CN104333716B - Radiation-resistant video image system and method - Google Patents
Radiation-resistant video image system and method Download PDFInfo
- Publication number
- CN104333716B CN104333716B CN201410618197.5A CN201410618197A CN104333716B CN 104333716 B CN104333716 B CN 104333716B CN 201410618197 A CN201410618197 A CN 201410618197A CN 104333716 B CN104333716 B CN 104333716B
- Authority
- CN
- China
- Prior art keywords
- noise
- image
- video
- video image
- pixel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a kind of radiation-resistant video image system and method；The system is used to realize this method；The system includes video image acquisition device (1) and video image processing device (3)；Video image processing device (3) includes digital processing unit (31) and video image N6ise deletion device (32)；Video image N6ise deletion device (32) is placed in digital processing unit (31), video image N6ise deletion device (32) includes median method denoising module, homodyne video noise image zooming-out module, mutual deviation method extracts video noise image module forward, mutual deviation method extracts video noise image module backward, secondary video noise image module is synthesized forward, secondary video noise image module is synthesized backward, preceding poor noise figure picture defines module, poor noise figure picture defines module afterwards, noise image demarcating module, noise eliminating module, median method video image correcting module.The video image noise handled by it is few, image clearly.
Description
Technical field
The present invention relates to a kind of radiation environment hypograph sensing technology, particularly a kind of radioresistance image sensing method and it is
System.
Background technology
With the various operation tasks continually developed under utilization, radiation environment increasingly deepened with nuclear power source of space probation
Also it is more and more.But the harm of Radiation On Human body is very big, therefore the engineer operation under strong radiation environment is needed by controlling at a distance
System or automatic job are completed.Various imaging sensors be widely used in the fields such as space science and nuclear science as core
Detector.Therefore the radiation hardening of imaging sensor has important actual application value.
Existing vision sensor（That is imaging sensor, similarly hereinafter）Radioresistance technology mainly uses pixel electronic circuit internal junction
Structure is reinforced and two methods of radiation isolation, i.e., improved using different hardware radiation proof reinforcings or the anti-Irradiation Design technique of chip
The capability of resistance to radiation of sensor.Both imaging sensor radioresistance technologies, often add the volume and weight of sensor, and
Complicated design technology can sharply increase imaging sensor cost, limit its popularization and application.Imaging sensor is by space spoke
Penetrate the influence with nuclear radiation and occur damage effect, so that the image scene directly obtained is often with stronger pollution noise.
When high energy particle is beaten on CCD image sensor array under intense radiation, the pixel hit, its pixel value can increase suddenly,
Make to produce local pulse noise in the image that it is obtained.The energy of other high energy particle can be transmitted between adjacent picture elements, so that
Noise regional area is caused to spread.For these problems, some researchers propose to eliminate strong spoke using the method for image procossing
Picture noise under penetrating, reaches the purpose that the imaging sensor based on software is reinforced, for example, king's Heng etc. is in document《A kind of strong spoke
Penetrate the new method of environmental monitoring hypograph noise reduction》（I.e.《Sensor and micro-system》, the phase of volume 30 11,59-61 pages in 2011）In carry
Go out a kind of to come improved adaptive median filter under intense radiation with the compound noise-removed filtering method that wavelet transformation is combined
Image carries out recovery processing.But this simple big to noise intensity, the densely distributed image effect of filtering is not good.
The content of the invention
The purpose of the present invention is to overcome the above-mentioned not enough of prior art and provide a kind of radiation-resistant video image system, its
The first-class redundance unit of bimirror is not needed to make a return journey noise, it is not required that ruggedized equipment reduces noise, and simple in construction, cost is low.
It is a further object of the present invention to provide a kind of radioresistance method of video image system, said system has used the party
Method, this method high treating effect, speed is fast, and step is simple, and the equipment that this method need not be complicated, use cost is low.
The technical scheme is that：A kind of radiation-resistant video image system；It include video image acquisition device and
Video image processing device；Video image acquisition device includes amasthenic lens and video image sensors, video image sensors
In the light path that amasthenic lens is focused on, video image acquisition device can obtain the sequence of video images of detection target, each regard
Frequency image is the image of the detection target obtained at certain a moment；Pass through between video image acquisition device and video image processing device
Communication system is connected；Video image processing device includes digital processing unit and video image N6ise deletion device；Digital processing is filled
Put including input interface, memory, processor and output interface；Video image N6ise deletion device is placed in digital processing unit,
Video image N6ise deletion device include median method denoising module, homodyne video noise image zooming-out module, forward mutual deviation method carry
Take video noise image module, mutual deviation method extracts video noise image module, synthesis secondary video noise figure is as mould forward backward
Block, backward synthesis secondary video noise image module, preceding poor noise figure picture define module, rear poor noise figure picture and define module, make an uproar
Acoustic image demarcating module, noise eliminating module, median method video image correcting module；
Median method denoising module, its each frame video image for being used to gather to come to video image sensors (12) is carried out just
Denoising is walked, each pixel that will each in video image is with itself and periphery r row's pixel compositions（2r+1）x（2r+1）It is individual
The median pixel value of pixel is substituted, and r=1~5 obtain each intermediate value video image；
Homodyne video noise image zooming-out module, for each pixel value of each width video image to be subtracted in
Corresponding pixel value in the intermediate value video image obtained after value method denoising resume module, and take absolute value, obtain homodyne noise
Image；
Mutual deviation method extracts video noise image module forward, tight before it for will be come in every width video image and video
Adjacent r1 width video images, which are compared, obtains each poor picture, r1=1~3, then by each difference as be added obtain each video image to
Preceding noise figure picture；
Mutual deviation method extracts video noise image module backward, for by will be come thereafter in every width video image and video
Face close to r1 width video images be compared and obtain each poor picture, r1=1~3, then each difference as being added is obtained into each video image
Noise figure picture backward；
Secondary video noise image module is synthesized forward, for by the homodyne noise figure corresponding to each width video image
As with noise figure picture forward in proportionSecondary video noise figure picture forward is synthesized, i.e., by the every of each width homodyne noise figure picture
One noise pixel value withBe multiplied, then by each noise pixel value of noise figure picture forward withIt is multiplied, takes this
Two products and each noise pixel value as secondary video noise figure picture forward；
Secondary video noise image module is synthesized backward, for by the corresponding homodyne noise figure picture of each width video image
With noise figure picture backward in proportionSecondary video noise figure picture backward is synthesized, i.e., by each of each width homodyne noise figure picture
Position noise pixel value withBe multiplied, then by each noise pixel value of noise figure picture backward withBe multiplied, take this two
Individual product and each noise pixel value as secondary video noise figure picture backward；
Wherein=0.1~0.5；
Preceding poor noise figure picture defines module, for by the value of each noise pixel in secondary video noise figure picture forward
Threshold value is defined with oneIt is compared, threshold value is defined if being more than, then assert that pixel position noise is present, value is
1, otherwise assert that pixel position noise is not present, its value is 0, and poor noise defines image before obtaining；
Poor noise figure picture defines module afterwards, and it is used for each noise pixel in secondary video noise figure picture backward
Value defines threshold value with oneIt is compared, threshold value is defined if being more than, then assert that pixel position noise is present, value
For 1, otherwise assert that pixel position noise is not present, its value is 0, obtains rear poor noise and defines image；
Noise defines threshold value=20~80；
Noise image demarcating module, the corresponding noise picture of image is defined for judging that preceding poor noise defines image and rear poor noise
Vegetarian refreshments be noise when, then demarcate the point for noise pixel, represent that otherwise noise does not regard as noise, with " 0 " with " 1 "
Represent, obtain Noise Calibration image；
Noise eliminating module, for video image to be extended to the periphery into a pixel, obtains extending video image, then will prolong
The pixel for the position for being demarcated as noise in video image in correspondence Noise Calibration image is stretched, corresponding picture in extension video image is taken
Four, vegetarian refreshments periphery summit pixel value sum synthesizes new pixel value according to parameter a, b with four, its periphery pixel value sum, obtains
To N6ise deletion video image；Namely it is exactly that will be demarcated as making an uproar in correspondence Noise Calibration image to extend in video image in fact
Centered on the pixel of the position of sound 3x3 pixel values composition array withCorrespondence position be multiplied take again and, as
The pixel value of corresponding pixel in N6ise deletion video image, obtains N6ise deletion video image；Wherein a+b=0.25.
Further technical scheme is the present invention：Described noise eliminating module is additionally operable to re-use 1~20 time, often
Carry out next time in use, being that the N6ise deletion video image for obtaining preceding first use is used as regarding using beginning next time
Frequency image is used, and obtains next N6ise deletion video image.
Further technical scheme of the invention is：It also includes median method video image correcting module, median method video
Image modification module is identical with the structure of above-mentioned median method denoising module, i.e., handle image using identical median method, its
For carrying out a median method denoising to N6ise deletion video image, final denoising video image is obtained.
Further technical scheme is the present invention：Described r=3；r1=2.
The also further technical scheme of the present invention is：Described=0.25；=40。
Further technical scheme is the present invention：Described a=0.073235, b=0.176765.
Further technical scheme is the present invention：Described communication system is cable and its interface circuit, or is channel radio
News system, wireless telecommunication system includes the parts such as wireless transceiver, interface circuit；Video image sensors, communication system are also set
There is hardware reinforcement technique；Digital processing unit, which is one, can carry out the computer of Digital Signal Processing.
The present invention another technical scheme be：A kind of radioresistance method of video image system；It comprises the following steps：
First, video image is obtained by the video image acquisition device of video image system；
2nd, image noise is tentatively gone using median method：Each frame video image that video image acquisition device is obtained
Each pixel arranges what pixel was constituted with it with periphery r（2r+1）x（2r+1）The median pixel value of individual pixel is substituted, and r=1~5 are obtained
To each intermediate value video image；
3rd, video noise image is extracted using homodyne：Each pixel value of each width video image is subtracted in
Corresponding pixel value in the intermediate value video image that value method is obtained, and take absolute value, obtain homodyne noise figure picture；
4th, video noise image is extracted using mutual deviation method forward：It is tight before it by being come in every width video image and video
Adjacent r1 width video images, which are compared, obtains each poor picture, r1=1~3, then by each difference as be added obtain each video image to
Preceding noise figure picture；
5th, video noise image is extracted using mutual deviation method backward, by will be come thereafter in every width video image and video
Face close to r1 width video images be compared and obtain each poor picture, r1=1~3, then each difference as being added is obtained into each video image
Noise figure picture backward；
6th, using secondary video noise figure picture is synthesized forward, by the homodyne noise figure corresponding to each width video image
As with noise figure picture forward in proportionSecondary video noise figure picture forward is synthesized, i.e., by the every of each width homodyne noise figure picture
One noise pixel value withBe multiplied, then by each noise pixel value of noise figure picture forward withIt is multiplied, takes this
Two products and each noise pixel value as secondary video noise figure picture forward；
Using synthesis secondary video noise figure picture backward：By the corresponding homodyne noise figure picture of each width video image with to
Noise figure picture is in proportion afterwardsSecondary video noise figure picture, i.e. each by each width homodyne noise figure picture are made an uproar backward for synthesis
Sound pixel value withBe multiplied, then by each noise pixel value of noise figure picture backward withIt is multiplied, takes this two to multiply
Each noise pixel value long-pending and as secondary video noise figure picture backward；
Wherein=0.1~0.5；
7th, define that preceding poor noise defines image and rear poor noise defines image：By secondary video noise figure forward as in
The value of each noise pixel defines threshold value with oneIt is compared, threshold value is defined if being more than, then the pixel is assert
Position noise is present, and value is 1, otherwise assert that pixel position noise is not present, its value is 0, and poor noise defines figure before obtaining
Picture；
The value of each noise pixel in secondary video noise figure picture backward is defined into threshold value with oneCompared
Compared with if more than defining threshold value, then assert that pixel position noise is present, value is 1, otherwise assert pixel position noise
It is not present, its value is 0, obtains rear poor noise and defines image；
Noise defines threshold value=20~80；
8th, noise image is demarcated：Judge that preceding difference noise defines image and rear poor noise and defines the corresponding noise pixel of image
Be noise when, then demarcate the point for noise pixel, represent that otherwise noise does not regard as noise with " 1 ", represented with " 0 ",
Obtain Noise Calibration image；
9th, cancelling noise：For video image to be extended to the periphery into a pixel, obtain extending video image, then will prolong
The pixel for the position for being demarcated as noise in video image in correspondence Noise Calibration image is stretched, corresponding picture in extension video image is taken
Four, vegetarian refreshments periphery summit pixel value sum synthesizes new pixel value according to parameter a, b with four, its periphery pixel value sum, obtains
To N6ise deletion video image；Namely it is exactly that will be demarcated as making an uproar in correspondence Noise Calibration image to extend in video image in fact
Centered on the pixel of the position of sound 3x3 pixel values composition array withCorrespondence position be multiplied take again and, as
The pixel value of corresponding pixel in N6ise deletion video image, obtains N6ise deletion video image；Wherein a+b=0.25.
The present invention has following features compared with prior art：
1. the system of the present invention does not need the first-class redundance unit of bimirror to make a return journey noise, it is not required that ruggedized equipment is made an uproar to reduce
Sound, simple in construction, cost is low；
2. the system and method high treating effect of the present invention, speed is fast, step is simple, and this method can in a computer easily
Realize, it is not necessary to which complicated equipment, use cost is low.
The detailed construction of the present invention is further described below in conjunction with the drawings and specific embodiments.
Brief description of the drawings
Fig. 1 is the general construction block diagram of radiation-resistant video image system of the present invention；
Fig. 2 is the structured flowchart of the video image acquisition device of radiation-resistant video image system of the present invention；
Fig. 3 is the structured flowchart of the video image processing device of radiation-resistant video image system of the present invention；
Fig. 4 is the detailed block diagram of radiation-resistant video image system of the present invention；
A width field image image in the video that Fig. 5 obtains for the video image acquisition device of the present invention；
The intermediate value video image that Fig. 6 is obtained after being the median method denoising resume module by the present invention；
The video image that Fig. 7 finally gives after being the radioresistance video image system processing by the present invention.
Embodiment
Embodiment 1
As Figure 1-4, a kind of radiation-resistant video image system；It includes video image acquisition device 1 and video figure
As processing unit 3；As shown in Fig. 2 video image acquisition device 1 includes amasthenic lens 11 and video image sensors 12, focus on
Camera lens 11 is convex lens, and its focus adjustable, video image sensors 12 are CCD（CCD, English full name：Charge-coupled
Device, Chinese full name：Charge coupled cell；Ccd image sensor is properly termed as, image controller is also；CCD is a kind of half
Optical image, can be converted into data signal by conductor device）, the light that video image sensors 12 are focused on positioned at amasthenic lens 11
Lu Shang, video image acquisition device 1 can obtain the sequence of video images of detection target（I.e. continuous multiple video images, continuously
Multiple video images constitute video）, each video image is the one of the pixel array, i.e. video of the detection target obtained at certain a moment
Frame, namely a video image, are designated as, it isPixel array (i.e. M rows N row pixel composition video figure
Picture), all videos are expressed as, wherein,,, in display
Each pixelValue be pixel value, video image acquisition device 1 be used for obtain composition video sequence of video images,
Whole sequence of video images is designated as, i.e., described video is by n widthThe video image of sizeComposition.
When under radiation environment, when high energy particle is beaten on the array of video image sensors 12 under intense radiation, being hit
Array pixel, its respective pixel value returned can increase suddenly（I.e. it is seen that video image can be especially bright, namely by radiating
Disturb the video image spot produced in some position of video image）, that is, form video image noise, same width video image
In whole noises constitute video noise image, video image sensors 12（That is CCD）The noise produced under radiation environment is
Random distribution：1. noise is in the same video image of videoIn position distribution be it is random, 2. noises in video before
Different video image afterwardsDistribution be also random, the i.e. complete phase of noise of same position different video image
Same probability is seldom；The video image processing device 3 of the present invention is namely based on this principle and designed.
Connected between video image acquisition device 1 and video image processing device 3 by communication system, described communication system
System can be cable and its interface circuit, or wireless telecommunication system, and wireless telecommunication system includes wireless transceiver, interface
The parts such as circuit；Communication system should also include the circuits such as level conversion, I/V conversions, signal amplification, A/D conversions.
The parts such as video image sensors 12, communication system can also use hardware reinforcement technique, reduction radiation interference production
Raw influence, to reduce image noise.
Video image processing device 3 includes digital processing unit 31 and video image N6ise deletion device 32；Digital processing is filled
Putting 31 includes input interface, memory, processor and output interface, and digital processing unit 31 can be special in the form of expression
Digital processing unit customized, or one can carry out the common computer of Digital Signal Processing（Such as single-chip microcomputer, notes
This computer etc.）；Video image N6ise deletion device 32 is placed in digital processing unit 31, during video image N6ise deletion device 32 includes
Value method denoising module, homodyne video noise image zooming-out module, forward mutual deviation method extract video noise image module, it is mutual backward
Poor method extracts video noise image module, forward synthesis secondary video noise image module, backward synthesis secondary video noise figure
Module, noise image demarcating module, noise eliminating mould are defined as module, preceding poor noise figure picture define module, rear poor noise figure picture
Block, median method video image correcting module.
Median method denoising module, it is used to gather each frame video image of coming to video image sensors 12（It is i.e. each
Width video image）Preliminary denoising is carried out, i.e., it is by each video imageIn
Each pixel with its periphery r arrange（R rows in periphery refer to be located at the pixel each r rows pixel up and down, common including the pixel
Constitute the pixel array of 2r+1 rows 2r+1 row）The median of pixel（Median refers to arrange these pixel values according to size order
Row, centrally located pixel value）Substitute, obtain each intermediate value video image and be designated as,；For example, with
Centered on pixel to be processed, its periphery three row is taken（r=3）Pixel, i.e., all around take three rows to constitute seven rows seven row altogether
Pixel array（7x7 pixel arrays）, then the value of this 49 pixels is arranged from small to large, median pixel value is taken（Arrange the 25th
The pixel value of position）As the pixel value of current pixel point, when the point is king-sized value（I.e. by produced by radiation interference
Value）It will be removed, reach the purpose of denoising；When pixel is located at video imageBoundary value at when（It is less than r from border
Row）, typically without processing, following boundary processing method can also be carried out and handled：Current pixel point is less than r to some direction
When arranging (such as 3 rows), no longer toward value on side, after the number of pixels actually got by size order arrangement, then centre is taken
The pixel value of position as current pixel pixel value；Certainly, it can be taken more than or less than three row's pixels to periphery, typically
The row of desirable r=1~5, in practice it has proved that take the progress median method of r=3 to go noise result preferable.
Homodyne video noise image zooming-out module（So-called " noise figure picture " refers to the figure of the noise components formation in image
Picture, i.e., be the noise pixel value of correspondence noise figure picture beyond the part of normal pixel value, the pixel without noise is corresponding to make an uproar
The level of noise of sound image is zero, similarly hereinafter）, for by each width video imageEach pixel value subtract it and gone by median method
The intermediate value video image obtained after resume module of making an uproarIn corresponding pixel value, and take absolute value, obtain homodyne noise figure
Picture , i.e.,, wherein。
Mutual deviation method extracts video noise image module forward, and it is used for every width video imageWith being come in video before it
Face close to r1 width video images be compared, r1=1~3, r1 is generally equivalent to 2, estimates this frame video imageForward
Noise figure picture, wherein；For example, r1=2 are taken, by each width video imageRespectively with before it
Two width video imagesPoor picture is formed, then by two differences as superposition obtains each width video imageMake an uproar forward
Sound image,, i.e.,,,
Mutual deviation method video noise image zooming-out module does not handle the most preceding two width video image of all videos forward, k values since 3,
Only take, certain this boundary value problem can also handle as follows：WillWithUseSubstitute,
All boundary value problems can make similar processing below, repeat no more, and no matter the method for which kind of processing boundary values does not influence this
The main body of invention is innovative, i.e., the present invention is not limited to boundary values processing form.
Mutual deviation method extracts video noise image module backward, and it can be by by every width video imageWith coming it in video
Immediately behind r1 width video images be compared, r1=1~3, r1 is generally equivalent to 2, estimates this frame video imageTo
Noise figure picture afterwards；For example, r1=2 are taken, by each width video imageRespectively with two width video images behindPoor picture is formed, then by two differences as superposition obtains the noise figure picture backward of each width video image, i.e.,,, whenBoundary values facture similar to the above Shi Kezuo (willWithUseSubstitute), it can not also handle.
Secondary video noise image module is synthesized forward, for by the homodyne noise figure picture corresponding to each width video imageWith noise figure picture forwardIn proportionSynthesize secondary video noise figure picture forward, i.e.,,
Wherein（Certain k can also get boundary value, i.e.,, but for convenience, boundary image can not
Deal with）, i.e., by each width homodyne noise figure pictureEach noise pixel value withIt is multiplied, then will noise forward
ImageEach noise pixel value withIt is multiplied, take two products and is used as secondary video noise figure forward
PictureEach noise pixel value, whereinFor constant, span can between 0.1 to 0.5, in practice it has proved thatTake
Effect is best when 0.25.
Secondary video noise image module is synthesized backward, for by the corresponding homodyne noise figure picture of each width video image
With noise figure picture backwardIn proportionSynthesize secondary video noise figure picture backward, i.e.,,
Wherein, i.e., by each width homodyne noise figure pictureEach noise pixel value withIt is multiplied, then
To noise figure picture backwardEach noise pixel value withIt is multiplied, take two products and is used as backward two
Secondary video noise imageEach noise pixel value, whereinFor constant, span between 0.1 to 0.5 can,
It facts have provedEffect is best when taking 0.25.
Preceding poor noise figure picture defines module, and it is used for secondary video noise figure picture forwardIn each noise picture
The value of vegetarian refreshments defines threshold value with oneIt is compared, threshold value is defined if being more than, then assert that pixel position noise is deposited
Value is 1, otherwise assert that pixel position noise is not present, its value is 0, and poor noise defines image before obtaining,
I.e.：
Wherein,,If assuming above-mentioned r1=2,, boundary is not to handle, i.e.,, can not handle, " close on and replace also referring to above-mentioned certainly
For method " simply handled；Noise defines threshold valueCan between taking 20 to 80, in practice it has proved that effect is best when taking 40.
Poor noise figure picture defines module afterwards, and it is used for secondary video noise figure picture backwardIn each noise picture
The value of vegetarian refreshments defines threshold value with oneIt is compared, threshold value is defined if being more than, then assert that pixel position noise is deposited
, value is 1, otherwise assert the pixel position noise be not present, its value be 0, obtain rear poor noise and define image,
I.e.：
Wherein,,If assuming above-mentioned r1=2,, boundary is not to handle, i.e.,, can not handle, can also refer to certainly above-mentioned
" closing on method of substitution " is simply handled；Noise defines threshold valueCan between taking 20 to 80, in practice it has proved that take 40 timeliness
It is really best.
Noise image demarcating module, for judging that preceding poor noise defines imageImage is defined with rear poor noiseCorresponding noise pixel be noise when, then demarcate the point for noise pixel, represented with " 1 "（It is not limited to use
" 1 " is represented）, otherwise noise do not regard as noise, with " 0 " represent, obtain Noise Calibration image, i.e.,, wherein,,,
OnlyWithWhen taking " 1 " value simultaneously," 1 " value can just be taken.
Noise eliminating module, for by video imageA pixel is extended to the periphery, obtains extending video image,
It is（M+2）X (N+2) image, extends video imageThe pixel and video image of middle extensionThe pixel value phase closed on
Together, i.e.,:
Wherein
Video image will be extended againMiddle correspondence Noise Calibration imageIn be demarcated as noise（As " 1 " value）Position
The pixel put, takes extension video imageIn corresponding four summit pixel value sums in pixel periphery and four, its periphery picture
Element value sum is proportionally（a,b）The new pixel value of synthesis, obtains N6ise deletion video image, i.e.,:
Wherein, a and b sums be equal to a quarter, i.e. a+b=0.25, and preferably a=0.073235, b=
0.176765, being exactly in fact will be to extend video imageMiddle correspondence Noise Calibration imageIn be demarcated as noise（As
" 1 " value, that is）Position pixel centered on 3x3 pixel values composition arrayWithCorrespondence position be multiplied take again and, as making an uproar
Sound rejects the pixel value of corresponding pixel in video image, obtains N6ise deletion video image, wherein a+b=0.25.
Noise eliminating module can be also used for re-using 1~20 time, often carry out next time in use, be will be previous
It is secondary to use obtained N6ise deletion video imageAs next time using the video image startedUsed, then
Obtain next N6ise deletion video image。
Median method video image correcting module, median method video image correcting module and above-mentioned median method denoising module
Structure is identical, and it is used for N6ise deletion video imageA median method denoising is carried out, final denoising is obtained and regards
Frequency image, wherein, , in practice it has proved that final goes
Make an uproar video imageSubstantially the noise that radiation is produced is eliminated, image is totally accurate.
Each " module " that video image N6ise deletion device 32 is included is placed in digital processing unit 31, ordinary skill
As long as personnel describe according to above-mentioned function, method and structure, each " module " can be designed to the hard of logic circuit easily
Part, can also be designed to software form, so applicant is not necessarily to provide each " module " specific logical circuitry or soft again
Part flow chart.
Experiment：Fig. 5 is the width field image image in the video that video image acquisition device is obtained；Fig. 6 is process
The intermediate value video image obtained after median method denoising resume module；Fig. 7 is radiation-resistant video image by the present invention
The denoising video image finally given after system processing, from above-mentioned 3 pictures it can be clearly seen that the present invention can be by
The Computer Vision for being radiated and having " glittering " starlight obtains very clean.
Embodiment 2
A kind of radioresistance method of video image system；It comprises the following steps：
First, video image is obtained by the video image acquisition device of video image system；
2nd, image noise is tentatively gone using median method：Each frame video image that video image acquisition device is obtained
Each pixel arranges what pixel was constituted with it with periphery r（2r+1）x（2r+1）The median pixel value of individual pixel is substituted, and r=1~5 are obtained
To each intermediate value video image；
3rd, video noise image is extracted using homodyne：Each pixel value of each width video image is subtracted in
Corresponding pixel value in the intermediate value video image that value method is obtained, and take absolute value, obtain homodyne noise figure picture；
4th, video noise image is extracted using mutual deviation method forward：It is tight before it by being come in every width video image and video
Adjacent r1 width video images, which are compared, obtains each poor picture, r1=1~3, then by each difference as be added obtain each video image to
Preceding noise figure picture；
5th, video noise image is extracted using mutual deviation method backward, by will be come thereafter in every width video image and video
Face close to r1 width video images be compared and obtain each poor picture, r1=1~3, then each difference as being added is obtained into each video image
Noise figure picture forward；
6th, using secondary video noise figure picture is synthesized forward, by the homodyne noise figure corresponding to each width video image
As with noise figure picture forward in proportionSecondary video noise figure picture forward is synthesized, i.e., by the every of each width homodyne noise figure picture
One noise pixel value withBe multiplied, then by each noise pixel value of noise figure picture forward withIt is multiplied, takes this
Two products and each noise pixel value as secondary video noise figure picture forward；
Using synthesis secondary video noise figure picture backward：By the corresponding homodyne noise figure picture of each width video image with to
Noise figure picture is in proportion afterwardsSecondary video noise figure picture, i.e. each by each width homodyne noise figure picture are made an uproar backward for synthesis
Sound pixel value withBe multiplied, then by each noise pixel value of noise figure picture backward withIt is multiplied, takes this two to multiply
Each noise pixel value long-pending and as secondary video noise figure picture backward；
Wherein=0.1~0.5；
7th, define that preceding poor noise defines image and rear poor noise defines image：By secondary video noise figure forward as in
The value of each noise pixel defines threshold value with oneIt is compared, threshold value is defined if being more than, then the pixel is assert
Position noise is present, and value is 1, otherwise assert that pixel position noise is not present, its value is 0, and poor noise defines figure before obtaining
Picture；
The value of each noise pixel in secondary video noise figure picture backward is defined into threshold value with oneCompared
Compared with if more than defining threshold value, then assert that pixel position noise is present, value is 1, otherwise assert pixel position noise
It is not present, its value is 0, obtains rear poor noise and defines image；
Noise defines threshold value=20~80；
8th, noise image is demarcated：Judge that preceding difference noise defines image and rear poor noise and defines the corresponding noise pixel of image
Be noise when, then demarcate the point for noise pixel, represent that otherwise noise does not regard as noise with " 1 ", represented with " 0 ",
Obtain Noise Calibration image；
9th, cancelling noise：For video image to be extended to the periphery into a pixel, obtain extending video image, then will prolong
The pixel for the position for being demarcated as noise in video image in correspondence Noise Calibration image is stretched, corresponding picture in extension video image is taken
Four, vegetarian refreshments periphery summit pixel value sum synthesizes new pixel value according to parameter a, b with four, its periphery pixel value sum, obtains
To N6ise deletion video image, a is equal to a quarter, i.e. a+b=0.25 with b sums；Namely being exactly in fact will be to extend video
The array for the 3x3 pixel values composition being demarcated as in image in correspondence Noise Calibration image centered on the pixel of the position of noise withCorrespondence position is multiplied to be taken and as the pixel value of corresponding pixel in N6ise deletion video image, obtains again
N6ise deletion video image, wherein a+b=0.25.
Claims (10)
1. a kind of radiation-resistant video image system；It includes video image acquisition device (1) and video image processing device
(3)；Video image acquisition device (1) includes amasthenic lens (11) and video image sensors (12), video image sensors
(12) it is located in the light path that amasthenic lens (11) is focused on, video image acquisition device (1) can obtain the video image of detection target
Sequence, each video image is the image of the detection target obtained at certain a moment；It is characterized in that：Video image acquisition device (1) with
Video image processing device is connected between (3) by communication system；Video image processing device (3) includes digital processing unit
And video image N6ise deletion device (32) (31)；Digital processing unit (31) includes input interface, memory, processor and output
Interface；Video image N6ise deletion device (32) is placed in digital processing unit (31), and video image N6ise deletion device (32) includes
Median method denoising module, homodyne video noise image zooming-out module, forward mutual deviation method extract video noise image module, backward
Mutual deviation method extracts video noise image module, forward synthesis secondary video noise image module, backward synthesis secondary video noise
Image module, preceding poor noise figure picture define module, rear poor noise figure picture and define module, noise image demarcating module, noise eliminating
Module, median method video image correcting module；
Median method denoising module, it is used to tentatively remove each frame video image that video image sensors (12) collection comes
Make an uproar processing, what each pixel that will each in video image was constituted with itself and periphery r row's pixels（2r+1）x（2r+1）Individual pixel
Median pixel value substitute, r=1~5 obtain each intermediate value video image；
Homodyne video noise image zooming-out module, for each pixel value of each width video image to be subtracted by median method
Corresponding pixel value in the intermediate value video image obtained after denoising resume module, and take absolute value, obtain homodyne noise figure picture；
Mutual deviation method extracts video noise image module forward, for by come in every width video image and video before it close to
R1 width video images, which are compared, obtains each poor picture, r1=1~3, then by each difference as addition obtains making an uproar forward for each video image
Sound image；
Mutual deviation method extracts video noise image module backward, for by the way that every width video image is tight behind with being come in video
Adjacent r1 width video images, which are compared, obtains each poor picture, r1=1~3, then by each difference as be added obtain each video image to
Noise figure picture afterwards；
Forward synthesize secondary video noise image module, for by the homodyne noise figure picture corresponding to each width video image with
Noise figure picture is in proportion forwardSynthesis secondary video noise figure picture forward, i.e., by each width homodyne noise figure picture each
Noise pixel value withBe multiplied, then by each noise pixel value of noise figure picture forward withIt is multiplied, takes this two
Product and each noise pixel value as secondary video noise figure picture forward；
Backward synthesize secondary video noise image module, for by the corresponding homodyne noise figure picture of each width video image with to
Noise figure picture is in proportion afterwardsSecondary video noise figure picture, i.e. each by each width homodyne noise figure picture are made an uproar backward for synthesis
Sound pixel value withBe multiplied, then by each noise pixel value of noise figure picture backward withIt is multiplied, takes this two to multiply
Each noise pixel value long-pending and as secondary video noise figure picture backward；
Wherein=0.1~0.5；
Preceding poor noise figure picture defines module, for by the value and one of each noise pixel in secondary video noise figure picture forward
It is individual to define threshold valueIt is compared, threshold value is defined if being more than, then assert that pixel position noise is present, value is 1, no
Then assert that pixel position noise is not present, its value is 0, and poor noise defines image before obtaining；
Poor noise figure picture defines module afterwards, its be used for by the value of each noise pixel in secondary video noise figure picture backward with
One is defined threshold valueIt is compared, threshold value is defined if being more than, then assert that pixel position noise is present, value is 1,
Otherwise assert that pixel position noise is not present, its value is 0, obtains rear poor noise and defines image；
Noise defines threshold value=20~80；
Noise image demarcating module, the corresponding noise pixel of image is defined for judging that preceding poor noise defines image and rear poor noise
Be noise when, then demarcate the point for noise pixel, represent that otherwise noise does not regard as noise with " 1 ", represented with " 0 ",
Obtain Noise Calibration image；
Noise eliminating module, for video image to be extended to the periphery into a pixel, obtains extending video image, then extension is regarded
It is demarcated as the pixel of the position of noise in frequency image in correspondence Noise Calibration image, takes corresponding pixel in extension video image
Four, periphery summit pixel value sum synthesizes new pixel value according to parameter a, b with four, its periphery pixel value sum, is made an uproar
Sound rejects video image；The position for being demarcated as noise in Noise Calibration image will be corresponded in video image to extend in fact
Centered on pixel 3x3 pixel values composition array withCorrespondence position is multiplied to be taken and is regarded as N6ise deletion again
The pixel value of corresponding pixel in frequency image, obtains N6ise deletion video image；Wherein a+b=0.25.
2. radiation-resistant video image system according to claim 1, it is characterized in that：Described noise eliminating module is also used
In re-using 1~20 time, often carry out next time in use, being the N6ise deletion video image for obtaining preceding first use
Used as the video image next time using beginning, obtain next N6ise deletion video image.
3. radiation-resistant video image system according to claim 1 or 2, it is characterized in that：It also includes median method video
Image modification module, median method video image correcting module is identical with the structure of above-mentioned median method denoising module, i.e., use
Identical median method handles image, and it is used to carry out a median method denoising to N6ise deletion video image, obtains final go
Make an uproar video image.
4. radiation-resistant video image system according to claim 1 or 2, it is characterized in that：Described r=3；r1=2.
5. radiation-resistant video image system according to claim 3, it is characterized in that：Described r=3；r1=2.
6. radiation-resistant video image system according to claim 1 or 2, it is characterized in that：Described=0.25；=
40。
7. radiation-resistant video image system according to claim 5, it is characterized in that：Described=0.25；=40。
8. radiation-resistant video image system according to claim 7, it is characterized in that：Described a=0.073235, b=
0.176765。
9. video image system according to claim 1 or 2, it is characterized in that：Described communication system is cable and its connect
Mouth circuit, or be wireless telecommunication system, wireless telecommunication system includes the parts such as wireless transceiver, interface circuit；Video image is passed
Sensor (12), communication system are additionally provided with hardware reinforcement technique；Digital processing unit (31), which is one, can carry out Digital Signal Processing
Computer.
10. a kind of radioresistance method of video image system；It is characterized in that：It comprises the following steps：
First, video image is obtained by the video image acquisition device (1) of video image system；
2nd, image noise is tentatively gone using median method：By the every of each frame video image of video image acquisition device (1) acquisition
Individual pixel arranges what pixel was constituted with it with periphery r（2r+1）x（2r+1）The median pixel value of individual pixel is substituted, and r=1~5 are obtained
Each intermediate value video image；
3rd, video noise image is extracted using homodyne：Each pixel value of each width video image is subtracted by median method
Corresponding pixel value in obtained intermediate value video image, and take absolute value, obtain homodyne noise figure picture；
4th, video noise image is extracted using mutual deviation method forward：By come in every width video image and video before it close to
R1 width video images, which are compared, obtains each poor picture, r1=1~3, then by each difference as addition obtains making an uproar forward for each video image
Sound image；
5th, video noise image is extracted using mutual deviation method backward, by the way that every width video image is tight behind with being come in video
Adjacent r1 width video images, which are compared, obtains each poor picture, r1=1~3, then by each difference as be added obtain each video image to
Noise figure picture afterwards；
6th, using forward synthesize secondary video noise figure picture, by the homodyne noise figure picture corresponding to each width video image with
Noise figure picture is in proportion forwardSynthesis secondary video noise figure picture forward, i.e., by each width homodyne noise figure picture each
Noise pixel value withBe multiplied, then by each noise pixel value of noise figure picture forward withIt is multiplied, takes this two
Product and each noise pixel value as secondary video noise figure picture forward；
Using synthesis secondary video noise figure picture backward：By the corresponding homodyne noise figure picture of each width video image with making an uproar backward
Sound image is in proportionSecondary video noise figure picture backward is synthesized, i.e., by each noise picture of each width homodyne noise figure picture
Element value withBe multiplied, then by each noise pixel value of noise figure picture backward withIt is multiplied, takes two products
With each noise pixel value as secondary video noise figure picture backward；
Wherein=0.1~0.5；
7th, define that preceding poor noise defines image and rear poor noise defines image：Will be each in secondary video noise figure picture forward
The value of noise pixel defines threshold value with oneIt is compared, threshold value is defined if being more than, then assert that pixel position is made an uproar
Sound is present, and value is 1, otherwise assert that pixel position noise is not present, its value is 0, and poor noise defines image before obtaining；
The value of each noise pixel in secondary video noise figure picture backward is defined into threshold value with oneIt is compared, if
More than defining threshold value, then assert that pixel position noise is present, value is 1, otherwise assert that pixel position noise is not deposited
It is 0 in, its value, obtains rear poor noise and define image；
Noise defines threshold value=20~80；
8th, noise image is demarcated：Judge that preceding difference noise defines image and rear poor noise and defines the equal of the corresponding noise pixel of image
During for noise, then the point is demarcated for noise pixel, represents that otherwise noise does not regard as noise with " 1 ", is represented, obtained with " 0 "
Noise Calibration image；
9th, cancelling noise：For video image to be extended to the periphery into a pixel, obtain extending video image, then extension is regarded
It is demarcated as the pixel of the position of noise in frequency image in correspondence Noise Calibration image, takes corresponding pixel in extension video image
Four, periphery summit pixel value sum synthesizes new pixel value according to parameter a, b with four, its periphery pixel value sum, is made an uproar
Sound rejects video image；Namely noise will be demarcated as in correspondence Noise Calibration image to extend in video image in fact
Centered on the pixel of position 3x3 pixel values composition array withCorrespondence position is multiplied to be taken and is used as noise again
The pixel value of corresponding pixel in video image is rejected, N6ise deletion video image is obtained；Wherein a+b=0.25.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410618197.5A CN104333716B (en) | 2014-11-06 | 2014-11-06 | Radiation-resistant video image system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410618197.5A CN104333716B (en) | 2014-11-06 | 2014-11-06 | Radiation-resistant video image system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104333716A CN104333716A (en) | 2015-02-04 |
| CN104333716B true CN104333716B (en) | 2017-07-28 |
Family
ID=52408358
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410618197.5A Expired - Fee Related CN104333716B (en) | 2014-11-06 | 2014-11-06 | Radiation-resistant video image system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104333716B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112261474A (en) * | 2020-10-26 | 2021-01-22 | 安徽众新电子科技有限公司 | Multimedia video image processing system and processing method |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102075669A (en) * | 2009-11-24 | 2011-05-25 | 富士通微电子(上海)有限公司 | Method and system for adaptively recursively denoising digital video signal |
| CN102238316A (en) * | 2010-04-29 | 2011-11-09 | 北京科迪讯通科技有限公司 | Self-adaptive real-time denoising scheme for 3D digital video image |
| CN102263982A (en) * | 2010-05-31 | 2011-11-30 | 北京创毅视讯科技有限公司 | Method and device for improving moving visibility of analogue television |
| CN102263965A (en) * | 2010-05-31 | 2011-11-30 | 北京创毅视讯科技有限公司 | Method and device for eliminating noises from analogue television signal |
| CN102479383A (en) * | 2010-11-30 | 2012-05-30 | 上海银晨智能识别科技有限公司 | Method and device for removing salt and pepper noise |
| JP2012129617A (en) * | 2010-12-13 | 2012-07-05 | Canon Inc | Image processing device and method |
| CN103501401A (en) * | 2013-10-01 | 2014-01-08 | 中国人民解放军国防科学技术大学 | Real-time video de-noising method for super-loud noises based on pre-filtering |
-
2014
- 2014-11-06 CN CN201410618197.5A patent/CN104333716B/en not_active Expired - Fee Related
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102075669A (en) * | 2009-11-24 | 2011-05-25 | 富士通微电子(上海)有限公司 | Method and system for adaptively recursively denoising digital video signal |
| CN102238316A (en) * | 2010-04-29 | 2011-11-09 | 北京科迪讯通科技有限公司 | Self-adaptive real-time denoising scheme for 3D digital video image |
| CN102263982A (en) * | 2010-05-31 | 2011-11-30 | 北京创毅视讯科技有限公司 | Method and device for improving moving visibility of analogue television |
| CN102263965A (en) * | 2010-05-31 | 2011-11-30 | 北京创毅视讯科技有限公司 | Method and device for eliminating noises from analogue television signal |
| CN102479383A (en) * | 2010-11-30 | 2012-05-30 | 上海银晨智能识别科技有限公司 | Method and device for removing salt and pepper noise |
| JP2012129617A (en) * | 2010-12-13 | 2012-07-05 | Canon Inc | Image processing device and method |
| CN103501401A (en) * | 2013-10-01 | 2014-01-08 | 中国人民解放军国防科学技术大学 | Real-time video de-noising method for super-loud noises based on pre-filtering |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104333716A (en) | 2015-02-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20180262744A1 (en) | Systems, methods and apparatuses for stereo vision | |
| EP4109392A1 (en) | Image processing method and image processing device | |
| CN104240240B (en) | A kind of infrared small target detection method and system based on FPGA | |
| CN104580858B (en) | A kind of unmanned plane is imaged load with multi-source | |
| CN110473185A (en) | Image processing method and device, electronic equipment, computer readable storage medium | |
| CN112651903B (en) | Thermal infrared imager image preprocessing system and preprocessing method thereof | |
| CN102509304A (en) | Intelligent optimization-based camera calibration method | |
| JP2000512790A (en) | Data processing system and method | |
| US8817121B2 (en) | Image apparatus and imaging method | |
| CN113222819B (en) | Remote sensing image super-resolution reconstruction method based on deep convolution neural network | |
| CN105869132A (en) | Method for detail enhancement of infrared image | |
| CN110880163A (en) | Low-light color imaging method based on deep learning | |
| Qiu et al. | Deep end-to-end alignment and refinement for time-of-flight RGB-D module | |
| Zhang et al. | An effective decomposition-enhancement method to restore light field images captured in the dark | |
| CN103907134A (en) | Image processing device, image processing method, image-capturing device, computer program, and recording medium | |
| CN104333716B (en) | Radiation-resistant video image system and method | |
| CN104463774B (en) | A kind of three tunnel image co-registration processor design methods based on DM642 | |
| CN113936138A (en) | Target detection method, system, equipment and medium based on multi-source image fusion | |
| CN104735349A (en) | Synchronous multi-focus Bayer video picture processing system and method | |
| CN111866476B (en) | Image shooting method and device and electronic equipment | |
| Zhou et al. | Raw bayer pattern image synthesis for computer vision-oriented image signal processing pipeline design | |
| CN114125319A (en) | Image sensor, camera module, image processing method and device and electronic equipment | |
| CN103198462B (en) | Intense radiation resistant visual sensing device based on information fusion | |
| Palani et al. | Pixel Optimization Using Iterative Pixel Compression Algorithm for Complementary Metal Oxide Semiconductor Image Sensors | |
| Xu et al. | Deep Learning-Based Pedestrian Detection Using RGB Images and Sparse LiDAR Point Clouds |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170728 Termination date: 20181106 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |