step (1) is obtained and second portion carries out sum operation, enters step (3) afterwards;
(3) result with step (2) sum operation carries out wavelet field regularization processing, output procedure image;
(4) the JND value and the mtf value of the procedural image of output in the calculation procedure (3), judge whether to satisfy the JND value greater than 0.4 and mtf value greater than 0.1, if satisfy, input parameter a when then calculating nuclear H (u) in the first in the step (1) and b are one group of best nuclear parameter, the procedural image of output is an optimized image in the step (3) simultaneously, with best nuclear parameter and optimized image output; If do not satisfy, then reselect the value of input parameter a and b, by formula H (u)=1+a (u-1) sin
2(bu) the symmetry rotation generates two-dimensional convolution nuclear P.Turn back to afterwards and carry out wavelet field MTFC in the first of step (1) and handle, carry out wavelet field MTFC and enter step (2) after handling.
2. a kind of small echo MTF compensation method according to claim 1 based on best nuclear shape, it is characterized in that: described wavelet field regularization is handled and is meant that image is carried out wavelet decomposition earlier obtains coefficient of wavelet decomposition, the coefficient of wavelet decomposition that obtains and the hard-threshold factor of correction are multiplied each other, at last multiplied result is carried out wavelet inverse transformation, and then the output procedure image; The hard-threshold factor of described correction is
Wherein, the hard-threshold multiplier factor of δ for revising, w is a coefficient of wavelet decomposition, and u is the yardstick of wavelet decomposition, and σ is meant the integer between 0～0.1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105435177A CN102044068B (en) | 2010-11-12 | 2010-11-12 | Compensation method of wavelet modulation transfer function based on optimal kernel shape |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105435177A CN102044068B (en) | 2010-11-12 | 2010-11-12 | Compensation method of wavelet modulation transfer function based on optimal kernel shape |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102044068A true CN102044068A (en) | 2011-05-04 |
| CN102044068B CN102044068B (en) | 2012-05-09 |
Family
ID=43910182
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2010105435177A Expired - Fee Related CN102044068B (en) | 2010-11-12 | 2010-11-12 | Compensation method of wavelet modulation transfer function based on optimal kernel shape |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102044068B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102510444A (en) * | 2011-11-23 | 2012-06-20 | 青岛市光电工程技术研究院 | Medium-frequency and high-frequency MTF (modulation transfer function) online compensating and reinforcing method of optical image camera |
| CN103983343A (en) * | 2014-05-29 | 2014-08-13 | 武汉大学 | Satellite platform chattering detection method and system based on multispectral image |
| CN106022354A (en) * | 2016-05-07 | 2016-10-12 | 浙江大学 | SVM-based image MTF measurement method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020186772A1 (en) * | 2001-06-08 | 2002-12-12 | Xin Li | Wavelet domain motion compensation system |
| CN1804657A (en) * | 2006-01-23 | 2006-07-19 | 武汉大学 | Small target super resolution reconstruction method for remote sensing image |
| CN1964433A (en) * | 2006-11-21 | 2007-05-16 | 华为技术有限公司 | A method and device to reduce image acquisition device noise |
-
2010
- 2010-11-12 CN CN2010105435177A patent/CN102044068B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020186772A1 (en) * | 2001-06-08 | 2002-12-12 | Xin Li | Wavelet domain motion compensation system |
| CN1804657A (en) * | 2006-01-23 | 2006-07-19 | 武汉大学 | Small target super resolution reconstruction method for remote sensing image |
| CN1964433A (en) * | 2006-11-21 | 2007-05-16 | 华为技术有限公司 | A method and device to reduce image acquisition device noise |
Non-Patent Citations (2)
| Title |
|---|
| 《信息与信号处理》 20071231 胡利军 小波变换和SUSAN算子在图像处理中的应用 13-15 1-2 第37卷, 第8期 2 * |
| 《科技广场》 20090731 柏春岚 基于小波函数的航空遥感图像压缩研究 1-2 , 2 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102510444A (en) * | 2011-11-23 | 2012-06-20 | 青岛市光电工程技术研究院 | Medium-frequency and high-frequency MTF (modulation transfer function) online compensating and reinforcing method of optical image camera |
| CN102510444B (en) * | 2011-11-23 | 2013-12-18 | 青岛市光电工程技术研究院 | Medium-frequency and high-frequency MTF (modulation transfer function) online compensating and reinforcing method of optical image camera |
| CN103983343A (en) * | 2014-05-29 | 2014-08-13 | 武汉大学 | Satellite platform chattering detection method and system based on multispectral image |
| CN103983343B (en) * | 2014-05-29 | 2016-05-11 | 武汉大学 | A kind of satellite platform based on multispectral image tremble detection method and system |
| CN106022354A (en) * | 2016-05-07 | 2016-10-12 | 浙江大学 | SVM-based image MTF measurement method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102044068B (en) | 2012-05-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110163815B (en) | Low-illumination reduction method based on multi-stage variational self-encoder | |
| Al‐Ameen | Nighttime image enhancement using a new illumination boost algorithm | |
| CN101441764B (en) | MTFC remote sensing image restoration method | |
| Feng et al. | Speckle reduction via higher order total variation approach | |
| CN111583123A (en) | Wavelet transform-based image enhancement algorithm for fusing high-frequency and low-frequency information | |
| CN110533607B (en) | Image processing method and device based on deep learning and electronic equipment | |
| CN106204447A (en) | The super resolution ratio reconstruction method with convolutional neural networks is divided based on total variance | |
| CN106127688B (en) | A kind of super-resolution image reconstruction method and its system | |
| US10580122B2 (en) | Method and system for image enhancement | |
| CN107862666A (en) | Mixing Enhancement Methods about Satellite Images based on NSST domains | |
| CN104023166A (en) | Environment self-adaptation video image de-noising method and device | |
| CN111583113A (en) | Infrared image super-resolution reconstruction method based on generation countermeasure network | |
| CN102044068B (en) | Compensation method of wavelet modulation transfer function based on optimal kernel shape | |
| Li et al. | A novel brain image enhancement method based on nonsubsampled contourlet transform | |
| CN113052775A (en) | Image shadow removing method and device | |
| CN111383187B (en) | Image processing method and device and intelligent terminal | |
| WO2020248706A1 (en) | Image processing method, device, computer storage medium, and terminal | |
| CN108492264B (en) | Single-frame image fast super-resolution method based on sigmoid transformation | |
| CN106447616A (en) | Method and device for realizing wavelet de-noising | |
| CN113554615B (en) | Image refinement processing method and device, electronic equipment and storage medium | |
| CN111598115A (en) | SAR image fusion method based on cross cortical neural network model | |
| CN115619682A (en) | Deep learning-based denoising tone mapping method and device | |
| CN114743225A (en) | Retinex-ResNet network model-based fingerprint image enhancement method | |
| Al-Ameen | Improving the contrast of aerial images using a new multi-concept algorithm | |
| CN112967208A (en) | Image processing method and device, electronic equipment and storage medium |
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 | ||
| 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: 20120509 Termination date: 20211112 |