CN102663714A - Saliency-based method for suppressing strong fixed-pattern noise in infrared image - Google Patents
Saliency-based method for suppressing strong fixed-pattern noise in infrared image Download PDFInfo
- Publication number
- CN102663714A CN102663714A CN2012100853128A CN201210085312A CN102663714A CN 102663714 A CN102663714 A CN 102663714A CN 2012100853128 A CN2012100853128 A CN 2012100853128A CN 201210085312 A CN201210085312 A CN 201210085312A CN 102663714 A CN102663714 A CN 102663714A
- Authority
- CN
- China
- Prior art keywords
- image
- marking area
- algorithm
- saliency map
- pattern noise
- 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
Abstract
The invention relates to a saliency-based method for suppressing a strong fixed-pattern noise in an infrared image. The method comprises the following steps: (1), inputting an infrared image having a strong fixed-pattern noise; (2), calculating a saliency map of the image, carrying out salient region detection and dividing the image into a salient region and a non-salient region; (3), employing a dual-platform histogram equalization algorithm to carry out processing on the salient region and for the non-salient region, employing a non-linear grey level mapping algorithm weighted by saliency to suppress a fixed-pattern noise; (4), eliminating a fake edge between local regions and obtaining values of pixel points on a boundary by weighted summation of values of pixel points that are in the salient region and the non-salient region and are near the boundary, thereby enabling the boundary to be in a smooth transition state and thus eliminating the fake edge; and (5), outputting a processing result. According to the invention, the method has advantages of simple operation, low usage cost, good versatility, high precision and substantial improvement on the image quality and the like.
Description
Technical field
The present invention is mainly concerned with fields such as Digital Image Processing, Video processing, computer vision, and the vision noticing mechanism that refers in particular to a kind of human visual system of utilization improves the method for infrared image or video quality and visual effect in real time.
Background technology
Owing to reasons such as manufacture crafts, different probe units will show different output responses for same infrared incident radiation amount, produces unevenness in the space, and then causes fixed pattern noise; Especially when the little target of observation, in order to improve the ability of discovery to little target, can improve the response gain of probe unit usually, thereby cause more tangible fixed pattern noise, this will badly influence the quality of infrared video.Provide the instance of the strong fixed pattern noise infrared images of two width of cloth among Fig. 1 a and Fig. 1 b, can find out that therefrom large-area strong fixed pattern noise has had a strong impact on the observation to echo signal in the image in the image, made visual quality of images descend.
In the prior art, the algorithm that suppresses for fixed pattern noise mainly is divided into two types: technological and technological based on the inhibition of scene based on the inhibition of reference source.
Inhibition Technology Need based on reference source is demarcated detector array elements in advance, obtains the Nonuniformity Correction factor, and the utilization correction factor is accomplished the inhibition to fixed pattern noise in actual application.Because fixed pattern noise slowly changes with the variation of environment in time, so these class methods need termly equipment is stopped, and probe unit is demarcated again, so it needing to be not suitable for the occasion of long time continuous working; Or the high precision reference source need be set in object scene, and system's heat picture is carried out real-time calibration, its cost is very expensive.
Inhibition technology based on scene is the information that is used for nonuniformity correction of directly from actual scene, extracting; With correction parameter or its correcting result of direct estimation of statistical method estimation infrared focal plane array, true picture is recovered to come out from the natural mode noise through the relevant information between true picture.The scene method requires scene to change (like object of which movement), and algorithm complex is higher, and correction accuracy is not high, is prone to produce " ghost ".Especially for the infrared image of " little target, overall background " because target is comparatively faint and in picture field the position relative fixed, very easily be treated to the ground unrest composition, cause effective target information loss, can not suppress fixed pattern noise well.
" human visual system " has the selective attention characteristic.The regional area details of utilizing the outstanding targetedly human eye of human-eye visual characteristic to pay close attention to suppresses non-region-of-interest noise, can effectively promote the whole visual effect of digital video.Practitioner Gianluigi Ciocca etc. has proposed a kind of image enchancing method that is driven by picture material.This algorithm is at first classified to input picture, determines only algorithm for image enhancement according to picture material then.Algorithm flow chart is as shown in Figure 2, and its main process is:
(1) method of using R.Schettini etc. to propose is outdoor, indoor and the close-up shot image according to low-level features such as color, texture, edges with image classification.Wherein, the structure of image classification device approximately needs 4500 manual images that mark correct classifications as training sample set.
(2) to the method for outdoor images with propositions such as C.Cusano, the utilization support vector machine method is classified to image-region, and then detects the zone with semantic meaning.The indoor detection with skin detection and people's face with the close-up shot image decided the zone with semantic meaning.
(3) detect marking area with Itti and Koch model.
(4) use the method for propositions such as A.Capra to accomplish the local contrast enhancing.
(5) calculate high fdrequency component with Laplace operator.Based on the significance zone that the 3rd step obtained, calculate remarkable high frequency, in original image, deduct non-remarkable high frequency, add remarkable high frequency, realize strengthening selectively the edge.
Can find out the problem that Gianluigi Ciocca method exists from above process: the image that 1) needs the correct classification of manual mark is as training sample set.2) this method of Gianluigi Ciocca method the 2nd step explanation is not considered the infrared image processing problem, is not suitable for infrared non-linear and proofreaies and correct.3) algorithm can only strengthen the edge selectively, can not effectively suppress strong fixed pattern noise.
Other has the practitioner to propose a kind of algorithm for image enhancement based on regional area significance weighting statistics gray level histogram.This algorithm utilizes Itti visual attention computation model acquisition image overall significantly to scheme; Then, the overall situation is significantly schemed to be divided into big subregion such as some grades, ask for the average significantly value of all subregion, and do normalization and handle, obtain the weighting statistics coefficient of subregion; With the gray level weighting statistical value addition of all subregion, obtain gray-scale information amount histogram again; At last, according to the mapping function of histogram equalization, the dynamic range of adjustment gray level.But when the marking area object contained the gray-scale value identical with non-marking area background, said method is the information of Prwsnt targets of interest really.
Summary of the invention
The technical matters that the present invention will solve just is: to the technical matters that prior art exists, the present invention provide a kind of easy and simple to handle, use cost is low, versatility good, precision is high, can improve the strong cover half formula of the infrared image based on the conspicuousness noise suppressing method of picture quality greatly.
For solving the problems of the technologies described above, the present invention adopts following technical scheme:
The strong cover half formula of a kind of infrared image based on conspicuousness noise suppressing method the steps include:
(1) input has the infrared image of strong fixed pattern noise;
(2) image of importing to step (1), the saliency map of computed image, and carry out marking area in view of the above and detect, be marking area and non-marking area with image division;
(3) take different Processing Algorithm to resulting zones of different in the step (2):, adopt two platform algorithm of histogram equalization to promote the marking area contrast, to strengthen target effective information for the resulting marking area of step (2); For the resulting non-marking area of step (2), adopt the nonlinear gray level mapping algorithm of significance weighting to suppress fixed pattern noise;
(4) the false edge between the elimination regional area; The value of pixel on the border is asked in the weighted sum that is positioned at the pixel point value of boundary vicinity through marking area and non-marking area, makes boundary seamlessly transit, and eliminates false edge;
(5) output result.
As further improvement of the present invention:
In the said step (2); When the significance of computed image; Adopt multiple dimensioned frequency-domain analysis algorithm respectively and come the overall saliency map of calculating input image based on the local contrast analytical algorithm, and the final saliency map S that obtains through weighting, final saliency map S is expressed as:
S＝w
1S
sss+w
2S
RC (1)
Wherein, S
SssBe the saliency map that obtains through multiple dimensioned frequency-domain analysis algorithm, S
RCBe the saliency map through obtaining, w based on the local contrast analytical algorithm
1Be S
SssShared weight in final saliency map S, w
2Be S
RCShared weight in final saliency map S.
In the said step (3), for non-marking area, use the nonlinear gray level mapping algorithm of significance weighting to suppress strong fixed pattern noise, the expression formula of the nonlinear gray level mapping algorithm of said significance weighting is:
Wherein i is the gray-scale value on the corresponding pixel points, s '
SSSBe that L is a gradation of image progression through the significance value of this some correspondence after the normalization of multiple dimensioned frequency-domain analysis algorithm computation gained saliency map process, i ' is grey scale pixel value after shining upon.
Compared with prior art, the invention has the advantages that:
(1) the present invention is based on the strong cover half formula of the infrared image noise suppressing method of conspicuousness; Only solve the strong cover half formula of infrared image squelch problem based on single-frame images significance information; Than tradition based on radiation source or based on the inhibition of scene technology; The present invention need not carry out complicated non-homogeneous parameter estimation and correction, and algorithm complex is low, is easy to use.
(2) the present invention is based on the strong cover half formula of the infrared image noise suppressing method of conspicuousness,, extract the corresponding marking area of target from human visual system's selective attention characteristic; The observation intention that reflects most of users exactly; Than other content-based image processing methods, the present invention need not by the picture material classification, and extensive degree is high; Versatility is good; And do not need precondition based on the picture material semantic classifiers, the user does not need to be grasped the present technique knowledge in field, and automaticity is high.
(3) the present invention is based on the strong cover half formula of the infrared image noise suppressing method of conspicuousness; Through weighting fusion with two kinds of marking area detection method SSS and RC; Obtain the division of final remarkable, non-marking area, efficiently solve the problem of target area flase drop of target area omission or the RC method of SSS method.
(4) the present invention is based on the strong cover half formula of the infrared image noise suppressing method of conspicuousness, adopt mode cheaply, improved infrared image or video quality, especially the image of " little target, overall background ".This method need periodically not demarcated focal plane arrays (FPA) by means of black matrix; Need not purchase senior infrared imaging system yet; Can improve, improve " little target, overall background " infrared image or video quality; Obtain high-quality infrared image or video data, also can play interchannel noise simultaneously and suppress effect well.
(5) the present invention compares with general nonlinear gray level mapping algorithm, when can keep target effective information, suppresses close with target area brightness, the large stretch of fixed pattern noise of brightness effectively.
Description of drawings
Fig. 1 a and Fig. 1 b are the synoptic diagram of strong fixed pattern noise infrared image example in the prior art.
Fig. 2 is the schematic flow sheet of Gianluigi Ciocca method in the prior art.
Fig. 3 is a processing flow chart of the present invention.
Fig. 4 a～Fig. 4 i is the synoptic diagram of the present invention's processing procedure in a concrete application example, and wherein Fig. 4 a is an original image, and Fig. 4 b is the image of RC method after handling; Fig. 4 c is the image of SSS method after handling; Fig. 4 d is the synoptic diagram that remarkable and non-marking area is divided; Fig. 4 e is cut apart the echo signal that original image obtains; Fig. 4 f is the synoptic diagram after echo signal strengthens; Fig. 4 g is for separating the noise that original image obtains; Fig. 4 h is the image after the squelch; Fig. 4 i exports result's synoptic diagram for the present invention.
Fig. 5 a and Fig. 5 b are the result contrast synoptic diagram of the present invention in instance two, and wherein Fig. 5 a is the original image of instance two, and Fig. 5 b is the result images of instance two after the present invention handles.
Fig. 6 a and Fig. 6 b are the result contrast synoptic diagram of the present invention in instance three, and wherein Fig. 6 a is the original image of instance three, and Fig. 6 b is the result images of instance three after the present invention handles.
Fig. 7 a and Fig. 7 b are the result contrast synoptic diagram of the present invention in instance four, and wherein Fig. 7 a is the original image of instance four, and Fig. 7 b is the result images of instance four after the present invention handles.
Embodiment
Below will combine Figure of description and specific embodiment that the present invention is explained further details.
As shown in Figure 3, the present invention is based on the strong cover half formula of the infrared image noise suppressing method of conspicuousness, the steps include:
(1) input has the infrared image of strong fixed pattern noise; For example, the strong fixed pattern noise infrared image of input " little target, overall background ".
(2) image of importing to step (1), the saliency map of computed image, and carry out marking area in view of the above and detect, be marking area and non-marking area with image division;
(3) take different Processing Algorithm to resulting different zone in the step (2), thereby suppress strong fixed pattern noise, outstanding target effective information.That is: for the resulting marking area of step (2), adopt two platform algorithm of histogram equalization, be used for promoting the marking area contrast, and then strengthen target effective information; For the resulting non-marking area of step (2), promptly general area under assurance and the conforming prerequisite of marking area edge gradient, adopts the nonlinear gray level mapping algorithm of significance weighting to suppress fixed pattern noise.
(4) the false edge between the elimination regional area; The value of pixel on the border is asked in the weighted sum that is positioned at the pixel point value of boundary vicinity through marking area and non-marking area, thereby accomplishes that boundary seamlessly transits, and eliminates false edge.
(5) output result.
In the above-mentioned steps (2), when the significance of computed image, can adopt multiple dimensioned frequency-domain analysis algorithm (Spectrum Scale Space analysis, SSS respectively; Be called for short " SSS method ") and based on local contrast analytical algorithm (Region Contrast, RC; Be called for short " RC method ") come the overall saliency map of calculating input image, and the final saliency map S that obtains through weighting, final saliency map S can be expressed as:
S＝w
1S
sss+w
2S
RC (1)
Wherein, S
SssBe the saliency map that obtains through the SSS method, S
RCBe the saliency map that obtains through the RC method, w
1Be S
SssShared weight in final saliency map S, w
2Be S
RCShared weight in final saliency map S.
Marking area through the RC method obtains is bigger than normal, and the SSS method can quite good detecting go out target and target area, but when 2 or a plurality of comparatively tangible object are arranged in the picture, the omission phenomenon can occur.Therefore, flase drop and the omission that powers and functions are effectively avoided marking area that add through two kinds of marking area detection methods.
Then, the saliency map that obtains based on formula (1) is marking area and non-marking area with image division.
In the above-mentioned steps (3); Carrying out two platform histogram equalizations for marking area handles; Through two plateau value are set; Two platform algorithm of histogram equalization have effectively overcome the problems such as enhancing and valid gray level information dropout of crossing that traditional algorithm of histogram equalization exists, and can effectively promote contrast, strengthen the marking area detailed information.And, using the nonlinear gray level mapping algorithm of significance weighting to suppress strong fixed pattern noise for non-marking area, the gray-scale value of traditional nonlinear gray level mapping algorithm after with normalization shines upon gray level as weights; Fixed pattern noise in the little target infrared image is played certain inhibition effect; But because it has used gray-scale information, when more weak and fixed pattern noise was strong when echo signal in the image, it can not effectively suppress noise; Can weaken echo signal on the contrary; Here introduce significance information, shown in formula (2), effectively overcome the problems referred to above.Processing procedure is as shown in Figure 4.
Wherein i is the gray-scale value on the corresponding pixel points, s '
SSSCalculate the gained saliency map through the corresponding significance value of this point after the normalization for the SSS method, L is a gradation of image progression, and i ' is mapping back grey scale pixel value.
Referring to Fig. 4, have particular application as example with one.
The first step adopts the marking area of above-mentioned formula (1) calculating input image;
Fig. 4 a～Fig. 4 i is for calculating the process of marking area to the strong fixed pattern noise infrared image of a frame " little target, overall background ".Wherein, Fig. 4 a is an original image, and sign can find out that there is very serious fixed pattern noise in the zone beyond the target from figure.Fig. 4 b and Fig. 4 c are respectively and use RC method and SSS method to carry out the result that significance detects.Two kinds of testing results are carried out weighting and carry out binaryzation obtaining marking area, the division of non-marking area, see among Fig. 4 d to identify.And serve as to obtain corresponding marking area Fig. 4 e of target and the corresponding non-marking area Fig. 4 g of noise according to original image Fig. 4 a is cut apart with this binaryzation result.
In second step, the subregion is carried out digital picture and is strengthened;
Marking area is carried out two platform histogram equalizations processing obtain Fig. 4 f, can find out relatively that with Fig. 4 e echo signal intensity has obtained keeping and giving prominence to; The nonlinear gray level conversion that non-marking area is carried out the significance weighting suppresses strong fixed pattern noise and obtains Fig. 4 h, and the fixed pattern noise in Fig. 4 g background has obtained better inhibited.
In the 3rd step, eliminate the false edge between the regional area.
The method that Fig. 4 f and Fig. 4 h mode through the weighted array of boundary vicinity pixel is obtained pixel point value on the border combines and obtains result Fig. 4 i of the present invention.
The 4th step, the result after output is handled.
For verifying usefulness of the present invention; A large amount of experiments in various embodiment, have been carried out; For example having chosen a few width of cloth results again enumerates as shown in the figure; Be that Fig. 5 a and Fig. 5 b are the result contrast synoptic diagram of the present invention in instance two, wherein Fig. 5 a is the original image of instance two, and Fig. 5 b is the result images of instance two after the present invention handles; Fig. 6 a and Fig. 6 b are the result contrast synoptic diagram of the present invention in instance three, and wherein Fig. 6 a is the original image of instance three, and Fig. 6 b is the result images of instance three after the present invention handles; Fig. 7 a and Fig. 7 b are the result contrast synoptic diagram of the present invention in instance four, and wherein Fig. 7 a is the original image of instance four, and Fig. 7 b is the result images of instance four after the present invention handles.Relatively handle forward and backward data in each instance; To " little target, overall background " strong fixed pattern noise infrared image; The present invention is only from single image information; Accomplish strong cover half formula Noise Suppression preferably, effectively given prominence to the target effective detailed information simultaneously, verified validity of the present invention.
Below only be preferred implementation of the present invention, protection scope of the present invention also not only is confined to the foregoing description, and all technical schemes that belongs under the thinking of the present invention all belong to protection scope of the present invention.Should be pointed out that for those skilled in the art some improvement and retouching not breaking away under the principle of the invention prerequisite should be regarded as protection scope of the present invention.
Claims (3)
1. the strong cover half formula of the infrared image based on conspicuousness noise suppressing method is characterized in that step is:
(1) input has the infrared image of strong fixed pattern noise;
(2) image of importing to step (1), the saliency map of computed image, and carry out marking area in view of the above and detect, be marking area and non-marking area with image division;
(3) take different Processing Algorithm to resulting zones of different in the step (2):, adopt two platform algorithm of histogram equalization to promote the marking area contrast, to strengthen target effective information for the resulting marking area of step (2); For the resulting non-marking area of step (2), adopt the nonlinear gray level mapping algorithm of significance weighting to suppress fixed pattern noise;
(4) eliminate false edge between the regional area, the value of pixel on the border is asked in the weighted sum that is positioned at the pixel point value of boundary vicinity through marking area and non-marking area, makes boundary seamlessly transit, and eliminates false edge;
(5) output result.
2. the strong cover half formula of the infrared image based on conspicuousness according to claim 1 noise suppressing method; It is characterized in that; In the said step (2), when the significance of computed image, adopt multiple dimensioned frequency-domain analysis algorithm respectively and come the overall saliency map of calculating input image based on the local contrast analytical algorithm; And the final saliency map S that obtains through weighting, final saliency map S is expressed as:
S＝w
1S
sss+w
2S
RC (1)
Wherein, S
SssBe the saliency map that obtains through multiple dimensioned frequency-domain analysis algorithm, S
RCBe the saliency map through obtaining, w based on the local contrast analytical algorithm
1Be S
SssShared weight in final saliency map S, w
2Be S
RCShared weight in final saliency map S.
3. the strong cover half formula of the infrared image based on conspicuousness according to claim 1 and 2 noise suppressing method; It is characterized in that; In the said step (3); For non-marking area, use the nonlinear gray level mapping algorithm of significance weighting to suppress strong fixed pattern noise, the expression formula of the nonlinear gray level mapping algorithm of said significance weighting is:
Wherein i is the gray-scale value on the corresponding pixel points, s '
SSSBe the significance value of multiple dimensioned frequency-domain analysis algorithm computation gained saliency map through this some correspondence after the normalization, L is a gradation of image progression, and i ' is mapping back grey scale pixel value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210085312.8A CN102663714B (en) | 2012-03-28 | 2012-03-28 | Saliency-based method for suppressing strong fixed-pattern noise in infrared image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210085312.8A CN102663714B (en) | 2012-03-28 | 2012-03-28 | Saliency-based method for suppressing strong fixed-pattern noise in infrared image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102663714A true CN102663714A (en) | 2012-09-12 |
| CN102663714B CN102663714B (en) | 2014-06-25 |
Family
ID=46773193
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210085312.8A Active CN102663714B (en) | 2012-03-28 | 2012-03-28 | Saliency-based method for suppressing strong fixed-pattern noise in infrared image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102663714B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103413303A (en) * | 2013-07-29 | 2013-11-27 | 西北工业大学 | Infrared target segmentation method based on joint obviousness |
| CN103679697A (en) * | 2012-09-13 | 2014-03-26 | 欧姆龙株式会社 | Image processing device, and image processing method |
| CN104951440A (en) * | 2014-03-24 | 2015-09-30 | 联想(北京)有限公司 | Image processing method and electronic device |
| CN105890768A (en) * | 2016-03-31 | 2016-08-24 | 浙江大华技术股份有限公司 | Infrared image non-uniformity correction method and device |
| CN106716990A (en) * | 2014-09-30 | 2017-05-24 | 富士胶片株式会社 | Infrared imaging device, fixed pattern noise calculation method, and fixed pattern noise calculation program |
| WO2017185442A1 (en) * | 2016-04-26 | 2017-11-02 | 成都市晶林科技有限公司 | Infrared image stripe filtering method based on image laying |
| CN107705324A (en) * | 2017-10-20 | 2018-02-16 | 中山大学 | A kind of video object detection method based on machine learning |
| CN109523477A (en) * | 2018-11-06 | 2019-03-26 | 武汉高德智感科技有限公司 | A kind of adaptive infrared image dynamic range transform method |
| WO2019100270A1 (en) * | 2017-11-22 | 2019-05-31 | 深圳市大疆创新科技有限公司 | Image noise calibration method and apparatus, image noise reduction method and apparatus, and image processing apparatus |
| CN111161177A (en) * | 2019-12-25 | 2020-05-15 | Tcl华星光电技术有限公司 | Image self-adaptive noise reduction method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7242018B2 (en) * | 2002-03-22 | 2007-07-10 | Fujifilm Corporation | Method and apparatus for acquiring radiation image signals |
| CN102129675A (en) * | 2011-02-24 | 2011-07-20 | 中国兵器工业系统总体部 | Nonlinear adaptive infrared image enhancing method |
-
2012
- 2012-03-28 CN CN201210085312.8A patent/CN102663714B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7242018B2 (en) * | 2002-03-22 | 2007-07-10 | Fujifilm Corporation | Method and apparatus for acquiring radiation image signals |
| CN102129675A (en) * | 2011-02-24 | 2011-07-20 | 中国兵器工业系统总体部 | Nonlinear adaptive infrared image enhancing method |
Non-Patent Citations (2)
| Title |
|---|
| TANG HUANG等: "Fixed Pattern Noise Suppression Algorithm Based on Background Modeling", 《2012 INTERNATIONAL WORKSHOP ON INFORMATION AND ELECTRONICS ENGINEERING (IWIEE)》 * |
| 吴锦杰等: "基于双目相机的图像拼接", 《计算机工程》 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103679697A (en) * | 2012-09-13 | 2014-03-26 | 欧姆龙株式会社 | Image processing device, and image processing method |
| CN103413303B (en) * | 2013-07-29 | 2016-02-10 | 西北工业大学 | Based on the infrared object segmentation method of associating conspicuousness |
| CN103413303A (en) * | 2013-07-29 | 2013-11-27 | 西北工业大学 | Infrared target segmentation method based on joint obviousness |
| CN104951440A (en) * | 2014-03-24 | 2015-09-30 | 联想(北京)有限公司 | Image processing method and electronic device |
| CN104951440B (en) * | 2014-03-24 | 2020-09-25 | 联想(北京)有限公司 | Image processing method and electronic equipment |
| CN106716990B (en) * | 2014-09-30 | 2019-12-20 | 富士胶片株式会社 | Infrared imaging device and fixed pattern noise calculation method |
| CN106716990A (en) * | 2014-09-30 | 2017-05-24 | 富士胶片株式会社 | Infrared imaging device, fixed pattern noise calculation method, and fixed pattern noise calculation program |
| US10536626B2 (en) | 2014-09-30 | 2020-01-14 | Fujifilm Corporation | Infrared imaging device, fixed pattern noise calculation method, and fixed pattern noise calculation program |
| CN105890768A (en) * | 2016-03-31 | 2016-08-24 | 浙江大华技术股份有限公司 | Infrared image non-uniformity correction method and device |
| CN105890768B (en) * | 2016-03-31 | 2019-02-12 | 浙江大华技术股份有限公司 | A kind of method and device of Infrared Image Non-uniformity Correction |
| WO2017185442A1 (en) * | 2016-04-26 | 2017-11-02 | 成都市晶林科技有限公司 | Infrared image stripe filtering method based on image laying |
| CN107705324A (en) * | 2017-10-20 | 2018-02-16 | 中山大学 | A kind of video object detection method based on machine learning |
| WO2019100270A1 (en) * | 2017-11-22 | 2019-05-31 | 深圳市大疆创新科技有限公司 | Image noise calibration method and apparatus, image noise reduction method and apparatus, and image processing apparatus |
| CN109523477A (en) * | 2018-11-06 | 2019-03-26 | 武汉高德智感科技有限公司 | A kind of adaptive infrared image dynamic range transform method |
| CN109523477B (en) * | 2018-11-06 | 2021-07-30 | 武汉高德智感科技有限公司 | Self-adaptive infrared image dynamic range transformation method |
| CN111161177A (en) * | 2019-12-25 | 2020-05-15 | Tcl华星光电技术有限公司 | Image self-adaptive noise reduction method and device |
| CN111161177B (en) * | 2019-12-25 | 2023-09-26 | Tcl华星光电技术有限公司 | Image self-adaptive noise reduction method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102663714B (en) | 2014-06-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102663714B (en) | Saliency-based method for suppressing strong fixed-pattern noise in infrared image | |
| Wang et al. | Haze removal based on multiple scattering model with superpixel algorithm | |
| Huang et al. | An advanced single-image visibility restoration algorithm for real-world hazy scenes | |
| JP5458905B2 (en) | Apparatus and method for detecting shadow in image | |
| Shiau et al. | Weighted haze removal method with halo prevention | |
| Luan et al. | Fast single image dehazing based on a regression model | |
| CN102138157B (en) | Color constancy method and system | |
| CN101959018A (en) | Image processing equipment, image processing method and program | |
| CN102222323A (en) | Histogram statistic extension and gradient filtering-based method for enhancing infrared image details | |
| CN101626454B (en) | Method for intensifying video visibility | |
| El Khoury et al. | Color and sharpness assessment of single image dehazing | |
| CN105136308A (en) | Adaptive correction method under variable integral time of infrared focal plane array | |
| CN104200457A (en) | Wide-angle camera shooting based discrete type canopy leaf area index detection system and method | |
| CN105139364A (en) | Image enhancement method and application thereof | |
| CN106934770A (en) | A kind of method and apparatus for evaluating haze image defog effect | |
| CN104574312A (en) | Method and device of calculating center of circle for target image | |
| CN103295205A (en) | Low-light-level image quick enhancement method and device based on Retinex | |
| CN106600615A (en) | Image edge detection algorithm evaluation system and method | |
| CN110111347A (en) | Logos extracting method, device and storage medium | |
| Purohit et al. | Enhancing the surveillance detection range of image sensors using HDR techniques | |
| CN113870150B (en) | Method for inverting spacecraft low-frequency vibration parameters based on continuous multiple remote sensing images | |
| CN103985089A (en) | Image stripe correction method in which weight edge analysis and intra-frame iteration are combined | |
| Lee et al. | Joint defogging and demosaicking | |
| CN107358616A (en) | SAR image edge detection method based on anisotropic morphology direction ratio | |
| Saberi et al. | Adaptive contrast enhancement of satellite images based on histogram and non-linear transfer function methods |
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 |