CN110363714A - The asymmetric correction method based on scene interframe registration of adjusting learning rate - Google Patents
The asymmetric correction method based on scene interframe registration of adjusting learning rate Download PDFInfo
- Publication number
- CN110363714A CN110363714A CN201910532213.1A CN201910532213A CN110363714A CN 110363714 A CN110363714 A CN 110363714A CN 201910532213 A CN201910532213 A CN 201910532213A CN 110363714 A CN110363714 A CN 110363714A
- Authority
- CN
- China
- Prior art keywords
- frame
- infrared image
- kth
- pixel
- value
- 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
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000006073 displacement reaction Methods 0.000 claims abstract description 16
- 239000011159 matrix material Substances 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 5
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 230000006399 behavior Effects 0.000 claims 1
- 230000004044 response Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000005855 radiation Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000005314 correlation function Methods 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 238000003331 infrared imaging Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000005622 photoelectricity Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000026676 system process Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
- G06T5/94—Dynamic range modification of images or parts thereof based on local image properties, e.g. for local contrast enhancement
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/80—Geometric correction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/38—Registration of image sequences
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20004—Adaptive image processing
- G06T2207/20012—Locally adaptive
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Transforming Light Signals Into Electric Signals (AREA)
Abstract
The invention discloses a kind of asymmetric correction methods based on scene interframe registration of adjusting learning rate, computing cross-correlation is carried out by the ranks projection vector to adjacent two field pictures, obtain image be expert at, the displacement of column direction, and then obtain the overlapping region of two field pictures, further according to the error and learning rate of two field pictures overlapping region, update is iterated to correction parameter.Wherein, learning rate depends on the spatial feature of image overlapping region, and higher learning rate is used in flat site, uses lower learning rate in details and the more region in edge.The method of the present invention has faster convergence rate compared with the asymmetric correction method based on scene for using fixed learning rate, while can effectively inhibit the generation of ghost phenomenon.
Description
Technical field
The invention belongs to Infrared Image Non-uniformity Correction fields, and in particular to a kind of adjusting learning rate based on field
The asymmetric correction method of scape interframe registration.
Background technique
Currently, infrared image has been widely used in the fields such as industry, medicine and military affairs to carry out detecing under low visibility
It surveys.In the ideal case, infrared imaging system is to the infrared light of homogeneous radiation, the ash of each pixel in the digital picture of acquisition
Angle value should be just the same.But in fact, being limited to the manufacturing process of solid-state electronic, the light-sensitive element (pixel) on detector is past
Toward uneven, the thickness etc. along with impurity concentration, effective photosensitive area is not accomplished the problems such as being absolutely averaged, the photoelectricity between pixel
Transfer efficiency is different, uneven to the imaging of the scenery of homogeneous radiation.In addition, between each channel of image data reading circuit
Difference, will lead to image occur be in column distribution fixation fringes noise.This requires our images to carry out Nonuniformity Correction, makes
Image obtains better visual effect.
Common Non Uniformity Correction of Infrared Image technology mainly has scaling method and two kinds of scene method.
Based on determining the technology that calibration method is applicability at present.But with the variation of external environment, such as focal plane temperature
The variation of degree, bias voltage, will lead to infrared focus plane response characteristic and drifts about, so that occurring on image after scaling correction new
FPN, reduce image quality, be originally used for correction parameter then be not suitable with variation after heterogeneity.It needs in actual operation
Periodically calibrate to update correction parameter.
Therefore, the Nonuniformity Correction based on scene in recent years, many scholars begin one's study.Based on scene
Reference source is not only omitted in correcting algorithm, is simplified system process flow, improves the stability of system, and can be with
The influence for effectively eliminating parameter characteristic drift, realizes the adaptive nonuniformity correction of high-precision, Larger Dynamic range.Based on scene
The algorithm of class is divided into constant constant statistics method, neural network algorithm and interframe method for registering etc..However neural network nonuniformity
Correcting algorithm is difficult the convergence rate with higher while guaranteeing calibration result, and engineer application difficulty is higher, and when figure
As it is static when, just will appear serious decay;Constant statistics method is stronger to the dependence of scene random motion, and is easy by field
Scape interference, the stability and convergence rate of algorithm cannot be taken into account;Interframe is registrated this method and is respectively detected according to detector in the short time
Member to the response of same scene should this consistent characteristic, obtain different pixels to same scene spoke using the mode of image registration
The response penetrated.Calculating is simpler, is easy to engineer application, but often cannot be considered in terms of convergence rate and stability.
Summary of the invention
The purpose of the present invention is to provide a kind of heterogeneity schools based on scene interframe registration of adjusting learning rate
Correction method, the asymmetric correction method convergence rate for solving to be generally basede on scene interframe registration is slow, and is easy to appear ghost
The problem of shadow.
The technical solution for realizing the aim of the invention is as follows: a kind of adjusting learning rate based on scene interframe registration
Asymmetric correction method realizes that steps are as follows:
Step 1, the infrared image sequence that real-time moving scene is obtained by thermal infrared imager are chosen wherein adjacent every time
Two frame infrared images;
Step 2, the mean value for calculating separately every a line all pixels value gray scale in above-mentioned two frames infrared image, and subtract respectively whole
The mean value of width gray value of image obtains the row projection value of every row;The mean value of above-mentioned each column all pixels value gray scale is calculated separately,
And subtract the mean value of entire image gray value respectively, obtain the column projection value of each column;
Step 3 makees computing cross-correlation to the row, column projection value in above-mentioned two frames infrared image;
The displacement of step 4, row, column direction when finding out the cross-correlation function maximum for making row and column direction, as two frames are red
The relative displacement of outer image.
Step 5, adjusting learning rate when obtaining the pixel correction according to the local variance of each pixel.
The row, column shift value of step 6, the two frame infrared images according to obtained in step 4, it can be deduced that two frame infrared images
Overlapping region, to overlapping one error matrix of regional structure.
Step 7 is updated correction parameter matrix;
Step 8 adds correction parameter with original infrared image, obtains output image.
Step 9, in chronological sequence sequence, two adjacent frames all to infrared image sequence, i.e. the 1st frame and the 2nd frame, the 2nd
Frame and the 3rd frame ..., -1 frame of kth and kth frame ..., are carried out the operation of step 2 to step 8, always to correction parameter matrix into
Row iteration updates.
Compared with prior art, the present invention its significant advantage are as follows:
(1) workload for reducing scaling method avoids repeatedly demarcating repeatedly.
(2) it only needs tens frames even more than ten frames that can restrain, substantially increases the rate of correction;
(3) due to using different renewal rates to different zones, general scene method is avoided due to using fixed study
The serious ghost problem of rate bring and the slow problem of convergence.
Detailed description of the invention
Fig. 1 is a kind of stream of the asymmetric correction method based on scene interframe registration of adjusting learning rate of the present invention
Cheng Tu.
Fig. 2 (a) is the 100th frame image having in heteropical original infrared image sequence;Fig. 2 (b) is using solid
The asymmetric correction method based on scene interframe registration for determining learning rate carries out that treated the 100th to original sequence
Frame image;Fig. 2 (c) is the asymmetric correction method based on scene interframe registration using adjusting learning rate of the invention
Carry out treated the 100th frame image.
Specific embodiment
It is further described with reference to the accompanying drawing.
The present invention is a kind of asymmetric correction method based on scene interframe registration of adjusting learning rate, Ke Yitong
The iteration for crossing multiframe is removed the heterogeneity in infrared image sequence.Its principle are as follows: first in infrared image sequence
The ranks projection vector of front and back two field pictures calculates cross-correlation information, then finds respectively so that cross-correlation information was maximized
Displacement horizontally and vertically, the as shift value of image, the overlapping region being superimposed, the root in overlapping region
The update of parameter is corrected to image according to learning rate, completes Nonuniformity Correction.Scene interframe is based on general herein
On the basis of the asymmetric correction method of registration, a kind of improved method is proposed, according to the local variance of each pixel, is calculated
Adaptive learning rate out can generate while improving convergence rate to avoid ghost.
In conjunction with Fig. 1, a kind of asymmetric correction method based on scene interframe registration of adjusting learning rate, including with
Lower step:
Step 1, the real-time infrared image sequence obtained by thermal infrared imager, the resolution ratio of infrared image are M × N,
Wherein M is line number, and N is columns;The correction parameter matrix O that a size is M × N is constructed, the correction ginseng of each pixel is stored
Number, initial value 0 choose adjacent two frames infrared image, such as kth frame and kth+1 for real-time infrared image sequence every time
Frame, k >=1;
Step 2, the mean value for calculating separately every a line all pixels value gray scale in+1 frame infrared image of above-mentioned kth frame and kth,
And subtract the mean value of corresponding whole picture infrared image gray value respectively, obtain the row projection value of every row;Calculate separately above-mentioned each column
The mean value of all pixels value gray scale, and subtract the mean value of corresponding whole picture infrared image gray value respectively, obtain the column projection of each column
Value, calculation formula are as follows:
Wherein i indicates that the row coordinate of pixel, j indicate the column coordinate of pixel,Indicate that the row of the i-th row of infrared image is thrown
Shadow value,Indicate the column projection value of infrared image jth column, XnCoordinate is the pixel of (i, j) in (i, j) expression n-th frame
Gray value;N indicates frame number, enables n=k or n=k+1;
Step 3 carries out computing cross-correlation to the row, column projection value of+1 frame infrared image of kth frame and kth:
Wherein, displacement h ∈ [0,2 × Δ of horizontal directioncol], displacement v ∈ [0,2 × Δ of vertical directionrow], ΔcolFor
+ 1 frame infrared image of kth is relative to the kth frame infrared image displacement upper limit in the horizontal direction, ΔrowFor+1 frame infrared image of kth
Relative to kth frame infrared image vertical direction the displacement upper limit;Crow(v) believe for the row cross-correlation of adjacent two frames infrared image
Cease function, CcolIt (h) is the column cross-correlation information function of adjacent two frames infrared image;
Step 4 finds and makes Crow(v) and Ccol(h) corresponding v when being maximized, h value find out+1 frame infrared image phase of kth
For kth frame infrared image line direction displacement di and column direction displacement dj:
Step 5, it is general based on scene interframe registration asymmetric correction method all pixels are all used it is same solid
Fixed learning rate when learning rate sets high, can generate ghost phenomenon, and when learning rate sets low, convergence rate is very slow, because
This can not obtain good balance in convergence rate and stability;The present invention is according to the local variance of each pixel, to each
Pixel calculates adjusting learning rate, takes into account convergence rate and stability, the specific steps are as follows:
The local mean value m (i, j) of each pixel 5-1) is calculated, centered on locally referring to the coordinate (i, j) with pixel, window is big
The small region for (2l+1) × (2l+1), wherein l is a preset integer, and X (a, b) indicates the gray value of pixel, is calculated public
Formula is as follows:
5-2) calculate the local variance sigma of each pixel2(i, j), calculation formula are as follows:
The learning rate of each pixel 5-3) is calculated according to local variance, calculation formula is as follows:
Wherein, αstFor initial learning rate, α (i, j) is the learning rate for the pixel that coordinate is (i, j).Use this public affairs
The calculated learning rate of formula, it is higher in image flat site, convergence rate can be improved, image detail and fringe region compared with
It is low, it can be generated to avoid ghost.
Row, column the shift value di and dj of+1 frame infrared image of step 6, the kth frame according to obtained in step 4 and kth, obtain
The overlapping region of adjacent two frames infrared image, to overlapping one error matrix ERR of regional structurek+1(i,j)2:
Wherein, Xk(i, j) indicates gray value of the coordinate for the pixel of (i, j), O in kth frame infrared imagek(i, j) is indicated
It is the value of (i, j) with coordinate in kth -1 and the calculated correction parameter matrix of kth frame, wherein k-1 minimum is equal to 1.
Step 7 is updated correction parameter matrix:
Ok+1(i, j)=Ok(i,j)-2×α(i,j)×ERRk+1(i,j)
Step 8 exports image Y after obtaining the correction of+1 frame of kth plus correction parameter matrix with+1 frame infrared image of kthk(i,
J):
Yk+1(i, j)=Xk+1(i,j)+Ok+1(i,j)。
Step 9, in chronological sequence sequence, two adjacent frames all to infrared image sequence, i.e. the 1st frame and the 2nd frame, the 2nd
Frame and the 3rd frame ..., -1 frame of kth and kth frame ..., are carried out the operation of step 2 to step 8, always to correction parameter matrix into
Row iteration updates.
For as shown in Fig. 2 (a) with the 100th frame image in heteropical original infrared image sequence, make
Original sequence is handled with the asymmetric correction method based on scene interframe registration of fixed learning rate, is obtained
The image as shown in Fig. 2 (b) it can clearly be seen that also remaining apparent heterogeneity, and produces ghost.And use this
The asymmetric correction method based on scene interframe registration for inventing the adjusting learning rate proposed is handled, obtained figure
As compared with Fig. 2 (b), it is more preferable to remove heteropical effect, and there is no ghost phenomenon as shown in Fig. 2 (c).
Claims (3)
1. a kind of asymmetric correction method based on scene interframe registration of adjusting learning rate, which is characterized in that including
Following steps:
Step 1, the real-time infrared image sequence obtained by thermal infrared imager, the resolution ratio of infrared image are M × N, wherein M
For line number, N is columns;The correction parameter matrix O that a size is M × N is constructed, the correction parameter of each pixel is stored,
Initial value is 0, for real-time infrared image sequence, chooses adjacent two frames infrared image every time, such as+1 frame of kth frame and kth, k >=
1;
Step 2, the mean value for calculating separately every a line all pixels value gray scale in+1 frame infrared image of above-mentioned kth frame and kth, and point
The mean value for not subtracting corresponding whole picture infrared image gray value obtains the row projection value of every row;It is all to calculate separately above-mentioned each column
The mean value of pixel value gray scale, and subtract the mean value of corresponding whole picture infrared image gray value respectively, the column projection value of each column is obtained, is counted
It is as follows to calculate formula:
Wherein i indicates that the row coordinate of pixel, j indicate the column coordinate of pixel,Indicate the row projection value of the i-th row of infrared image,Indicate the column projection value of infrared image jth column, Xn(i, j) indicates that coordinate is the gray scale of the pixel of (i, j) in n-th frame
Value;N indicates frame number, enables n=k or n=k+1;
Step 3 carries out computing cross-correlation to the row, column projection value of+1 frame infrared image of kth frame and kth:
Wherein, displacement h ∈ [0,2 × Δ of horizontal directioncol], displacement v ∈ [0,2 × Δ of vertical directionrow], ΔcolFor kth+
1 frame infrared image is relative to the kth frame infrared image displacement upper limit in the horizontal direction, ΔrowIt is opposite for+1 frame infrared image of kth
In kth frame infrared image vertical direction the displacement upper limit;CrowIt (v) is the row cross-correlation information letter of adjacent two frames infrared image
Number, CcolIt (h) is the column cross-correlation information function of adjacent two frames infrared image;
Step 4 finds and makes Crow(v) and Ccol(h) corresponding v when being maximized, h value, find out+1 frame infrared image of kth relative to
Displacement dj of the kth frame infrared image in the displacement di and column direction of line direction:
Step 5, obtained according to the local variance of each pixel the pixel update correction parameter when adjusting learning rate α (i,
j);
Row, column the shift value di and dj of+1 frame infrared image of step 6, the kth frame according to obtained in step 4 and kth, obtain adjacent
The overlapping region of two frame infrared images, to overlapping one error matrix ERR of regional structurek+1(i,j)2:
Wherein, Xk(i, j) indicates gray value of the coordinate for the pixel of (i, j), O in kth frame infrared imagek(i, j) is indicated with the
Coordinate is the value of (i, j) in k-1 and the calculated correction parameter matrix of kth frame, and wherein k-1 minimum is equal to 1;
Step 7 is updated correction parameter matrix:
Ok+1(i, j)=Ok(i,j)-2×α(i,j)×ERRk+1(i,j)
Step 8 exports image Y after obtaining the correction of+1 frame of kth plus correction parameter matrix with+1 frame infrared image of kthk(i, j):
Yk+1(i, j)=Xk+1(i,j)+Ok+1(i,j) 。
2. the asymmetric correction method based on scene interframe registration of adjusting learning rate according to claim 1,
It is characterized by: in above-mentioned steps 5, obtained according to the local variance of each pixel adaptive when the pixel updates correction parameter
Learning rate α (i, j), specific steps are as follows:
The local mean value m (i, j) of each pixel 5-1) is calculated, centered on locally referring to the coordinate (i, j) with any pixel, window is big
The small region for (2l+1) × (2l+1), wherein l is a preset integer, and X (a, b) indicates the gray value of pixel, is calculated public
Formula is as follows:
5-2) calculate the local variance sigma of each pixel2(i, j), calculation formula are as follows:
The learning rate of each pixel 5-3) is calculated according to local variance, calculation formula is as follows:
Wherein, αstFor initial learning rate, α (i, j) is the learning rate of the pixel of coordinate (i, j).
3. the asymmetric correction method based on scene interframe registration of adjusting learning rate according to claim 1,
It is characterized by: in chronological sequence sequence, two adjacent frames all to infrared image sequence, i.e. the 1st frame and the 2nd frame, the 2nd frame and
3rd frame ..., -1 frame of kth and kth frame ,+1 frame ... of kth frame and kth are carried out in claim 1 step 2 to the behaviour of step 8
Make, update is iterated to correction parameter matrix always.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910532213.1A CN110363714B (en) | 2019-06-19 | 2019-06-19 | Non-uniformity correction method based on scene interframe registration of self-adaptive learning rate |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910532213.1A CN110363714B (en) | 2019-06-19 | 2019-06-19 | Non-uniformity correction method based on scene interframe registration of self-adaptive learning rate |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110363714A true CN110363714A (en) | 2019-10-22 |
CN110363714B CN110363714B (en) | 2023-03-28 |
Family
ID=68216392
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910532213.1A Active CN110363714B (en) | 2019-06-19 | 2019-06-19 | Non-uniformity correction method based on scene interframe registration of self-adaptive learning rate |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110363714B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111507915A (en) * | 2020-04-13 | 2020-08-07 | 烟台艾睿光电科技有限公司 | Real-time infrared non-uniformity correction method, equipment and medium based on fuzzy registration |
CN113375808A (en) * | 2021-05-21 | 2021-09-10 | 武汉博宇光电***有限责任公司 | Infrared image non-uniformity correction method based on scene |
US20220319018A1 (en) * | 2021-03-30 | 2022-10-06 | Distech Controls Inc. | Method and computing device using a neural network to localize an overlap between two thermal images respectively generated by two infrared sensors |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102538973A (en) * | 2011-12-31 | 2012-07-04 | 南京理工大学 | Rapidly converged scene-based non-uniformity correction method |
CN104296876A (en) * | 2013-07-15 | 2015-01-21 | 南京理工大学 | FPGA-based scene non-uniformity correction method and device |
CN105318971A (en) * | 2014-07-07 | 2016-02-10 | 南京理工大学 | Self-adaptive non-uniformity correction method for carrying image registration on infrared video sequence |
US20170019616A1 (en) * | 2014-05-15 | 2017-01-19 | Huawei Technologies Co., Ltd. | Multi-frame noise reduction method, and terminal |
CN106570829A (en) * | 2015-10-09 | 2017-04-19 | 南京理工大学 | Infrared image non-uniformity correction method |
US20180204322A1 (en) * | 2017-01-13 | 2018-07-19 | Toshiba Medical Systems Corporation | Efficient variance-reduced method and apparatus for model-based iterative ct image reconstruction |
CN108665425A (en) * | 2018-03-28 | 2018-10-16 | 西安电子科技大学 | Infrared Image Non-uniformity Correction method based on interframe registration and adaptive step |
CN109741267A (en) * | 2018-12-05 | 2019-05-10 | 西安电子科技大学 | Infrared Image Non-uniformity Correction method based on three sides filtering and neural network |
-
2019
- 2019-06-19 CN CN201910532213.1A patent/CN110363714B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102538973A (en) * | 2011-12-31 | 2012-07-04 | 南京理工大学 | Rapidly converged scene-based non-uniformity correction method |
CN104296876A (en) * | 2013-07-15 | 2015-01-21 | 南京理工大学 | FPGA-based scene non-uniformity correction method and device |
US20170019616A1 (en) * | 2014-05-15 | 2017-01-19 | Huawei Technologies Co., Ltd. | Multi-frame noise reduction method, and terminal |
CN105318971A (en) * | 2014-07-07 | 2016-02-10 | 南京理工大学 | Self-adaptive non-uniformity correction method for carrying image registration on infrared video sequence |
CN106570829A (en) * | 2015-10-09 | 2017-04-19 | 南京理工大学 | Infrared image non-uniformity correction method |
US20180204322A1 (en) * | 2017-01-13 | 2018-07-19 | Toshiba Medical Systems Corporation | Efficient variance-reduced method and apparatus for model-based iterative ct image reconstruction |
CN108665425A (en) * | 2018-03-28 | 2018-10-16 | 西安电子科技大学 | Infrared Image Non-uniformity Correction method based on interframe registration and adaptive step |
CN109741267A (en) * | 2018-12-05 | 2019-05-10 | 西安电子科技大学 | Infrared Image Non-uniformity Correction method based on three sides filtering and neural network |
Non-Patent Citations (1)
Title |
---|
CHAO ZUO ET AL.: "Scene-based nonuniformity correction algorithm based on interframe registration", 《J.OPT.SOC.AM.A》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111507915A (en) * | 2020-04-13 | 2020-08-07 | 烟台艾睿光电科技有限公司 | Real-time infrared non-uniformity correction method, equipment and medium based on fuzzy registration |
CN111507915B (en) * | 2020-04-13 | 2024-03-29 | 烟台艾睿光电科技有限公司 | Real-time infrared non-uniformity correction method, device and medium based on fuzzy registration |
US20220319018A1 (en) * | 2021-03-30 | 2022-10-06 | Distech Controls Inc. | Method and computing device using a neural network to localize an overlap between two thermal images respectively generated by two infrared sensors |
US12020445B2 (en) * | 2021-03-30 | 2024-06-25 | Distech Controls Inc. | Method and computing device using a neural network to localize an overlap between two thermal images respectively generated by two infrared sensors |
CN113375808A (en) * | 2021-05-21 | 2021-09-10 | 武汉博宇光电***有限责任公司 | Infrared image non-uniformity correction method based on scene |
Also Published As
Publication number | Publication date |
---|---|
CN110363714B (en) | 2023-03-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105318971B (en) | The adaptive nonuniformity correction method of image registration is used to infrared video sequence | |
CN107255521B (en) | A kind of Infrared Image Non-uniformity Correction method and system | |
CN110363714A (en) | The asymmetric correction method based on scene interframe registration of adjusting learning rate | |
CN108846805B (en) | Infrared thermal image two-point non-uniform correction method based on scene self-adaption | |
CN108592884B (en) | A kind of general linear array satellite core line image generating method | |
CN106017695B (en) | Adaptive infrared asymmetric correction method based on state estimation | |
CN108981611B (en) | Digital projection grating image fitting correction method based on distortion global correction | |
CN102778296B (en) | Total variation-based self-adaptation non-uniformity correction method for infrared focal plane | |
US10931901B2 (en) | Method and apparatus for selectively correcting fixed pattern noise based on pixel difference values of infrared images | |
CN104296876A (en) | FPGA-based scene non-uniformity correction method and device | |
CN108665425A (en) | Infrared Image Non-uniformity Correction method based on interframe registration and adaptive step | |
CN109990731A (en) | The bearing calibration of nonlinear phase error based on digital projection three-dimension measuring system | |
CN109903235A (en) | A kind of removing method of infrared image fringes noise | |
CN109813442B (en) | Multi-frame processing-based internal stray radiation non-uniformity correction method | |
CN109801343A (en) | Based on annular artifact bearing calibration, the CT control system for rebuilding front and back image | |
CN106886983A (en) | Image non-uniform correction method based on Laplace operators and deconvolution | |
CN105931203A (en) | Infrared image stripe filtering method based on statistical relative stripe removal method | |
CN107486629A (en) | A kind of vision positioning laser marking system bearing calibration | |
JP6830712B1 (en) | Random sampling Consistency-based effective area extraction method for fisheye images | |
CN103164846A (en) | Infrared fringe non-uniformity correction method based on steering kernel estimation | |
CN110428376A (en) | Geometric correction method on a kind of line array CCD satellite image star based on FPGA | |
CN110501025A (en) | A kind of In-flight calibration method of Rolling shutter face battle array CMOS camera | |
CN111507915B (en) | Real-time infrared non-uniformity correction method, device and medium based on fuzzy registration | |
CN110033414A (en) | A kind of Infrared Image Non-uniformity Correction method and system based on equalization processing | |
Ratliff et al. | Algebraic scene-based nonuniformity correction in focal plane arrays |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |