CN109658357A - A kind of denoising method towards remote sensing satellite image - Google Patents

A kind of denoising method towards remote sensing satellite image Download PDF

Info

Publication number
CN109658357A
CN109658357A CN201811579851.0A CN201811579851A CN109658357A CN 109658357 A CN109658357 A CN 109658357A CN 201811579851 A CN201811579851 A CN 201811579851A CN 109658357 A CN109658357 A CN 109658357A
Authority
CN
China
Prior art keywords
image
denoising
pixel
remote sensing
sensing satellite
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.)
Pending
Application number
CN201811579851.0A
Other languages
Chinese (zh)
Inventor
高昆
李若娴
焦建超
韩璐
苏云
张晓典
王俊伟
张宇桐
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Institute of Technology BIT
Original Assignee
Beijing Institute of Technology BIT
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Institute of Technology BIT filed Critical Beijing Institute of Technology BIT
Priority to CN201811579851.0A priority Critical patent/CN109658357A/en
Publication of CN109658357A publication Critical patent/CN109658357A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00Image enhancement or restoration
    • G06T5/001Image restoration
    • G06T5/002Denoising; Smoothing
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10032Satellite or aerial image; Remote sensing
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20004Adaptive image processing

Abstract

The invention discloses a kind of denoising methods towards remote sensing satellite image, including are loaded into original remote sensing satellite image, obtain image to be denoised using Adaptive contrast enhancement method enhancing picture contrast;Image to be denoised is divided into M × N bounded discretization grid, obtain it is discrete after image to be denoised;The impulsive noise wait denoise in image after discrete is detected to obtain two-dimentional identity matrix F using intermediate value absolute difference decision criteria, pixel and the pixel of image to be denoised correspond in two-dimentional identity matrix F;Search window and neighborhood window are being set in image wait denoise after discrete;Window be slidably connected in the search window similarity calculation and Gauss weighted average calculation method in field carries out denoising, image after being denoised.The present invention can be realized excellent grain details holding and good image denoising effect, while algorithm complexity is lower.Treatment effeciency is higher.

Description

A kind of denoising method towards remote sensing satellite image
Technical field
The present invention relates to remote sensing satellite image processing technology fields, more particularly, to one kind towards remote sensing satellite image Denoising method.
Background technique
With the development of remote sensing technology, the target image obtained by remote sensing satellite is in military surveillance early warning, land resource GeneraI investigation, disaster monitoring, environmental monitoring, engineering construction and planning etc. are used widely.Based on the application, the mankind for The type and quantity demand of remote sensing images are gradually increased.However, people do not pursue data volume, remote sensing simply in practical applications The quality of image be only more directly influence obtain information accuracy and reliability factor, thus picture quality increasingly at For research focus.
An important factor for noise is generation image interference, during obtaining and transmitting remote sensing satellite image, image matter Amount will receive the adverse effect of various noises.Therefore, field of remote sensing image processing is become for the research of image de-noising method Important research direction.
Therefore it provides it is a kind of can be realized excellent grain details keep and good image denoising effect towards remote sensing The denoising method of satellite image is this field technical problem urgently to be resolved.
Summary of the invention
In view of this, it is excellent to solve realization the present invention provides a kind of denoising method towards remote sensing satellite image The technical issues of grain details holding and good image denoising effect.
In order to solve the above technical problem, the present invention provides a kind of denoising methods towards remote sensing satellite image, comprising:
It is loaded into original remote sensing satellite image, is obtained using Adaptive contrast enhancement method enhancing picture contrast wait denoise Image;
The image to be denoised is divided into M × N bounded discretization grid, obtain it is discrete after the image to be denoised, v =v (i, j) | and i ∈ M, j ∈ N }, v (i, j) indicates image pixel value;
The impulsive noise described in after discrete wait denoise in image is detected using intermediate value absolute difference decision criteria To two-dimentional identity matrix F, pixel and the pixel of the image to be denoised are corresponded in the two dimension identity matrix F, wherein in It is worth absolute difference decision criteria and uses following calculation formula:
Q (i, j)=| v (i, j)-MEDΩ(i, j) |, Ω is the neighborhood centered on v (i, j), MEDΩ(i, j) is institute in Ω There is the intermediate value of pixel grey scale;
Search window and neighborhood window are set in image wait denoise described in after discrete, wherein described search window with Centered on reference pixel v (i, j), i ∈ M, j ∈ N, with DsFor radius, described search window size is D × D, D=2Ds+1;It is described Neighborhood window is centered on pixel v (k, l) to be processed, k ∈ M, k ≠ i;L ∈ N, l ≠ j, with dsFor radius, the neighborhood window Size is d × d, d=2ds+1;
The field window is slidably connected similarity calculation and Gauss weighted average calculation method in described search window Carry out denoising, image u=after being denoised u (i, j) | i ∈ M, j ∈ N };Wherein, the denoising includes:
The pixel v (i, j) and the corresponding v with v (i, j) of the corresponding position of F (i, j)=1 in the two-dimentional identity matrix F (k, l), without similarity calculation and Gauss weighted average calculation;
V (i, j) and the intermediate value absolute difference of v (k, l) in the neighborhood window size are calculated, when the intermediate value absolute difference is big When decision threshold, v (i, j) is without similarity calculation and Gauss weighted average calculation;
The reference pixel and the pixel to be processed are without similarity calculation and Gauss weighted average calculation.
Optionally, it is loaded into original image, picture contrast is enhanced using Adaptive contrast enhancement method, is further wrapped It includes: picture contrast is enhanced using following formula,
Wherein, I (i, j) indicates image in the gray scale of the position (i, j), and μ is population mean, and σ is population standard deviation, y (i, j) For enhanced image grayscale, a and b are respectively the numerical value set according to feature of image.
Optionally, the impulsive noise in the image to be detected after discrete is carried out using intermediate value absolute difference decision criteria Detection obtains two-dimentional identity matrix F, comprising:
Decision threshold TH is set, when q (i, j) is greater than TH, then v (i, j) is impulsive noise pixel, enables F (i, j)=1;When When q (i, j) is less than or equal to TH, then v (i, j) is non-pulse noise pixel, enables F (i, j)=0.
Optionally, the calculation formula of image u=after the denoising { u (i, j) | i ∈ M, j ∈ N } includes:
Wherein, w (v1,v2) indicate two windows between similarity degree, v1Represent v (i, j), v2It represents v (k, l), V (i, j) Wait denoise in image centered on v (i, j) described in indicating after discrete, size is fixed as each pixel in the region of d × d Value;V (k, l) indicate after discrete described in wait denoise in image centered on v (k, l), size is fixed as the region of d × d Interior each pixel value, when being calculated, the location of pixels of V (i, j) and V (k, l) are corresponding.
Optionally,
Wherein, z () is the normalized parameter that weight is normalized;||·||2,α 2For Gauss weighted euclidean distance, α is Gaussian kernel standard deviation;H is similarity Gauss weighting parameters, and e is natural constant.
Optionally,
Optionally, D=15, d=3.
Optionally, decision threshold TH=85.
Compared with prior art, the denoising method provided by the invention towards remote sensing satellite image, at least realizes as follows The utility model has the advantages that
Denoising method provided by the invention carries out Adaptive contrast enhancement to original image first, and adaptive contrast increases Strong process can improve the visual effect of image, and adjustment image makes it be more suitable for people's observation and subsequent image analysis.Then to arteries and veins It rushes noise to be detected, the impulsive noise detected will not calculated in the follow-up process, to realize that removal pulse is made an uproar The purpose of sound.The Gaussian noise in image is removed in the prior art and generallys use gaussian filtering method, and it is good that this method denoises effect It is good, but image border and grain details are difficult to retain.The non-local mean filtering developed on its basis can effectively remove figure As Gaussian noise, but when handling image edge information or grain details, it was easy to appear smooth phenomenon, and the above method needs Calculating time that will be very long.Using denoising process provided by the invention, image border and grain details can be effectively avoided It crosses smoothly, not only effectively removes the Gaussian noise in remote sensing satellite image, also filtered out the impulsive noise detected in image.This Invention image denoising effect with excellent grain details holding and more preferably, and algorithm complexity is reduced, it is big to calculate the time It is big to shorten.
By referring to the drawings to the detailed description of exemplary embodiment of the present invention, other feature of the invention and its Advantage will become apparent.
Detailed description of the invention
It is combined in the description and the attached drawing for constituting part of specification shows the embodiment of the present invention, and even With its explanation together principle for explaining the present invention.
Fig. 1 is the denoising method flow chart provided by the invention towards remote sensing satellite image;
Fig. 2 is the original image chosen;
Fig. 3 is to add the noise image obtained after noise to original image in Fig. 2;
Fig. 4 is noise image contrast reinforcing effect figure in Fig. 3;
Fig. 5 is image after the denoising that noise image is obtained after being handled using denoising method provided by the invention in Fig. 3.
Specific embodiment
Carry out the various exemplary embodiments of detailed description of the present invention now with reference to attached drawing.It should also be noted that unless in addition having Body explanation, the unlimited system of component and the positioned opposite of step, numerical expression and the numerical value otherwise illustrated in these embodiments is originally The range of invention.
Be to the description only actually of at least one exemplary embodiment below it is illustrative, never as to the present invention And its application or any restrictions used.
Technology, method and apparatus known to person of ordinary skill in the relevant may be not discussed in detail, but suitable In the case of, the technology, method and apparatus should be considered as part of specification.
It is shown here and discuss all examples in, any occurrence should be construed as merely illustratively, without It is as limitation.Therefore, other examples of exemplary embodiment can have different values.
It should also be noted that similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi It is defined in a attached drawing, then in subsequent attached drawing does not need that it is further discussed.
The present invention provides a kind of denoising method towards remote sensing satellite image, and Fig. 1 defends to be provided by the invention towards remote sensing The denoising method flow chart of star chart picture.As shown in Figure 1, comprising:
Step S101: being loaded into original remote sensing satellite image, enhances picture contrast using Adaptive contrast enhancement method, To be easy to subsequent step progress.Enhancing picture contrast refers to the ash by all or part of pixels in certain rules modification image Degree, to change the operation of image grayscale dynamic range.The vision effect of image can be improved by picture superposition operation Fruit is conducive to subsequent image analysis.The image grayscale that photoelectric imaging sensor obtains on remote sensing satellite under normal conditions mainly collects In in a certain range centered on mean value, in the class Gaussian Profile of single peak form.
Optionally, the present invention enhances picture contrast using following formula,
Wherein, I (i, j) indicates image in the gray scale of the position (i, j), and μ is population mean, and σ is population standard deviation, y (i, j) For enhanced image grayscale, a and b are respectively the numerical value set according to feature of image, can repeatedly adjust a, b number in practice The selection of value carries out calculation processing, to obtain preferable treatment effect.In a kind of optional embodiment, a=0, b= 255。
The method for generalling use histogram equalization in the prior art carries out picture contrast before carrying out denoising Enhancing operation, but this method will not select data when handling remote sensing satellite image, and it is thin to may cause image The loss of section.The present invention uses the Adaptive contrast enhancement method of customized computation rule, and reinforcing effect is made to be more suitable for handling Object, to obtain better treatment effect.
Step S102: being divided into M × N bounded discretization grid for image to be denoised, obtain it is discrete after image to be denoised, V=v (i, j) | and i ∈ M, j ∈ N }, v (i, j) indicates image pixel value;
Step S103: the impulsive noise wait denoise in image after discrete is visited using intermediate value absolute difference decision criteria Two-dimentional identity matrix F is measured, pixel and the pixel of image to be denoised correspond in two-dimentional identity matrix F, wherein intermediate value is exhausted Following calculation formula is used to poor decision criteria:
Q (i, j)=| v (i, j)-MEDΩ(i, j) |, Ω is the neighborhood centered on v (i, j), MEDΩ(i, j) is institute in Ω There is the intermediate value of pixel grey scale;Q (i, j) indicates the value for the intermediate value absolute difference being calculated.
Two dimension identity matrix F is for recording wait denoise impulsive noise pixel and non-pulse noise picture in image in the present invention Element.Optionally, two-dimentional identity matrix F can be established according to following rule:
The decision threshold TH for setting impulsive noise first, when q (i, j) is greater than TH, then the corresponding pixel identification of v (i, j) For impulsive noise pixel, F (i, j)=1 is enabled;When q (i, j) is less than or equal to TH, then the corresponding pixel of v (i, j) regards as non-arteries and veins Noise pixel is rushed, F (i, j)=0 is enabled.
Optionally, decision threshold TH=85.
Step S104: search window and neighborhood window are being set in image wait denoise after discrete, wherein search window with Centered on reference pixel v (i, j), i ∈ M, j ∈ N, with DsFor radius, search box size is D × D, D=2Ds+1;Neighborhood window Centered on pixel v (k, l) to be processed, k ∈ M, k ≠ i;L ∈ N, l ≠ j, with dsFor radius, neighborhood window size is d × d, d =2ds+1。
Optionally, D=15, d=3.I.e. search box size is 15 × 15, and neighborhood window size is 3 × 3.
Step S105: field window is slidably connected similarity calculation and Gauss weighted average calculation method in the search window Carry out denoising, image u=after being denoised u (i, j) | i ∈ M, j ∈ N };Wherein, denoising includes:
Wherein, the present invention carries out denoising using improved non-local mean denoising method, and image can pass through after denoising Following formula calculates:
Wherein, w (v1,v2) indicate two windows between similarity degree, v1Represent v (i, j), v2It represents v (k, l), V (i, j) It indicates, wait denoise in image centered on v (i, j), size is fixed as each pixel value in the region of d × d after discrete;V (k, l) is indicated, wait denoise in image centered on v (k, l), size is fixed as each picture in the region of d × d after discrete Element value, when being calculated, the location of pixels of V (i, j) and V (k, l) are corresponding.Z () is to return to what weight was normalized One changes parameter;||·||2,α 2For Gauss weighted euclidean distance, α is Gaussian kernel standard deviation;H is similarity Gauss weighting parameters, e For natural constant.Optionally, h=10 σ.
In the denoising method that uses of the present invention, in order to guarantee to differ with reference pixel gray scale biggish impulsive noise pixel, Edge pixel etc. avoids the detailed information such as image border, complex texture by excess smoothness without weighted average calculation, realization Reason.During processing:
The pixel v (i, j) of the corresponding position of F (i, j)=1 and the corresponding v (k, l) with v (i, j) in two-dimentional identity matrix F, Without similarity calculation and Gauss weighted average calculation;It is identified as the reference pixel of impulsive noise pixel and its corresponding Pixel to be processed, without similarity calculation and Gauss weighted average calculation.
V (i, j) and the intermediate value absolute difference of v (k, l) in neighborhood window size are calculated, when intermediate value absolute difference is greater than decision threshold When being worth (TH), v (i, j) is without similarity calculation and Gauss weighted average calculation;
As search window and the reference pixel of neighborhood window center and pixel to be processed without similarity calculation and height This weighted average calculation.
Denoising method provided by the invention carries out Adaptive contrast enhancement to original remote sensing satellite image first, adaptively Contrast enhancing process can improve the visual effect of image, and adjustment image makes it be more suitable for people's observation and subsequent image analysis. Then impulsive noise is detected, the impulsive noise detected will not be calculated in the follow-up process, gone to realize Except the purpose of impulsive noise.The Gaussian noise removed in image in the prior art generallys use gaussian filtering method, and this method is gone It makes an uproar and works well, but image border and grain details are difficult to retain.The non-local mean filtering developed on its basis can have Effect removal image Gaussian noise, but when handling image edge information or grain details, it was easy to appear smooth phenomenon, Er Qieshang The calculating time that the method for stating needs to grow very much.Using denoising process provided by the invention, can effectively avoid image border and Grain details are excessively smooth, not only effectively remove the Gaussian noise in remote sensing satellite image, also filtered out the arteries and veins detected in image Rush noise.Present invention image denoising effect with excellent grain details holding and more preferably, and algorithm complexity is reduced, it counts Evaluation time greatly shortens.
In order to prove the validity of denoising method provided by the invention, inventor is simulated application test, and Fig. 2 is choosing The original image taken, Fig. 3 are that the noise image obtained after noise is added to original image in Fig. 2, and Fig. 4 is noise image in Fig. 3 Contrast reinforcing effect figure, after the denoising that Fig. 5 is obtained after being handled for noise image in Fig. 3 using denoising method provided by the invention Image.
In order to simulate the image effect of remote sensing satellite acquisition, in original image shown in Fig. 2 first addition intensity be σ= 15, p=2.5% Gauss-pulse mixed noise, wherein Gaussian noise is stable, 0 mean value white noise, and σ is in image Standard deviation, p be impulsive noise occur probability.Obtain that there is Gauss-pulse mixed noise image to be processed, such as Fig. 3 It is shown, can obvious sensation noise it is larger.Using denoising method provided by the invention, enhancing figure then is carried out to image in Fig. 3 Image contrast obtains effect picture shown in Fig. 4.Then it is slidably connected in the search window similarity meter by setting field window It calculates and Gauss weighted average calculation method carries out denoising, obtain image after denoising shown in fig. 5.The letter of analogue noise image Making an uproar than (SNR) is 20.7188dB, and the SNR of image is 37.6208dB after denoising method of the present invention processing, it is seen that image noise Than getting a promotion, thus denoising method of the invention has good denoising performance, and good denoising effect can be observed in human eye.
Through the foregoing embodiment it is found that the denoising method provided by the invention towards remote sensing satellite image, at least realizes It is following the utility model has the advantages that
Denoising method provided by the invention carries out Adaptive contrast enhancement to original remote sensing satellite image first, adaptively Contrast enhancing process can improve the visual effect of image, and adjustment image makes it be more suitable for people's observation and subsequent image analysis. Then impulsive noise is detected, the impulsive noise detected will not be calculated in the follow-up process, gone to realize Except the purpose of impulsive noise.The Gaussian noise removed in image in the prior art generallys use gaussian filtering method, and this method is gone It makes an uproar and works well, but image border and grain details are difficult to retain.The non-local mean filtering developed on its basis can have Effect removal image Gaussian noise, but when handling image edge information or grain details, it was easy to appear smooth phenomenon, Er Qieshang The calculating time that the method for stating needs to grow very much.Using denoising process provided by the invention, can effectively avoid image border and Grain details are excessively smooth, not only effectively remove the Gaussian noise in remote sensing satellite image, also filtered out the arteries and veins detected in image Rush noise.Present invention image denoising effect with excellent grain details holding and more preferably, and algorithm complexity is reduced, it counts Evaluation time greatly shortens.
Although some specific embodiments of the invention are described in detail by example, the skill of this field Art personnel it should be understood that example above merely to being illustrated, the range being not intended to be limiting of the invention.The skill of this field Art personnel are it should be understood that can without departing from the scope and spirit of the present invention modify to above embodiments.This hair Bright range is defined by the following claims.

Claims (8)

1. a kind of denoising method towards remote sensing satellite image characterized by comprising
It is loaded into original remote sensing satellite image, figure to be denoised is obtained using Adaptive contrast enhancement method enhancing picture contrast Picture;
The image to be denoised is divided into M × N bounded discretization grid, obtain it is discrete after the image to be denoised, v={ v (i, j) | i ∈ M, j ∈ N }, v (i, j) indicates image pixel value;
The impulsive noise described in after discrete wait denoise in image is detected to obtain two using intermediate value absolute difference decision criteria Identity matrix F is tieed up, pixel and the pixel of the image to be denoised correspond in the two dimension identity matrix F, wherein intermediate value is exhausted Following calculation formula is used to poor decision criteria:
Q (i, j)=| v (i, j)-MEDΩ(i, j) |, Ω is the neighborhood centered on v (i, j), MEDΩ(i, j) is all pictures in Ω The intermediate value of plain gray scale;
Search window and neighborhood window are set in image wait denoise described in after discrete, wherein described search window is to refer to Centered on pixel v (i, j), i ∈ M, j ∈ N, with DsFor radius, described search window size is D × D, D=2Ds+1;The neighborhood Window is centered on pixel v (k, l) to be processed, k ∈ M, k ≠ i;L ∈ N, l ≠ j, with dsFor radius, the neighborhood window size For d × d, d=2ds+1;
Be slidably connected in described search window similarity calculation and Gauss weighted average calculation method of the field window carries out Denoising, image u=after being denoised u (i, j) | i ∈ M, j ∈ N };Wherein, the denoising includes:
The pixel v (i, j) of the corresponding position of F (i, j)=1 and the corresponding v (k, l) with v (i, j) in the two-dimentional identity matrix F, Without similarity calculation and Gauss weighted average calculation;
V (i, j) and the intermediate value absolute difference of v (k, l) in the neighborhood window size are calculated, is sentenced when the intermediate value absolute difference is greater than When determining threshold value, v (i, j) is without similarity calculation and Gauss weighted average calculation;
The reference pixel and the pixel to be processed are without similarity calculation and Gauss weighted average calculation.
2. the denoising method according to claim 1 towards remote sensing satellite image, which is characterized in that
It is loaded into original image, picture contrast is enhanced using Adaptive contrast enhancement method, further comprises: using following public Formula enhances picture contrast,
Wherein, I (i, j) indicates image in the gray scale of the position (i, j), and μ is population mean, and σ is population standard deviation, and y (i, j) is to increase Image grayscale after strong, a and b are respectively the numerical value set according to feature of image.
3. the denoising method according to claim 1 towards remote sensing satellite image, which is characterized in that
The impulsive noise in the image to be detected after discrete is detected to obtain two using intermediate value absolute difference decision criteria Tie up identity matrix F, comprising:
Decision threshold TH is set, when q (i, j) is greater than TH, then v (i, j) is impulsive noise pixel, enables F (i, j)=1;When q (i, When j) being less than or equal to TH, then v (i, j) is non-pulse noise pixel, enables F (i, j)=0.
4. the denoising method according to claim 1 towards remote sensing satellite image, which is characterized in that
The calculation formula of image u=after the denoising { u (i, j) | i ∈ M, j ∈ N } includes:
Wherein, w (v1,v2) indicate two windows between similarity degree, v1Represent v (i, j), v2It represents v (k, l), V (i, j) is indicated Wait denoise in image centered on v (i, j) described in after discrete, size is fixed as each pixel value in the region of d × d;V (k, l) indicate after discrete described in wait denoise in image centered on v (k, l), size is fixed as in the region of d × d respectively A pixel value, when being calculated, the location of pixels of V (i, j) and V (k, l) are corresponding.
5. the denoising method according to claim 4 towards remote sensing satellite image, which is characterized in that
Wherein, z () is the normalized parameter that weight is normalized;||·||2,α 2For Gauss weighted euclidean distance, α is height This core standard deviation;H is similarity Gauss weighting parameters, and e is natural constant.
6. the denoising method according to claim 5 towards remote sensing satellite image, which is characterized in that
7. the denoising method according to claim 1 towards remote sensing satellite image, which is characterized in that D=15, d=3.
8. the denoising method according to claim 1 towards remote sensing satellite image, it is characterised in that decision threshold TH=85.
CN201811579851.0A 2018-12-24 2018-12-24 A kind of denoising method towards remote sensing satellite image Pending CN109658357A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811579851.0A CN109658357A (en) 2018-12-24 2018-12-24 A kind of denoising method towards remote sensing satellite image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811579851.0A CN109658357A (en) 2018-12-24 2018-12-24 A kind of denoising method towards remote sensing satellite image

Publications (1)

Publication Number Publication Date
CN109658357A true CN109658357A (en) 2019-04-19

Family

ID=66115792

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811579851.0A Pending CN109658357A (en) 2018-12-24 2018-12-24 A kind of denoising method towards remote sensing satellite image

Country Status (1)

Country Link
CN (1) CN109658357A (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102750675A (en) * 2012-05-31 2012-10-24 华中科技大学 Non-local means filtering method for speckle noise pollution image
CN106815818A (en) * 2017-01-17 2017-06-09 中国科学院上海高等研究院 A kind of image de-noising method
US10096109B1 (en) * 2017-03-31 2018-10-09 The Board Of Trustees Of The Leland Stanford Junior University Quality of medical images using multi-contrast and deep learning

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102750675A (en) * 2012-05-31 2012-10-24 华中科技大学 Non-local means filtering method for speckle noise pollution image
CN106815818A (en) * 2017-01-17 2017-06-09 中国科学院上海高等研究院 A kind of image de-noising method
US10096109B1 (en) * 2017-03-31 2018-10-09 The Board Of Trustees Of The Leland Stanford Junior University Quality of medical images using multi-contrast and deep learning

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
BARTOMEU COLL ETC.: ""A non-local algorithm for image denoising"", 《IEEE COMPUTER SOCIETY CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》 *
刘迎辉等: ""基于邻域特征匹配的通用噪声滤波器"", 《基于邻域特征匹配的通用噪声滤波器》 *
胡谋法等: ""一种实用的可见光图像自适应对比度增强方法"", 《光学与光电技术》 *

Similar Documents

Publication Publication Date Title
Zhu et al. Removing atmospheric turbulence via space-invariant deconvolution
Wang et al. Recent advances in image dehazing
US20170280073A1 (en) Systems and Methods for Reducing Noise in Video Streams
Wang et al. A unified tensor level set for image segmentation
Yan et al. Anomaly detection in images with smooth background via smooth-sparse decomposition
Zhao et al. Fuzzy c-means clustering with non local spatial information for noisy image segmentation
Kenig et al. Blind image deconvolution using machine learning for three-dimensional microscopy
CN104851086B (en) A kind of image detecting method for cable surface defect
CN107153816B (en) Data enhancement method for robust face recognition
Fan et al. A hierarchical image matting model for blood vessel segmentation in fundus images
US20130251246A1 (en) Method and a device for training a pose classifier and an object classifier, a method and a device for object detection
CN104200442B (en) Non-local mean MRI image denoising method based on improved canny rim detections
CN105976330B (en) A kind of embedded greasy weather real time video image stabilization
JP2008511366A (en) Feature-weighted medical object contour detection using distance coordinates
Huang et al. Faster R-CNN for marine organisms detection and recognition using data augmentation
Morillas et al. New adaptive vector filter using fuzzy metrics
US8103055B2 (en) Detection of blobs in images
CN104978715B (en) A kind of non-local mean image de-noising method based on filter window and parameter adaptive
Wei et al. Tensor voting guided mesh denoising
CN101996406A (en) No-reference structural sharpness image quality evaluation method
KR20130127654A (en) Image processing apparatus for moving image haze removal and method using that
CN104657945B (en) The infrared small target detection method of multiple dimensioned space-time Federated filter under complex background
US20150279021A1 (en) Video object tracking in traffic monitoring
CN103369209A (en) Video noise reduction device and video noise reduction method
CN105787889B (en) A kind of Fast Image Denoising based on non-local mean

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20190419

RJ01 Rejection of invention patent application after publication