CN101278839A - Method for tracking nighttime drive - Google Patents
Method for tracking nighttime drive Download PDFInfo
- Publication number
- CN101278839A CN101278839A CNA2008100478060A CN200810047806A CN101278839A CN 101278839 A CN101278839 A CN 101278839A CN A2008100478060 A CNA2008100478060 A CN A2008100478060A CN 200810047806 A CN200810047806 A CN 200810047806A CN 101278839 A CN101278839 A CN 101278839A
- Authority
- CN
- China
- Prior art keywords
- human eye
- image
- search
- reflective spot
- tracking
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention pertains to the technical field of automobile safety, which more particularly relates to a tracking method in night driving. The invention is characterized in that: firstly, a reflective spot of a human eye is searched as the drive tracking base and is carried out image sampling and image processing; the reflective spot of the human eye is searched by image identification and a binaryzation threshold value is stored as the initial binaryzation threshold value when the reflective spot of the human eye is searched, so as to ensure that the reflective spot of the human eye is the initial position of the human eye; then the drive tracking judge is carried out, that is to say, consequent original images are extracted from video and a face regional image is divided in the consequent original images according to the initial position of the human eye and is regarded as the image to be identified; the tracking of the reflective spot of the human eye is continued to judge the driving situation according to the search result. The method has the advantages of accuracy, real time performance and adaptability and can make judgment in danger time and help the driver, which is an ideal assistance for safety driving.
Description
Technical field
The invention belongs to the automotive safety technical field, particularly a kind of method for tracking nighttime drive.
Background technology
Ride safety of automobile is the problem that the driver is concerned about.China has the automobile in the whole world 1.9%, the traffic fatalities that cause have but accounted for 15% of the whole world, become one of the most multiple country of vehicle accident, causing vehicle accident a big chunk reason is relatively poor, thin the causing of driver safety consciousness because the security performance of vehicle compares, because nighttime driving is sleepy easily, the accident that causes is especially many.Pay close attention to automotive safety, be not only for driver and passenger, and for other people on the road.Realize that by video monitor driving tracking technique is an important research direction of field of automobile safety, but because the supervision of prior art judges that sensitivity is not high, the view data treating capacity is excessive, therefore practical function is not good, can't in time make reflection, also need to overcome these problems apart from practical application to driving dangerous situation.
Summary of the invention
The object of the invention is to provide a kind of method for tracking nighttime drive that can real-time tracking monitors the driving situation.
Technical scheme of the present invention is: at first, searches out the human eye reflective spot and follows the tracks of the basis, may further comprise the steps as driving,
Step 1, image sampling, the video and the preservation of promptly taking driver's seat in the car;
Step 2, Flame Image Process is promptly taken out frame and is obtained initial original image from video;
Step 3, image recognition, image to be identified adopts initial original image, concrete mode is to enumerate the mode conversion binary-state threshold, variation circulation with binary-state threshold is carried out following three step 31～33 up to searching the human eye reflective spot, and preserve binary-state threshold when searching the human eye reflective spot as initial binary-state threshold, guarantee that human eye reflective spot position is as initial position of human eye
Step 31 is treated the recognition image binaryzation, obtains bianry image;
Step 32 is carried out expansion process to bianry image, obtains the expansion plans picture;
Step 33, search human eye reflective spot in the expansion plans picture, described way of search is to search for facial positions earlier, searches for eye locations again in facial positions, search human eye reflective spot in eye locations;
Then, drive to follow the tracks of and judge, promptly from video, take out frame and obtain follow-up original image, in follow-up original image, be partitioned into the facial zone image as image to be identified according to initial position of human eye, with initial binary-state threshold execution in step 31～33 search human eye reflective spots, judge the driving situation according to Search Results.
And, infrared light sources is provided in car, take and adopt the infrared pick-up head.
And, to lay on the ride before windshield after the steering wheel and take the photographic head that adopts, the camera lens of photographic head is at the parallel position of crossing the steering wheel centrage perpendicular to ground, and in the face of driver's seat.
And the described mode of enumerating is for adjusting binary-state threshold from small to large.
And described search facial positions implementation is, seeks that white portion in the expansion plans picture is gone up most and the most left coordinate figure and preserving, and from left to right cuts apart by setting facial scope size from top to bottom, and the zone that is partitioned into is a facial positions; The implementation of described search eye locations is that the gray scale maximum region in the searching facial positions is eye locations; Described implementation of searching for the human eye reflective spot in eye locations is that the rectangle luminous point in the searching eye locations is the human eye reflective spot, and the pixel coordinate in the rectangle upper left corner is decided to be the centre coordinate of human eye reflective spot.
And described concrete mode according to Search Results judgement driving situation is as follows,
When in the facial zone image, searching the human eye reflective spot, be judged as normal driving, from video, extract next follow-up original image and continue to follow the tracks of;
When search is less than the human eye reflective spot in the facial zone image, carrying out facial zone deeply searches for, promptly adjust binary-state threshold and search for the human eye reflective spots with threshold transformation circulation execution in step 31～33 to enumerate mode, if search the human eye reflective spot then be judged as improper driving, on the Search Results basis, extract next follow-up original image and proceed to drive to follow the tracks of and judge; Adjusted binary-state threshold and still search for less than the human eye reflective spot then be judged as dangerous driving if enumerate.
And, set the head position higher limit, when being judged as dangerous driving, as picture search facial positions to be identified, the coordinate figure of white portion pixel peak is judged as the driver and loses consciousness during greater than the head position higher limit with the facial zone image.
And, set time delay, when search is less than the human eye reflective spot in the facial zone image, extracts next follow-up original image and be partitioned into facial zone image continuation search, when still not searching the human eye reflective spot, begin to carry out facial zone and deeply search for above time delay.
The present invention creatively proposes to utilize the human eye reflective spot to carry out driving condition judgement, accuracy rate height; And judge that data processing amount is little based on binary image, emphasis monitors that the facial zone image has further dwindled data processing task in tracking, therefore is swift in response, and satisfies to drive in real time and follows the tracks of needs.Using the present invention can carry out explication de texte to various driving situations, is convenient at processing, neither influences normal driving, can avoid dangerous again, satisfies accuracy, real-time and adaptability requirement, has great promotional value.
Description of drawings
The principle schematic of Fig. 1 embodiment of the invention.
The particular flow sheet of Fig. 2 embodiment of the invention.
The specific embodiment
Method for tracking nighttime drive of the present invention was divided into for two steps, at first carried out initial ranging, carried out then searching in the small area on the basis of initial search result, and efficient is high to be judged soon.Technical scheme provided by the invention is as follows:
At first, search out the human eye reflective spot and follow the tracks of the basis, may further comprise the steps as driving,
Step 1, image sampling, the video and the preservation of promptly taking driver's seat in the car.The most important condition of wanting images acquired is to have light source in the car, but considers according to visual theory, and only light is more outdoor when weak than car in car, and the driver could see through windshield observation road conditions.Therefore in car, provide light source can not use intensive visible light.What the dark more driver's pupil of light was put in addition is also just big more.It is big more that pupil is put, and it is also just clear more to observe things.So, the invention provides further technical scheme: infrared light sources is provided in car, takes and adopt the infrared pick-up head in order to guarantee traffic safety.Preferred implementation is, lays on the ride before windshield after the steering wheel and takes the photographic head that adopts, and the camera lens of photographic head is at the parallel position of crossing the steering wheel centrage perpendicular to ground, and in the face of driver's seat.
Step 2, Flame Image Process is promptly taken out frame and is obtained initial original image from video.In order to improve data-handling efficiency, can carry out format conversion to the view data of getting off and handle from camera collection, be converted into gray level image such as the colored original image that colour imagery shot is taken, can vast scale ground amount of compressed data.
Step 3, image recognition, image to be identified adopts initial original image, concrete mode is to enumerate the mode conversion binary-state threshold, variation circulation with binary-state threshold is carried out following three step 31～33 up to searching the human eye reflective spot, and preserve binary-state threshold when searching the human eye reflective spot as initial binary-state threshold, guarantee that human eye reflective spot position is as initial position of human eye.Enumerate in an orderly manner and can guarantee search efficiency, can adjust binary-state threshold from small to large.
Step 31 is treated the recognition image binaryzation, obtains bianry image;
Step 32 is carried out expansion process to bianry image, obtains the expansion plans picture;
Step 33, search human eye reflective spot in the expansion plans picture, described way of search is to search for facial positions earlier, searches for eye locations again in facial positions, search human eye reflective spot in eye locations;
Utilize the human eye reflective spot to carry out driving condition and judge it is important improvement point of the present invention, because no matter be at which type of photoenvironment, the reflective characteristic of eyes is impregnable.In image, the human eye reflective spot shows as 1 pixel luminous point, and after bianry image was carried out expansion process, the human eye reflective spot can show as the little rectangle luminous point of 2 * 2 pixels.The present invention utilizes the distinct performance of reflective characteristic in image, and the accuracy rate of carrying out image recognition is very high.The present invention also provides further technical scheme, realize the progressively search of face-eye-reflective spot, it is higher to carry out the point by point scanning judging efficiency compared with the view picture picture: described search facial positions implementation is, seeking the interior white portion of expansion plans picture goes up and the most left coordinate figure and preservation most, from left to right cut apart by setting facial scope size (can set as the case may be during enforcement) from top to bottom, the zone that is partitioned into is a facial positions; The implementation of described search eye locations is that the gray scale maximum region in the searching facial positions is eye locations; Described implementation of searching for the human eye reflective spot in eye locations is that the rectangle luminous point in the searching eye locations is the human eye reflective spot, and the pixel coordinate in the rectangle upper left corner is decided to be the centre coordinate of human eye reflective spot.Can adopt conventional images process chip to realize search identification during concrete enforcement based on MATLAB.
Then, drive to follow the tracks of and judge, promptly from video, take out frame and obtain follow-up original image, in follow-up original image, be partitioned into the facial zone image as image to be identified according to initial position of human eye, with initial binary-state threshold execution in step 31～33 search human eye reflective spots, judge the driving situation according to Search Results.
If because driver's normal driving, eye locations does not have significantly change, be partitioned into the facial zone image as image to be identified from follow-up original image, facial zone is continued Tracking Recognition, obviously the entire image efficient that photographs than monitoring camera always is higher.So-called facial zone is not to refer to complete face, can only choose the zone bigger slightly than eye locations during enforcement, determines that according to initial position of human eye it is rational selection that central point is cut apart.The original image that the embodiment of the invention is obtained from camera collection is the gray scale of 320 * 240 pixels, and image data amount is about 10kb; Through being the bianry image of 81 * 111 pixels after the binaryzation, image data amount is about 450 bytes; Behind the facial zone image binaryzation that is partitioned into, supply to differentiate search image be 41 * 111 pixels, data volume is 280 bytes.As seen the present invention has carried out effective control to the data treating capacity, and the image of fast processing photographic head shooting has good real time performance continuously.
As image to be identified,, not only can judge the driving situation according to Search Results with the facial zone image, can also obtain more accurate driving behavior by analysis according to described with initial binary-state threshold execution in step 31～33 search human eye reflective spots:
When in the facial zone image, searching the human eye reflective spot, be judged as the driving of normal condition, from video, extract next follow-up original image and continue to follow the tracks of;
When search is less than the human eye reflective spot in the facial zone image, carry out facial zone and deeply search for, promptly adjust binary-state threshold and search for the human eye reflective spots with threshold transformation circulation execution in step 31～33 to enumerate mode.If search the human eye reflective spot then be judged as improper driving, the expression driver has changed the sitting posture when driving, perhaps the extraneous light environment changes, on the Search Results basis, extract next follow-up original image and proceed to drive to follow the tracks of and judge and get final product, so the adaptability of technical solution of the present invention is also very good.If enumerate and adjusted binary-state threshold and still search for less than the human eye reflective spot then be judged as dangerous driving, may be that driver's eyes is closed or towards other places, dozed off in other words or the sight line run-off-road.
In driving safety, the situation that the driver bows needs vigilant especially, the invention provides further technical scheme: set the head position higher limit, when being judged as dangerous driving, with the facial zone image as picture search facial positions to be identified, the coordinate figure of white portion pixel peak is judged as the driver and loses consciousness during greater than the head position higher limit.
When considering normal driving, the driver can not stare at the road surface always.Look at that of short duration movement such as rearview mirror should drive to allow nictation once in a while, therefore there is no need this situation is made a response.The present invention proposes: set time delay, when search is less than the human eye reflective spot in the facial zone image, extract next follow-up original image and be partitioned into facial zone image continuation search, when still not searching the human eye reflective spot, just begin to carry out facial zone and deeply search for above time delay.Rationally set time delay length when needing only concrete enforcement, just can under the situation of processing resource that practices every conceivable frugality, neither influence normal driving, can avoid dangerous again.
Referring to Fig. 1, the invention provides the embodiment ultimate principle: adopt CMOS photographic head capture video, carry out video then and take out Flame Image Process such as frame, original image is preserved.Because video is made up of consecutive image, for the ease of real-time tracking, original image can be numbered in chronological order and deposit.Employing is carried out video identification based on the picture processing chip of MATLAB on the original image basis, preserve the image classification of various situations standby at last: the image of the image under the normal condition, doze or sight line run-off-road, the image when losing consciousness.Also can carry out other subsequent treatment during concrete the application according to tracking results, for example sounding prompting, driving adapter etc., the present invention will not give unnecessary details.Referring to Fig. 2, the present invention also provides the idiographic flow of embodiment, so that implement reference: call original image, binary-state threshold is made as 25, and (experiment finds to adopt 25 can search out facial positions immediately usually, be preferred empirical value), carry out image binaryzation, promote threshold size then gradually; If threshold value can directly be called next Zhang Houxu original image greater than 45 o'clock, to raise the efficiency, if between 25～45, found the human eye reflective spot, preserve current threshold value and center point coordinate, begin then to follow the tracks of and judge: the image after calling, call threshold value and center point coordinate, cut apart bianry image, judge whether to occur the eye reflective spot, if not do not occur then readjust threshold value and carry out the facial zone image and deeply search for.If do not occur the eye reflective spot yet, judge then whether head is sagging, so that the most dangerous driving situation has appearred in prompting.
Claims (8)
1. method for tracking nighttime drive is characterized in that:
At first, search out the human eye reflective spot and follow the tracks of the basis, may further comprise the steps as driving,
Step 1, image sampling, the video and the preservation of promptly taking driver's seat in the car;
Step 2, Flame Image Process is promptly taken out frame and is obtained initial original image from video;
Step 3, image recognition, image to be identified adopts initial original image, concrete mode is to enumerate the mode conversion binary-state threshold, variation circulation with binary-state threshold is carried out following three step 31～33 up to searching the human eye reflective spot, and preserve binary-state threshold when searching the human eye reflective spot as initial binary-state threshold, guarantee that human eye reflective spot position is as initial position of human eye
Step 31 is treated the recognition image binaryzation, obtains bianry image;
Step 32 is carried out expansion process to bianry image, obtains the expansion plans picture;
Step 33, search human eye reflective spot in the expansion plans picture, described way of search is to search for facial positions earlier, searches for eye locations again in facial positions, search human eye reflective spot in eye locations;
Then, drive to follow the tracks of and judge, promptly from video, take out frame and obtain follow-up original image, in follow-up original image, be partitioned into the facial zone image as image to be identified according to initial position of human eye, with initial binary-state threshold execution in step 31～33 search human eye reflective spots, judge the driving situation according to Search Results.
2. method for tracking nighttime drive according to claim 1 is characterized in that: infrared light sources is provided in car, takes and adopt the infrared pick-up head.
3. method for tracking nighttime drive according to claim 1, it is characterized in that: lay on the ride before windshield after the steering wheel and take the photographic head that adopts, the camera lens of photographic head is at the parallel position of crossing the steering wheel centrage perpendicular to ground, and in the face of driver's seat.
4. method for tracking nighttime drive according to claim 1 is characterized in that: the described mode of enumerating is for adjusting binary-state threshold from small to large.
5. method for tracking nighttime drive according to claim 1, it is characterized in that: described search facial positions implementation is, seeking the interior white portion of expansion plans picture goes up and the most left coordinate figure and preservation most, from left to right cut apart by setting facial scope size from top to bottom, the zone that is partitioned into is a facial positions; The implementation of described search eye locations is that the gray scale maximum region in the searching facial positions is eye locations; Described implementation of searching for the human eye reflective spot in eye locations is that the rectangle luminous point in the searching eye locations is the human eye reflective spot, and the pixel coordinate in the rectangle upper left corner is decided to be the centre coordinate of human eye reflective spot.
6. according to claim 1 or 2 or 3 or 4 or 5 described method for tracking nighttime drive, it is characterized in that:
Described concrete mode according to Search Results judgement driving situation is as follows,
When in the facial zone image, searching the human eye reflective spot, be judged as normal driving, from video, extract next follow-up original image and continue to follow the tracks of;
When search is less than the human eye reflective spot in the facial zone image, carrying out facial zone deeply searches for, promptly adjust binary-state threshold and search for the human eye reflective spots with threshold transformation circulation execution in step 31～33 to enumerate mode, if search the human eye reflective spot then be judged as improper driving, on the Search Results basis, extract next follow-up original image and proceed to drive to follow the tracks of and judge; Adjusted binary-state threshold and still search for less than the human eye reflective spot then be judged as dangerous driving if enumerate.
7. method for tracking nighttime drive according to claim 6, it is characterized in that: set the head position higher limit, when being judged as dangerous driving, with the facial zone image as picture search facial positions to be identified, the coordinate figure of white portion pixel peak is judged as the driver and loses consciousness during greater than the head position higher limit.
8. according to claim 6 or 7 described method for tracking nighttime drive, it is characterized in that: set time delay, when search is less than the human eye reflective spot in the facial zone image, extract next follow-up original image and be partitioned into facial zone image continuation search, when still not searching the human eye reflective spot, begin to carry out facial zone and deeply search for above time delay.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008100478060A CN101278839B (en) | 2008-05-22 | 2008-05-22 | Method for tracking nighttime drive |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008100478060A CN101278839B (en) | 2008-05-22 | 2008-05-22 | Method for tracking nighttime drive |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101278839A true CN101278839A (en) | 2008-10-08 |
| CN101278839B CN101278839B (en) | 2010-10-13 |
Family
ID=40011642
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2008100478060A Expired - Fee Related CN101278839B (en) | 2008-05-22 | 2008-05-22 | Method for tracking nighttime drive |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101278839B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103116890A (en) * | 2013-02-27 | 2013-05-22 | 中山大学 | Video image based intelligent searching and matching method |
| CN108392180A (en) * | 2017-02-07 | 2018-08-14 | 株式会社岛津制作所 | Time-activity curve measurement device |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1131699C (en) * | 1999-06-09 | 2003-12-24 | 现代自动车株式会社 | Method for detecting driver's eyes in dozing driving alarm system |
| US7206435B2 (en) * | 2002-03-26 | 2007-04-17 | Honda Giken Kogyo Kabushiki Kaisha | Real-time eye detection and tracking under various light conditions |
| CN1830389A (en) * | 2006-04-21 | 2006-09-13 | 太原理工大学 | Device for monitoring fatigue driving state and its method |
| CN101090482B (en) * | 2006-06-13 | 2010-09-08 | 唐琎 | Driver fatigue monitoring system and method based on image process and information mixing technology |
| CN101030316B (en) * | 2007-04-17 | 2010-04-21 | 北京中星微电子有限公司 | Safety driving monitoring system and method for vehicle |
-
2008
- 2008-05-22 CN CN2008100478060A patent/CN101278839B/en not_active Expired - Fee Related
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103116890A (en) * | 2013-02-27 | 2013-05-22 | 中山大学 | Video image based intelligent searching and matching method |
| CN103116890B (en) * | 2013-02-27 | 2015-11-18 | 中山大学 | A kind of intelligent search matching process based on video image |
| CN108392180A (en) * | 2017-02-07 | 2018-08-14 | 株式会社岛津制作所 | Time-activity curve measurement device |
| CN108392180B (en) * | 2017-02-07 | 2021-07-30 | 株式会社岛津制作所 | Time intensity curve measuring device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101278839B (en) | 2010-10-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210004619A1 (en) | Driving state analysis method and apparatus, driver monitoring system and vehicle | |
| CN101950355B (en) | Method for detecting fatigue state of driver based on digital video | |
| US20060204042A1 (en) | System and method for determining eye closure state | |
| US6304187B1 (en) | Method and device for detecting drowsiness and preventing a driver of a motor vehicle from falling asleep | |
| CN107323338B (en) | Vehicle steering lamp control system, control method and vehicle | |
| CN111079475A (en) | Driving state detection method and device, driver monitoring system and vehicle | |
| WO2020029444A1 (en) | Method and system for detecting attention of driver while driving | |
| US7746235B2 (en) | System and method of detecting eye closure based on line angles | |
| US20180232587A1 (en) | Vehicle monitoring method and apparatus, processor, and image acquisition device | |
| CN111062292B (en) | Fatigue driving detection device and method | |
| Li et al. | Yawning detection for monitoring driver fatigue based on two cameras | |
| US9418287B2 (en) | Object detection apparatus | |
| CN101593425A (en) | A kind of fatigue driving monitoring method and system based on machine vision | |
| CN105739705A (en) | Human-eye control method and apparatus for vehicle-mounted system | |
| CN102306293A (en) | Method for judging driver exam in actual road based on facial image identification technology | |
| CN111032413A (en) | Method for operating a screen of a motor vehicle and motor vehicle | |
| Lin et al. | Adaptive IPM-based lane filtering for night forward vehicle detection | |
| CN106627529A (en) | Intelligent monitoring method and device for vehicle | |
| CN109466286A (en) | A kind of intelligent anti-glare method and device in automobile disengaging low-light (level) region | |
| CN105835761A (en) | Method and device for controlling on and off of atmosphere lamps in car | |
| KR20130106640A (en) | Apparatus for trace of wanted criminal and missing person using image recognition and method thereof | |
| CN101278839B (en) | Method for tracking nighttime drive | |
| CN114954307A (en) | Driving assistance system based on artificial intelligence | |
| CN107640160B (en) | Vehicle safe driving auxiliary system and control method thereof | |
| CN107480629A (en) | A kind of method for detecting fatigue driving and device based on depth information |
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 | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20101013 Termination date: 20110522 |