CN103747213B - A kind of real-time defogging method of the Traffic Surveillance Video based on moving target - Google Patents
A kind of real-time defogging method of the Traffic Surveillance Video based on moving target Download PDFInfo
- Publication number
- CN103747213B CN103747213B CN201410017407.5A CN201410017407A CN103747213B CN 103747213 B CN103747213 B CN 103747213B CN 201410017407 A CN201410017407 A CN 201410017407A CN 103747213 B CN103747213 B CN 103747213B
- Authority
- CN
- China
- Prior art keywords
- video
- frame
- image
- mist elimination
- traffic surveillance
- 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
- 238000000034 method Methods 0.000 title claims abstract description 53
- 239000003595 mist Substances 0.000 claims abstract description 61
- 238000003379 elimination reaction Methods 0.000 claims abstract description 50
- 230000008030 elimination Effects 0.000 claims abstract description 40
- 238000002835 absorbance Methods 0.000 claims description 20
- 238000001914 filtration Methods 0.000 claims description 11
- 230000005540 biological transmission Effects 0.000 claims description 8
- 230000003247 decreasing effect Effects 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 15
- 238000003672 processing method Methods 0.000 abstract description 6
- 239000003086 colorant Substances 0.000 abstract description 3
- 230000000694 effects Effects 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000012467 final product Substances 0.000 description 1
- 239000008187 granular material Substances 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 230000026676 system process Effects 0.000 description 1
Landscapes
- Closed-Circuit Television Systems (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of real-time defogging method of the Traffic Surveillance Video based on moving target, according to the characteristics of Traffic Surveillance Video, the content of frame of video is divided into into foreground moving object and background parts, different methods are respectively adopted to foreground and background carries out mist elimination process, and wherein emphasis processes foreground moving object.Meanwhile, using the dependency between video consecutive frame, to method of the background using not estimating or estimating less, substantially increase processing speed.SD video for 720 × 576 can reach the processing speed of 25 frames/second, meet the demand of real-time processing, overcome present in prior art because defogging method processes the problem that complexity height cannot meet Traffic Surveillance Video real time handling requirement.Mist elimination processing method proposed by the present invention is more targeted, can more efficiently retain the true colors of foreground moving object, reduces cross-color, is that the intelligent processing method of further traffic video is laid a good foundation.
Description
Technical field
The invention belongs to image/video signals process field, is related to a kind of Traffic Surveillance Video reality based on moving target
When defogging method.
Background technology
Serious haze weather can cause serious impact to the quality of Traffic Surveillance Video, and mist elimination process is to traffic monitoring
The follow-up intelligent analysis work of video is significant.The key of mist elimination process is to estimate absorbance and air light intensity.
If estimate and, it is possible to by observed image obtain mist elimination after image.At present, the mist elimination based on dark primary priori
Method is main stream approach.
Dark primary priori rule is pointed out：In a regional area of most natural images, some pixels are at least
One Color Channel has very low value.According to dark primary priori rule, it is estimated that air light intensity and absorbance.
The method of estimation of air light intensity is：It is in dark primary image, first that the brightness value of each pixel is suitable according to successively decreasing
Sequence according to sequence, it is then determined that numerical values recited be front 0.1% point in dark primary image location, finally find out these
Valuation of the maximum in original fog image region corresponding to position as air light intensity.
The method of estimation of absorbance is：First, each pixel for having mist image to observing is taken in tri- passages of RGB
Minima, obtain a width gray level image.Next mini-value filtering operation is carried out to gray level image, is then subtracted with air light intensity
Go filtered image each point gray scale be worth to transmission plot.Transmission plot obtains final product normalized transmission divided by air light intensity
Rate.
The patent of Application No. CN201210125321.5 discloses a kind of video image mist elimination based on self adaptation tolerance
Method.Estimation of the method using guiding filtering to absorbance is refined, and so as to obtain more careful transmission plot, makes mist elimination
As a result it is finer and smoother.But the process of guiding filtering will consume the substantial amounts of time, it is impossible to meet the requirement of real-time of Video processing.
The patent of Application No. CN201110134572.5 discloses a kind of real-time video defogging system.The system does not have
Have time-consuming huge guiding filtering process, and realize above-mentioned mist elimination step using digital integrated electronic circuit, by it is hardware-accelerated come
Lift the speed that mist elimination is processed.The video rate of system process 288 × 352 processes 720 × 576 SD up to 60 frames/second
Video rate is up to 15 frames/second.The method can the less video of real-time processing size, but the SD video for 720 × 576
Then can only substantially meet real time handling requirement.
In addition, above two method does not differentiate between the content of image when processing.It is for Traffic Surveillance Video, this etc.
Same processing method effect is unsatisfactory.This is because in order to ensure image it is overall remove fog effect, often cause to monitor field
Moving target in scape produces cross-color, produces certain impact to follow-up intelligent analysis.
The content of the invention
The present invention proposes a kind of real-time defogging method of Traffic Surveillance Video based on moving target.The method is according to traffic
The characteristics of monitor video, the content of frame of video is divided into into foreground moving object and background parts, foreground and background is respectively adopted
Different methods are processed.The color characteristics of foreground target so can be retained well, wanting for real-time processing can be met again
Ask, be that follow-up transport information intelligent processing method lays the first stone.
For achieving the above object, the present invention is employed the following technical solutions.
1. the division of video content
The intelligent processing method of Traffic Surveillance Video is more concerned with moving target, therefore should ensure emphatically this partial content
Mist elimination treatment effect.To the Traffic Surveillance Video for collecting, frame of video is divided into fortune initially with neighbor frame difference method by the present invention
Two parts of moving-target and background.For this two parts content, different defogging methods are respectively adopted and are processed, to ensure to go
Video quality after mist.
2. the mist elimination of different content is processed
The value of air light intensity and absorbance is estimated first, and the image after mist elimination is obtained by observed image then.Absorbance
Estimation is most time-consuming part during whole mist elimination.The content for how being directed to Traffic Surveillance Video is described below, air is estimated
The value of light intensity and absorbance, meets the requirement of real-time processing.
（1）The estimation of air light intensity
Air light intensity generally keeps constant within long period of time, it is therefore not necessary to estimate greatly to each frame of video
The value of gas light intensity.The present invention estimates the value of an air light intensity at set intervals to frame of video, so as to greatly reduce air
The computation complexity of light intensity valuation.
（2）The estimation of absorbance
For the frame of video that have estimated air light intensity, the estimation of absorbance is carried out, and whole frame of video is carried out at mist elimination
Reason.For other frame of video, moving target and background parts are respectively processed.
1）The mist elimination of moving target is processed
In order to the process to foreground moving object is more targeted, also for mist elimination processing speed is lifted, the present invention is only
The estimation and mist elimination that absorbance is carried out to foreground moving object is processed, and eliminates mini-value filtering process, to improve absorbance
The speed of estimation.
In real life, some granules in air, are constantly present, people still can be felt when the object of distant place is observed
Feel the impact of mist, the presence of mist in addition can allow people to feel the presence of the depth of field, it is therefore necessary to retain certain when mist elimination
The mist of degree, to strengthen sense of reality and the depth of field sense of image.For this purpose, present invention introduces a decay factor is controlling mist elimination power
Degree.
2）The mist elimination of background is processed
For Traffic Surveillance Video, its background is substantially stationary, changes little between consecutive frame, therefore the present invention
Directly using the mist elimination result in previous frame, the value of absorbance is no longer reevaluated, can so greatly promote mist elimination process
Speed.
Moving target and background area after mist elimination is combined, and obtains a complete mist elimination frame of video.
Compared with prior art, the present invention has following obvious advantage and beneficial effect：
1. computation complexity is low, it is fast to perform speed.It is high that existing defogging method processes complexity, it is impossible to meets traffic monitoring
The demand of video real-time processing.The content of frame of video is divided into foreground moving mesh according to the characteristics of Traffic Surveillance Video by the present invention
Mark and background two parts, are respectively processed, and wherein emphasis processes foreground moving object.Meanwhile, using between video consecutive frame
Dependency, to background using the method not estimating or estimate less, substantially increase processing speed.Mark for 720 × 576
Clear video can reach the processing speed of 25 frames/second, meet the demand of real-time processing.
2. mist elimination is processed more targetedly, can more efficiently retain the true colors of foreground moving object.This
Bright that foreground moving object and background are treated with a certain discrimination, emphasis processes foreground moving object, is at follow-up transport information intellectuality
Reason is laid a good foundation.
Description of the drawings
Fig. 1 is the flow chart of defogging method involved in the present invention；
Fig. 2 is to carry out the image after Traffic Surveillance Video mist elimination before processing using the present invention,（a）For the image before mist elimination,
（b）For the image after mist elimination；
Fig. 3 is the color contrast image that automobile is moved forward and backward using the present invention and guiding filtering method mist elimination,（a）For mist elimination
Front automobile image,（b）Be using the automobile image after guiding filtering method mist elimination,（c）It is using the vapour after mist elimination of the present invention
Car image.
Specific embodiment
Below in conjunction with the drawings and specific embodiments, the present invention will be further described.
The flow chart of the real-time defogging method of the Traffic Surveillance Video based on moving target proposed by the present invention as shown in figure 1,
Specifically include following steps：
Step 1, carries out mist elimination process to the first frame of Traffic Surveillance Video.
Step 1.1, estimates air light intensity.
（1）To there is mist image to take the minimum pixel value in tri- Color Channels of RGB and carry out mini-value filtering, obtain dark former
Color image：
In formula, JdarkX () is dark primary image；Ic(y) be observe have mist image, y ∈ Ω (x), Ω (x) are with picture
A boxed area centered on vegetarian refreshments y.
（2）By the brightness value of pixel in dark primary image by order sequence of successively decreasing, determine that numerical values recited is front 0.1%
Point location in dark primary image；
（3）Find out the maximum of the air light intensity in the original fog image region corresponding to these positions, i.e. air light intensity
Valuation.
Step 1.2, estimates absorbance.
To there is mist image to take the minimum pixel value in tri- Color Channels of RGB, a width gray level image is obtained, then uses atmosphere light
Strong valuation A deducts gray value of the image each point and obtains transmission plot with the product of attenuation quotient, finally with transmission plot divided by A
Obtain normalized absorbance：
In formula,For normalized absorbance valuation；ω is an attenuation quotient, for controlling the intensity of mist elimination, 0 ＜
ω≤1, ω generally take 0.7.
Step 1.3, carries out mist elimination process.
Image after mist elimination is：
In formula, t0It is the lower limit of absorbance, is usually arranged as 0.1；K is tolerance, is unsatisfactory for dark primary elder generation for amendment
The bright areas of hypothesis are tested, K generally takes 0.5-0.6 times of A values.
Frame of video, from the beginning of the second frame, is divided into foreground moving object and background two using neighbor frame difference method by step 2
Point, different defogging methods are respectively adopted to this two parts then and are processed.
Step 2.1, the frame for calculating present frame and former frame are poor, for the three-channel differences of RGB are no more than the portion of threshold value T
Point, it is judged as background area, remainder is judged as foreground moving object region.Threshold value T is usually taken to be 2.
Step 2.2, if foreground moving object region, then execution step 1.2,1.3, carry out mist elimination to which；If background
Region, then directly substituted using the mist elimination result of relevant position in previous frame, no longer carries out defogging.
Step 2.3, repeat step 2, until having processed all frame of video.
In order to ensure that actual value is more nearly to the estimation of air light intensity A, step 1 and 2 are repeated at set intervals,
Estimation is re-started to A.
Shown in Fig. 2 is the Traffic Surveillance Video mist elimination effect contrast figure obtained using method proposed by the present invention.From figure
In as can be seen that using method proposed by the present invention, it is possible to obtain remove fog effect well.
In order to comparison it is of the invention compared with prior art remove fog effect, respectively using of the invention and of the prior art draw
Leading filtering method carries out mist elimination process to one section of Traffic Surveillance Video.Fig. 3 is the comparison that two methods remove fog effect.Can from figure
It is to find out, higher using the picture contrast after the mist elimination of guiding filtering method, but the color of automobile is too deep, occurs in that substantially
Cross-color.Image after the mist elimination of the present invention preferably remains the true colors of foreground moving object, is further
Traffic video intelligent processing method is laid a good foundation.
The SD video rate of the patent process 720 × 576 of Application No. CN201110134572.5 is up to 15 frames/second.
Experiment shows that the SD video rate of method process proposed by the present invention 720 × 576 meets SD video up to 25 frames/second
Real time handling requirement., up to 13-14 frames/second, processing speed is very fast for the HD video speed of process 1280 × 720.
Claims (2)
1. the real-time defogging method of a kind of Traffic Surveillance Video based on moving target, it is characterised in that according to Traffic Surveillance Video
The characteristics of, the content of frame of video is divided into into foreground moving object and background parts, different sides are respectively adopted to foreground and background
Method carries out mist elimination process, comprises the following steps：
Step 1, carries out mist elimination process to the first frame of Traffic Surveillance Video；
Step 1.1, estimates air light intensity；
(1) to there is mist image to take the minimum pixel value in tri- Color Channels of RGB and carry out mini-value filtering, obtain dark primary figure
Picture：
In formula, JdarkX () is dark primary image；Ic(y) be observe have mist image, y ∈ Ω (x), Ω (x) are with pixel y
Centered on a boxed area；
(2) by order sequence of successively decreasing, the brightness value of pixel in dark primary image is determined into that the point that numerical values recited is front 0.1% exists
Location in dark primary image；
(3) max pixel value in the original fog image region corresponding to these positions, as the valuation A of air light intensity are found out；
Step 1.2, estimates absorbance；
To there is mist image to take the minimum pixel value in tri- Color Channels of RGB, a width gray level image is obtained, then is estimated with air light intensity
Value A deducts the product of gray value and the attenuation quotient of the image each point and obtains transmission plot, is finally obtained divided by A with transmission plot
Normalized absorbance：
In formula,For normalized absorbance valuation；ω is an attenuation quotient, for controlling the intensity of mist elimination, 0 ＜ ω≤
1；
Step 1.3, carries out mist elimination process；
Image after mist elimination is：
In formula, t0It is the lower limit of absorbance, is set to 0.1；K is tolerance, is unsatisfactory for the bright of dark primary a priori assumption for amendment
Bright area, K take 0.5-0.6 times of A values；
Frame of video, from the beginning of the second frame, is divided into foreground moving object and background two parts using neighbor frame difference method, so by step 2
Different defogging methods are respectively adopted to this two parts afterwards to process；
Step 2.1, the frame for calculating present frame and former frame are poor, for the three-channel differences of RGB are no more than the part of threshold value T,
It is judged as background area, remainder is judged as foreground moving object region；Threshold value T takes 2；
Step 2.2, if foreground moving object region, then execution step 1.2,1.3, carry out mist elimination to which；If background area,
Then directly substituted using the mist elimination result of relevant position in previous frame, no longer carried out defogging；
Step 2.3, repeat step 2, until having processed all frame of video.
2. the real-time defogging method of a kind of Traffic Surveillance Video based on moving target according to claim 1, its feature exist
In, in order to ensure that actual value is more nearly to the estimation of air light intensity A, at set intervals between repeat step 1 and 2, to A
Estimation is re-started, ω takes 0.7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410017407.5A CN103747213B (en) | 2014-01-15 | 2014-01-15 | A kind of real-time defogging method of the Traffic Surveillance Video based on moving target |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410017407.5A CN103747213B (en) | 2014-01-15 | 2014-01-15 | A kind of real-time defogging method of the Traffic Surveillance Video based on moving target |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103747213A CN103747213A (en) | 2014-04-23 |
| CN103747213B true CN103747213B (en) | 2017-04-05 |
Family
ID=50504199
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410017407.5A Expired - Fee Related CN103747213B (en) | 2014-01-15 | 2014-01-15 | A kind of real-time defogging method of the Traffic Surveillance Video based on moving target |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103747213B (en) |
Families Citing this family (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105282421B (en) * | 2014-07-16 | 2018-08-24 | 宇龙计算机通信科技(深圳)有限公司 | A kind of mist elimination image acquisition methods, device and terminal |
| CN104200435A (en) * | 2014-08-29 | 2014-12-10 | 武汉大学 | Single-image defogging method based on fog imaging model |
| CN104408757B (en) * | 2014-11-07 | 2017-11-14 | 吉林大学 | The method and system of haze effect are added in a kind of video to Driving Scene |
| CN104717400A (en) * | 2015-02-03 | 2015-06-17 | 北京理工大学深圳研究院 | Real-time defogging method of monitoring video |
| CN104732497A (en) * | 2015-03-31 | 2015-06-24 | 北京交通大学 | Image defogging method, FPGA and defogging system including FPGA |
| US9508129B1 (en) * | 2015-11-06 | 2016-11-29 | Adobe Systems Incorporated | Dehazing photos and videos using visual artifact suppression |
| CN105550999A (en) * | 2015-12-09 | 2016-05-04 | 西安邮电大学 | Video image enhancement processing method based on background reuse |
| CN105635526A (en) * | 2015-12-25 | 2016-06-01 | 国网重庆市电力公司电力科学研究院 | Real-time video defogging method and system applied in video monitoring system |
| CN105959510B (en) * | 2016-04-22 | 2019-04-26 | 湖南工学院 | A kind of video rapid defogging method |
| CN106340009B (en) * | 2016-08-18 | 2018-12-11 | 河海大学常州校区 | A kind of electric power line detecting method and system based on parallel binocular |
| CN107194894A (en) * | 2017-05-25 | 2017-09-22 | 河南师范大学 | A kind of video defogging method and its system |
| CN107360344B (en) * | 2017-06-27 | 2020-02-21 | 西安电子科技大学 | Rapid defogging method for monitoring video |
| CN107277299B (en) * | 2017-07-27 | 2020-08-18 | Oppo广东移动通信有限公司 | Image processing method, image processing device, mobile terminal and computer readable storage medium |
| CN107451969B (en) * | 2017-07-27 | 2020-01-10 | Oppo广东移动通信有限公司 | Image processing method, image processing device, mobile terminal and computer readable storage medium |
| CN107454318B (en) * | 2017-07-27 | 2020-01-10 | Oppo广东移动通信有限公司 | Image processing method, image processing device, mobile terminal and computer readable storage medium |
| CN108830803A (en) * | 2018-05-17 | 2018-11-16 | 昆明理工大学 | A kind of traffic video image defogging optimization algorithm |
| CN108921805B (en) * | 2018-07-06 | 2020-11-03 | 厦门理工学院 | Image and video haze removing method, computer device and storage medium |
| CN110852965A (en) * | 2019-10-31 | 2020-02-28 | 湖北大学 | Video illumination enhancement method and system based on generation countermeasure network |
| CN112750089B (en) * | 2020-12-27 | 2023-03-28 | 同济大学 | Optical remote sensing image defogging method based on local block maximum and minimum pixel prior |
| CN114638763B (en) * | 2022-03-24 | 2024-05-24 | 华南理工大学 | Image defogging method, system, computer device and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103079062A (en) * | 2013-02-05 | 2013-05-01 | 武汉科技大学 | Intelligent video surveillance system |
| CN103093202A (en) * | 2013-01-21 | 2013-05-08 | 信帧电子技术（北京）有限公司 | Car logo locating method and car logo locating device |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103186887B (en) * | 2011-12-30 | 2016-08-03 | 株式会社日立制作所 | Image demister and image haze removal method |
-
2014
- 2014-01-15 CN CN201410017407.5A patent/CN103747213B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093202A (en) * | 2013-01-21 | 2013-05-08 | 信帧电子技术（北京）有限公司 | Car logo locating method and car logo locating device |
| CN103079062A (en) * | 2013-02-05 | 2013-05-01 | 武汉科技大学 | Intelligent video surveillance system |
Non-Patent Citations (2)
| Title |
|---|
| 数字图像去雾与超分辨率重建算法研究;侯天峰;《中国优秀硕士学位论文全文数据库（电子期刊）》;20110915;正文第7-21页 * |
| 视频监控图像去雾方法研究;邵长艳;《中国优秀硕士学位论文全文数据库（电子期刊）》;20121015;正文第35-75页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103747213A (en) | 2014-04-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103747213B (en) | A kind of real-time defogging method of the Traffic Surveillance Video based on moving target | |
| CN102750674B (en) | Video image defogging method based on self-adapting allowance | |
| CN107301623B (en) | Traffic image defogging method and system based on dark channel and image segmentation | |
| CN103077504B (en) | A kind of image defogging method capable based on self-adaptation illumination calculation | |
| CN102202163B (en) | Adaptive enhancement method and device for monitored video | |
| CN103049888A (en) | Image/video demisting method based on combination of dark primary color of atmospheric scattered light | |
| CN103996178A (en) | Sand and dust weather color image enhancing method | |
| CN107301624A (en) | The convolutional neural networks defogging algorithm pre-processed based on region division and thick fog | |
| CN103985091A (en) | Single image defogging method based on luminance dark priori method and bilateral filtering | |
| CN104036466A (en) | Video defogging method and system | |
| CN105046677A (en) | Enhancement processing method and apparatus for traffic video image | |
| CN107358585A (en) | Misty Image Enhancement Method based on fractional order differential and dark primary priori | |
| CN105631831A (en) | Video image enhancement method under haze condition | |
| CN104166968A (en) | Image dehazing method and device and mobile terminal | |
| CN107360344A (en) | Monitor video rapid defogging method | |
| CN102306307B (en) | Positioning method of fixed point noise in color microscopic image sequence | |
| CN104331867B (en) | The method, device and mobile terminal of image defogging | |
| CN102523364A (en) | Document image strike-through eliminating method and system | |
| CN103020921A (en) | Single image defogging method based on local statistical information | |
| CN103729828A (en) | Video rain removing method | |
| CN111598814B (en) | Single image defogging method based on extreme scattering channel | |
| CN103020930A (en) | Nighttime monitoring video enhancing method | |
| CN104021527A (en) | Rain and snow removal method in image | |
| CN104809700A (en) | Low-light video real-time enhancement method based on bright channel | |
| CN105046670A (en) | Image rain removal method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| 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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170405 |