CN100546380C - Target detection and tracking at night based on visual characteristic - Google Patents
Target detection and tracking at night based on visual characteristic Download PDFInfo
- Publication number
- CN100546380C CN100546380C CNB2005101306960A CN200510130696A CN100546380C CN 100546380 C CN100546380 C CN 100546380C CN B2005101306960 A CNB2005101306960 A CN B2005101306960A CN 200510130696 A CN200510130696 A CN 200510130696A CN 100546380 C CN100546380 C CN 100546380C
- Authority
- CN
- China
- Prior art keywords
- target
- night
- frame
- contrast
- frames
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 41
- 230000000007 visual effect Effects 0.000 title claims abstract description 11
- 238000000034 method Methods 0.000 claims abstract description 29
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 16
- 230000003044 adaptive effect Effects 0.000 claims abstract description 7
- 238000004364 calculation method Methods 0.000 claims description 11
- 230000008859 change Effects 0.000 claims description 10
- 238000012935 Averaging Methods 0.000 claims 1
- 230000008569 process Effects 0.000 abstract description 10
- 238000004458 analytical method Methods 0.000 abstract description 6
- 238000012545 processing Methods 0.000 abstract description 5
- 238000001914 filtration Methods 0.000 abstract description 2
- 238000012544 monitoring process Methods 0.000 description 18
- 238000005516 engineering process Methods 0.000 description 9
- 238000011160 research Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000005286 illumination Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 230000002411 adverse Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000003331 infrared imaging Methods 0.000 description 2
- 230000005855 radiation Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000035800 maturation Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000008430 psychophysiology Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
Abstract
A kind of target detection and tracking at night based on visual characteristic comprises step: catch the video at night; Adopt adaptive algorithm decision frame-to-frame differences; Determine the target area according to contrast; Current location and next position constantly according to the velocity information target of prediction.The present invention is the similarity between judgment frame exactly, helps the processing of subsequent process.Introduced these mankind's of contrast basic vision characteristic and expressed image, made algorithm not only at night but also also can be effective at atrocious weather such as greasy weather situation.The method of some trajectory analysises is used to feed back to detection-phase, to the variation filtering preferably of noise and light, has further improved the accuracy that detects.
Description
Technical field
The present invention relates to pattern recognition, particularly based on vision night target the detection and tracking method.
Background technology
Along with the develop rapidly of modern science and technology, utilize video camera to monitor the every aspect that dynamic scene is widely used in modern society already, particularly those are to the occasion of safety requirements sensitivity, as national defence, community, bank, parking lot, military base etc.The vision monitoring of dynamic scene is the forward position research direction that receives much concern in recent years, and its detection from the video camera sequences of images captured, identification, tracking target are also understood its behavior.Although the present rig camera that extends as human vision ubiquity in commerce is used is not given full play to its initiatively effect of supervision media in real time.Therefore, develop automaticity with practical significance, intelligent visual monitor system becomes urgent and necessary day by day.This just requires and can not only replace human eye with video camera, and the general-purpose computers contributor, replaces the people, monitors or control task to finish.
At present, most researcher is primarily focused on the key technology that solves in the scene monitoring on daytime.Yet the ratio that night and adverse weather account for the annual time is quite big, will account for basically more than the annual half the time.Simultaneously, because the influence of adverse weather, especially night low-light (level) influence, the effect that common monitoring equipment (ccd video camera) was taken in a such time period is very bad, be unfavorable for very much occasion security personnel's such as community, mansion monitoring, more be unfavorable for analysis and evidence obtaining after case takes place.And the purpose of nighttime vision monitoring is exactly to improve the visuality of night monitoring image, give the visual capacity of computer scene information around correct understanding at night (evening), this is for (for example improving those security-sensitive occasions, the parking lot, residential quarters), has great Research Significance in the fail safe at night.Nighttime vision monitoring at first utilizes equipment such as video camera monitored scene to be carried out the collection of raw image data, therefore, nighttime vision monitoring comprises and utilizes image processing techniques that the image that collects is strengthened or preliminary treatment, night background modeling and then carry out motion target detection, tracking, at last, exercise data to resulting moving target carries out semantic analysis, judges and understand the behavior of moving object, wherein night target detection and tracking be vital.
Early stage night monitoring utilizes common camera to add the combination of high strength searchlight, by searchlight illumination is carried out in the monitoring area at night and strengthens, and then by common camera IMAQ is carried out in the monitoring area.This mainly is the influence that the image illumination deficiency is caught in compensation.Corresponding post-processed is exactly directly to improve enhancement algorithms such as brightness, contrast to handle nighttime image to obtain the image of better quality.The factor that influences the picture quality that the common CCD camera takes is more, mainly comprises low contrast, low-light (level), low color saturation and The noise.More external companies have designed corresponding special equipment.Real-time low-light (level) video such as the Enpiction Image Technologies company of the U.S. strengthens module IPM-1000TM[2] and the product NightView[3 of DynaPel Systems Inc company] or the like all be to strengthen consideration from video, but these products are mainly from improving visual angle, and this enhancement process strengthened noise, do not help follow-up target detection and follows the tracks of and handle.Utilize infrared imagery technique to catch the target at night, infrared video carries out target detection and track algorithm research then.Because infrared technique day by day maturation impels it constantly to strengthen in the monitoring Application for Field, but still is subjected to the restriction of price.Infrared imaging mainly divides two kinds of active infrared and passive infrareds.At present, the image quality of passive infrared is better than active infrared.Infrared imaging equipment has many advantages: it does not rely on illumination condition, and by day or all can use evening, it has extended the field range of camera in the daytime; Compare with common camera, it is subjected to the influence of shade less [4].Need only the thermal radiation property of moving object and the thermal radiation property of background and there are differences, then general equal can from thermal infrared images, the detecting of moving object.But cost an arm and a leg it can not be popularized.From the visible light angle, it is also few that common CCD camera video captured is detected the present research of target automatically.
Summary of the invention
The purpose of this invention is to provide a kind of based on the next automatic detection and tracking target at night of visual characteristic.
For achieving the above object, a kind of target detection and tracking at night based on visual characteristic comprises step:
A. catch the video at night;
B. adopt adaptive algorithm decision frame period, choose the frame enough big with the present frame otherness; This adaptive algorithm is judged the interval of selecting automatically by the similarity of calculating between two frames, if the correlation between two frames is less than threshold value, show that so two frame differences are little, then continue to compare the correlation of next frame and present frame, up to the frame period that finds between the two enough big frames of otherness;
C. determine the target area according to contrast in two frames that the otherness that finds is enough big in step b;
D. further determine moving target based on the contrast change information of two enough big frames of the otherness that in step b, finds;
E. according to the current location of velocity information target of prediction and next position constantly: consider uniform motion position calculation model, obtain the prediction reference point by detected target trajectory point sequence before; Calculate the movement velocity of target again by before detected target location; Obtain the predicted position of target then by prediction reference point and movement velocity.
The present invention is the similarity between judgment frame exactly, helps the processing of subsequent process.Introduced these mankind's of contrast basic vision characteristic and expressed image, made algorithm not only at night but also also can be effective at atrocious weather such as greasy weather situation.The method of some trajectory analysises is used to feed back to detection-phase, and to the variation filtering preferably of noise and light, this has further improved the accuracy that detects.
Description of drawings
Fig. 1 is target detection tracking at a night block diagram, comprises target detection and follows the tracks of two parts;
Fig. 2 is target contrast calculating at night;
Fig. 3 is a visual contrast degree result of calculation;
Fig. 4 is a contrast change calculations instance graph;
Fig. 5 is target following forecast model figure;
Fig. 6 is the vision monitoring platform;
Fig. 7 is traffic scene vehicle at night detection and tracking exemplary plot;
Fig. 8 is people's at night target detection and trace example figure;
Fig. 9 is that faint light is shone target detection and trace example figure down;
Figure 10 is a target detection exemplary plot under the dense fog condition;
Figure 11 is a target detection and follow the tracks of evaluation graph at night.
Embodiment
The present invention be mainly concerned with human vision property extraction, night target signature extraction, night target problems such as detection and tracking.Because it is a strong tool that human vision property has been verified at image processing, computer vision field; In addition, directly perceived go up night the people be easy to tell mobile target, still can introduce human vision property, in the detection and tracking process, introduce the change information of these characteristics then along with the time.Consider that to sum up the present invention has realized a kind of target detection and track algorithm at night based on human vision property, the technical scheme flow chart of whole invention is seen shown in the accompanying drawing 1.
Whole proposal of the present invention mainly comprises target detection and follows the tracks of two processes.Different with the monitoring of routine, intelligent monitoring need replace the automatic detection and tracking target of people with computer, it can night automatically detect mobile target and further evaluating objects whether unusually, such as whether entering important departments such as bank, airport and reporting to the police and handle.With regard to the target detection process, our purpose is to detect target in real time accurately.At first utilize rig camera to catch the video at night, and enter our executive software by video frequency collection card.Utilize above-mentioned frame period selecting technology, carry out object detection process than technology such as variations based on local feature treatment technology, the contrast of contrast.In a word, the result after the target detection is when the geometric position information that comprises a plurality of people's profiles.
With regard to the target following process,, determine the position relation of previous frame target and present frame target by forecast model, thereby determine its correspondence through after the above target detection.In addition, we have also utilized the dimension information of target, and the multiframe match information is further filtered because noise and light intensity change the flase drop operation that causes.If target can be detected accurately and trace into, and enter into the forbidden zone that we stipulate, then report to the police.And follow the tracks of this target always, show simultaneously and write down this track, be beneficial to subsequent analysis.
Provide the explanation of each related in this invention technical scheme detailed problem below in detail:
1. the adaptive frame difference is chosen
From the motion of extracting the target at night the background is a critical step for subsequent analysis.At present change detecting method mainly contains three kinds of background subtractions, time difference, light stream and since night scene background complexity, we adopt the time difference algorithm, but with common anchor-frame between the algorithm of number different, we adopt adaptive algorithm decision frame-to-frame differences.Think on directly perceived that the just expression that is more or less the same of two two field pictures does not have too many motion to take place, thus we to portray frame with similarity poor.
MAD (Mean Absolute Difference) and SAD (Sum of AbsoluteDifference) method are used to describe similarity, but this method is responsive for comparisons such as noise, illumination, and NCC (Normal Correlation Coefficient) incites somebody to action robust more comparatively speaking.Making I represent a size is the sequence of the two field picture of NXM, and then similarity NCC can be expressed as
Wherein, I is an average.
Can calculate frame period by following operation
NCC(I
i，I
j)≥T
R?then?j＝j+1?else?i＝j (2)
If the similarity of two frames is less than T
R, second frame continues to choose backward so, up to greater than T
R, operate from the frame period that present frame begins to calculate next time this moment.
2. express based on the target signature at night of contrast
Even the mankind can find target target hovering sometimes easily at night, the ability of people's this resolution object comprises the difference of gray value, difference of color or the like mainly according to the difference of target and background.According to the research of psychophysiology, people's this visual characteristic can be described with contrast.Common contrast has weber contrast and Michelson-contrast, and adopted partial statistics characteristic to define contrast here.Result of calculation is referring to accompanying drawing 2.It is inequality that the zone of target and the contrast major part in aimless zone are arranged as can be seen, and for the stagnant zone of texture-rich, contrast is also enough big, but this part we will adopt another characteristic of people-motion sensitive is removed.
Had as accompanying drawing 2 contrast result of calculations, we think our interesting areas according to the bigger zone of contrast, therefore can the passing threshold processing obtain.
Wherein, T1 is a threshold value.Use example that contrast detects target as shown in Figure 3 for one.
3. based on contrast change information motion detection technique
Contrast information has just provided the target of contrast greater than certain threshold value.In interesting target is also included within, but also have a large amount of uninterested targets, as many texture informations and marginal information, but the target that moves in night monitoring is that we are more interested.We think the mobile certain rules that has of target.Specific to contrast metric is exactly that moving of target must be brought the change of contrast on every side.The example of utilizing contrast to change accurately to monitor moving target as shown in Figure 4.
4. based on the target following technology of predicting
Target following is a problem of judging that on the basis of target detection different period targets are whether corresponding, particularity owing to the scene at night, the present invention only utilizes velocity information to come current location and next moment position of target of prediction, reach the purpose of target following, simultaneously, follow the tracks of the trace information that obtains and also be fed the detection that is used for target, make target detection more accurate.
When following the tracks of, at first former detected target is carried out motion prediction.Because the frame per second of video is 30FPS, therefore can be similar to think each moving target within several frames in front and back at the uniform velocity, the predicted position of target just draws according to the position of target in previous frame and its movement velocity and does not consider the problem of acceleration.Certainly, this is the most basic idea.Because the size of detected same target also is not quite similar in each frame, center position is shake often, the mobile route of target is not a smooth curve, if directly adopt the position of target in the previous frame may produce bigger error, therefore need obtain the prediction reference point with the method for time domain average as prediction reference.Its computing formula is as follows.
Wherein, RefPtk, t are k target at t prediction reference point constantly, and { ObjPtk} is the tracing point sequence of this target, the window size that adopts when N is time domain average.
When calculating the movement velocity of target, relate to the problem of anti-jitter equally.We be not simple from two frames of front and back its translational speed of position calculation of target, the choosing of reference point needs certain time interval (silent know choose 11 frames at interval) here, and reference point also is that the mean value computation by target location in the N continuous frame obtains.The formula that calculates target velocity is as follows.
Wherein, M is a frame number at interval, the window size that N adopts during for time domain average.
According to the hypothesis of uniform motion, the predicted position of target is shown below.
Because need target trace information in the number frame before present frame in the aforementioned calculation process, this just requires at least to have in the target trajectory (M+N) individual point.When not enough this number of the tracing point of target, system directly uses the tracing point of target in previous frame as its predicted position.
After above-mentioned prediction computing finishes, calculate the distance between the predicted position of each foreground object and each target in the present frame one by one.Suppose in the present frame that detected foreground object is S, number of targets is K, then can constitute the K*S distance matrix | Dks, t|, Dks wherein, t be t constantly k target resemble and s foreground object between Euclidean distance, that is:
D
ks，t＝|PredPt
k，t-ForePt
s，t| (8)
Only work as Dks, t is k capable and s row minimum value and Dks simultaneously, and t is not more than predicated error in limited time, thinks that s foreground object is k the continuity of target in present frame.Promptly must satisfy k target simultaneously and be from the nearest target of s foreground object and s foreground object also is to think that just both are same targets during from the nearest foreground object of k target.The predicated error of native system limit acquiescence gets 10.0, and the target location then gets 20.0 during predicted position in as present frame in using previous frame.Judge that the foreground object that is not included into any target after finishing is considered to emerging target, and deposit object queue in.For the target of failing to find the coupling foreground object at present frame, deposit its predicted position the tracing point sequence of target in as the appearance position of this target in present frame, have only and think just that after target is lost the F1 frame continuously target loses.Such setting is that the track that causes of the omission (causing one of them or two target generation omissions thereby tend to be detected as a foreground object when two targets are overlapped) when preventing foreground detection is discontinuous.Threshold value F1 acquiescence value is 30 in the native system.For emerging target, have only and when it occurs the F2 frame continuously, just be considered to effective target, white noise character according to picture noise, the erroneous judgement foreground object that is caused by noise generally can not occur at same position or the continuous several frames of its adjacent locations, therefore adopts this strategy can further restrain noise effect.The acquiescence value of F2 is 6 in the native system.
Introduced after some ins and outs in the embodiment, described test effect of the present invention below in detail:
In order to verify the validity of this method, we have selected real-time monitoring platform as shown in Figure 6.This experiment porch comprises 19 cameras.Scene comprises indoor verandas, hall, typical situation such as outdoor traffic scene.All videos are with the speed acquisition of 30 frame per seconds, and original size is 320x240.We test the data of different periods at night.
Accompanying drawing 7 and accompanying drawing 8 have been showed target detection and the result of tracking at night, and under the enough situation of visibility, the influence of night lights is to make to detect the main former of failure and answer.In the night traffic scene, not only need to detect accurately vehicle, and should avoid because the flase drop that car light causes at the light of road reflection.At accompanying drawing 8, the reflection of window light can produce wrong detection equally shown in accompanying drawing (a).(b) be our testing result, can see to detect target accurately and not disturbed by other factors.
Accompanying drawing 9 is utmost point low light level detection and trace examples according to target under the condition.Accompanying drawing 10 is that our invention is applied to testing result under the dense fog condition.We utilize appraisal procedure that comparison is made in our invention and the method in the technical literature simultaneously, comparative result such as accompanying drawing 11.As can be seen: 1) effective, detect the accuracy rate height.Under the fair situation of light, the detection accuracy rate (greater than 95%) of our invention generally is higher than prior art (less than 80%), can not work for other method under the situations such as faint light photograph, dense fog simultaneously and our invention can detect moving target effectively.2) adaptability: not only effective to evening and bad weather, under normal weather condition, especially the target detection under the sunlight is effective equally, and because considered the partial statistics characteristic, can remove target shadow effectively.3) calculation cost: minimum calculation cost is one of advantage of this invention, and this mainly has benefited from simple more feature selecting.The primitive character that we use is based on block statistics contrast information, for original size is the image of 320x240, if we select the window of 4x4 to calculate, then last calculative picture size only is 80x60, with in the pointwise modeling mentioned compare, consider other calculating, the fast at least 4-6 of speed doubly.Our algorithm is all having certain superiority aspect feature selecting, calculation cost, accuracy in detection and the adaptability.
In a word, based on human vision property, the present invention proposes target detection and track algorithm at a kind of simple and effective night.Improved interFrameGap selecting technology is used to choose two enough not big frames of postman; Then, realize the extraction of target signature at night, consider temporal information simultaneously, further determine moving target with the contrast change information by contrast information; At last, tracing process adopts speed prediction model, and the multiframe coupling of tracking target is used to the detection of conclusive judgement target.Result of the test on real monitoring experiment database has been verified the validity of our algorithms.
Claims (5)
1. target detection and tracking at night based on visual characteristic comprises step:
A. catch the video at night;
B. adopt adaptive algorithm decision frame period, choose the frame enough big with the present frame otherness; This adaptive algorithm is judged the interval of selecting automatically by the similarity of calculating between two frames, if the correlation between two frames is less than threshold value, show that so two frame differences are little, then continue to compare the correlation of next frame and present frame, up to the frame period that finds between the two enough big frames of otherness;
C. determine the target area according to contrast in two frames that the otherness that finds is enough big in step b;
D. further determine moving target based on the contrast change information of two enough big frames of the otherness that in step b, finds;
E. according to the current location of velocity information target of prediction and next position constantly: consider uniform motion position calculation model, obtain the prediction reference point by detected target trajectory point sequence before; Calculate the movement velocity of target again by before detected target location; Obtain the predicted position of target then by prediction reference point and movement velocity.
2. method according to claim 1 is characterized in that adopting video camera to catch the video at night.
3. method according to claim 2 is characterized in that described video camera is at least one.
4. method according to claim 1 is characterized in that described contrast is defined by partial statistics characteristic.
5. method according to claim 1 is characterized in that obtaining the prediction reference point with the time domain averaging method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005101306960A CN100546380C (en) | 2005-12-21 | 2005-12-21 | Target detection and tracking at night based on visual characteristic |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005101306960A CN100546380C (en) | 2005-12-21 | 2005-12-21 | Target detection and tracking at night based on visual characteristic |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1988653A CN1988653A (en) | 2007-06-27 |
| CN100546380C true CN100546380C (en) | 2009-09-30 |
Family
ID=38185255
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005101306960A Expired - Fee Related CN100546380C (en) | 2005-12-21 | 2005-12-21 | Target detection and tracking at night based on visual characteristic |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100546380C (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101409825B (en) * | 2007-10-10 | 2011-04-13 | 中国科学院自动化研究所 | Nighttime vision monitoring method based on information fusion |
| CN101827204B (en) * | 2010-04-19 | 2013-07-17 | 成都索贝数码科技股份有限公司 | Method and system for detecting moving object |
| CN104660954B (en) * | 2013-11-18 | 2019-10-25 | 深圳力维智联技术有限公司 | Brightness of image method for improving and device under low-illumination scene based on background modeling |
| CN104079881B (en) * | 2014-07-01 | 2017-09-12 | 中磊电子（苏州）有限公司 | The relative monitoring method of supervising device |
| CN108960190B (en) * | 2018-07-23 | 2021-11-30 | 西安电子科技大学 | SAR video target detection method based on FCN image sequence model |
| CN111476065A (en) * | 2019-01-23 | 2020-07-31 | 北京奇虎科技有限公司 | Target tracking method and device, computer equipment and storage medium |
| CN110363197B (en) * | 2019-06-22 | 2023-05-23 | 东北电力大学 | Video region of interest extraction method based on improved visual background extraction model |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5999877A (en) * | 1996-05-15 | 1999-12-07 | Hitachi, Ltd. | Traffic flow monitor apparatus |
| CN1404695A (en) * | 2000-12-06 | 2003-03-19 | 皇家菲利浦电子有限公司 | Method and apparatus to select the best video frame to transmit to a remote station for closed circuit television (CCTV)based residential area security monitoring |
-
2005
- 2005-12-21 CN CNB2005101306960A patent/CN100546380C/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5999877A (en) * | 1996-05-15 | 1999-12-07 | Hitachi, Ltd. | Traffic flow monitor apparatus |
| CN1404695A (en) * | 2000-12-06 | 2003-03-19 | 皇家菲利浦电子有限公司 | Method and apparatus to select the best video frame to transmit to a remote station for closed circuit television (CCTV)based residential area security monitoring |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1988653A (en) | 2007-06-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106652465B (en) | Method and system for identifying abnormal driving behaviors on road | |
| Javed et al. | Tracking and object classification for automated surveillance | |
| CN100546380C (en) | Target detection and tracking at night based on visual characteristic | |
| CN100545867C (en) | Aerial shooting traffic video frequency vehicle rapid checking method | |
| CN105744232A (en) | Method for preventing power transmission line from being externally broken through video based on behaviour analysis technology | |
| CN101295405A (en) | Portrait and vehicle recognition alarming and tracing method | |
| Barmpoutis et al. | Smoke detection using spatio-temporal analysis, motion modeling and dynamic texture recognition | |
| Tian et al. | Smoke detection in videos using non-redundant local binary pattern-based features | |
| CN105678803A (en) | Video monitoring target detection method based on W4 algorithm and frame difference | |
| CN102968625A (en) | Ship distinguishing and tracking method based on trail | |
| CN105046719B (en) | A kind of video frequency monitoring method and system | |
| Malhi et al. | Vision based intelligent traffic management system | |
| CN102610104B (en) | Onboard front vehicle detection method | |
| Kawakatsu et al. | Traffic surveillance system for bridge vibration analysis | |
| Xia et al. | Automatic multi-vehicle tracking using video cameras: An improved CAMShift approach | |
| Monteiro et al. | Wrongway drivers detection based on optical flow | |
| Lee | An accident detection system on highway using vehicle tracking trace | |
| Santos et al. | Car recognition based on back lights and rear view features | |
| CN1426020A (en) | Far distance identity identifying method based on walk | |
| Shammi et al. | An automated way of vehicle theft detection in parking facilities by identifying moving vehicles in CCTV video stream | |
| Hai-Feng et al. | Vehicle abnormal behavior detection system based on video | |
| CN112115767B (en) | Tunnel foreign matter detection method based on Retinex and YOLOv3 models | |
| Lagorio et al. | Automatic detection of adverse weather conditions in traffic scenes | |
| CN106485713B (en) | Video foreground detection method | |
| Joshi | Vehicle speed determination using image processing |
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 | ||
| ASS | Succession or assignment of patent right |
Owner name: ZHONGKE WISH TECHNOLOGY (BEIJING) CO., LTD. Free format text: FORMER OWNER: INSTITUTE OF AUTOMATION, CHINESE ACADEMY OF SCIENCES Effective date: 20121206 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| COR | Change of bibliographic data |
Free format text: CORRECT: ADDRESS; FROM: 100080 HAIDIAN, BEIJING TO: 100081 HAIDIAN, BEIJING |
|
| TR01 | Transfer of patent right |
Effective date of registration: 20121206 Address after: 100081 Beijing City, Haidian District, No. 14 Building No. 1 hospital 206 Patentee after: In the Department of science and Technology (Beijing) Co., Ltd. realism Address before: 100080 Zhongguancun East Road, Beijing, No. 95, No. Patentee before: Institute of Automation, Chinese Academy of Sciences |
|
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20090930 Termination date: 20141221 |
|
| EXPY | Termination of patent right or utility model |