CN104866838B - A kind of front vehicles automatic testing method of view-based access control model - Google Patents
A kind of front vehicles automatic testing method of view-based access control model Download PDFInfo
- Publication number
- CN104866838B CN104866838B CN201510296813.4A CN201510296813A CN104866838B CN 104866838 B CN104866838 B CN 104866838B CN 201510296813 A CN201510296813 A CN 201510296813A CN 104866838 B CN104866838 B CN 104866838B
- Authority
- CN
- China
- Prior art keywords
- vehicle
- rectangle
- detection
- detected
- frame region
- 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
- 238000012360 testing method Methods 0.000 title claims abstract description 18
- 238000001514 detection method Methods 0.000 claims abstract description 86
- 238000000034 method Methods 0.000 claims abstract description 24
- 230000008569 process Effects 0.000 claims abstract description 22
- 238000012545 processing Methods 0.000 claims abstract description 13
- 238000009434 installation Methods 0.000 claims abstract description 10
- 230000011218 segmentation Effects 0.000 claims description 27
- 238000005070 sampling Methods 0.000 claims description 10
- 230000008859 change Effects 0.000 claims description 2
- 238000000605 extraction Methods 0.000 claims description 2
- 238000004321 preservation Methods 0.000 claims description 2
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 claims 1
- 238000003709 image segmentation Methods 0.000 claims 1
- 238000004364 calculation method Methods 0.000 abstract description 5
- 230000008901 benefit Effects 0.000 abstract description 4
- 238000007689 inspection Methods 0.000 description 4
- 230000007812 deficiency Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000005086 pumping Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of front vehicles automatic testing methods of view-based access control model, include the following steps：Image sequence is obtained by vehicle-mounted video camera；Before detection process, to image to be detected draw the proving operation of calibration line, and manual setting detection zone；It is detected using underbody shadow character, vehicle license plate characteristic or tail-light feature, and has been examined vehicle judgement；During being detected processing by frame successively, to not captured into between-line spacing with the vehicle lost.So that the installation site of video camera is no longer restricted by proving operation, may be mounted at any position of vehicle front, it is easy to operate, applied widely, convenient for flexibly installation and use replacement；Reduce the unnecessary follow-up judgement in part by the judgement of same vehicle, save operation time, effectively prevent flase drop, has the advantages that accuracy of detection is high, detection accuracy is strong, calculation amount is smaller, processing cost is low etc..
Description
Technical field
The present invention relates to a kind of detection methods, are examined automatically more particularly to a kind of front vehicles of the view-based access control model of vehicular
Survey method, belongs to technical field of intelligent traffic.
Background technology
Currently, the method for detection front vehicles is usually to set fixed vehicle detection region and detection threshold value, to vehicle
And the not enough discrimination of pseudo- vehicle, and be for camera it is fixed in the case of handled, not only limit camera shooting
Installation site and the installation site fixation of machine are inconvenient to change；Detection zone is divided according to the lane line detected in image simultaneously
Domain and detection threshold value, calculation amount are larger.
And existing processing method is usually detected to each frame using various features when detecting vehicle,
Therefore processing speed, detection accuracy, image procossing cost can not be taken into account.
Invention content
It is a primary object of the present invention to overcome deficiency in the prior art, provide a kind of front vehicles of view-based access control model
Automatic testing method, the bus especially suitable for special bus zone traveling carry out the monitoring of front vehicles.
Technical problem to be solved by the invention is to provide easy to operate, response quickly, processing efficiently, result is reliable, suitable
With the front vehicles automatic testing method of the wide view-based access control model of range, specific camera installation locations are not needed not only, are convenient for
It flexibly installs and using replacement；And it is high with accuracy of detection, that detection accuracy is strong, calculation amount is smaller, processing cost is low etc. is excellent
Point, the great utility value having in industry.
In order to achieve the above object, the technical solution adopted in the present invention is：
A kind of front vehicles automatic testing method of view-based access control model, includes the following steps：
1) continuous image sequence is obtained by vehicle-mounted video camera, and is transmitted to computer and is handled；
2) before detection process starts, according to the installation site of video camera, drafting calibration line is carried out to image to be detected
Proving operation, and according to front lane line and calibration line manual setting detection zone；Detection zone size is obtained by calibration line
With the quantitative relationship of actual size, to obtain the actual size of doubtful vehicle；
3) computer is detected processing by frame successively, is detected using underbody shadow character, primarily determines detection zone
Domain has, enters step 4) there may be the rectangle frame region R1 of front vehicles；Without the detection process for then entering next frame image；
4) judge in rectangle frame region R1 whether to be to have examined vehicle, be to track success, add one with vehicle frame number, preserve rectangle
For frame region R1 as the position for having examined vehicle, the gray scale map combining for preserving present frame contains the gray-scale map for having examined vehicle, frame inspection
Survey terminates, into the detection process of next frame image；Otherwise it enters step 5)；
5) judge in rectangle frame region R1 whether to be red doubtful vehicle, be detected using vehicle license plate characteristic, it is no
Then it is detected using tail-light feature；
Using vehicle license plate characteristic or tail-light feature is utilized, detects and successfully then finds new car, preserve rectangle frame region R1 conducts
The position for having examined vehicle preserves the gray-scale map of present frame as containing the gray-scale map for having examined vehicle, and frame detection terminates, under
The detection process of one frame image；Using vehicle license plate characteristic or tail-light feature is utilized, unsuccessfully frame detection also terminates for detection, under
The detection process of one frame image；
6) during being detected processing by frame successively, it is continuous set frame number and do not find examined vehicle and be judged as with losing, to
Inspection vehicle is initialized, and the rectangle frame region R1 that vehicle location has been examined containing the gray-scale map and expression of having examined vehicle is reset；
To not with losing the case where, with vehicle frame number it is shared set frame number and find examined vehicle and then judge that the vehicle as front vehicles, carries out it
It captures.
The present invention is further arranged to：The calibration line includes upper calibration line and lower calibration line totally 2, it is to be checked to be distributed in
In altimetric image；The detection zone be according to image to be detected select where detection vehicle dimension range by front lane line and
The trapezoid area that calibration line is constituted does not make detection process for vehicle existing other than detection zone.
The present invention is further arranged to：It is detected using underbody shadow character in the step 3), specifically, to current
The full figure of frame image carries out binarization segmentation, and full figure binarization segmentation threshold value and full figure binary map are obtained using Otsu algorithm；It will
Gray value less than full figure binarization segmentation threshold value point labeled as white point as foreground, remaining be labeled as black as background, and
Profile is extracted to full figure binary map, seeks the minimum enclosed rectangle R2 of profile, internal is shadow region to be determined.
Judge whether minimum enclosed rectangle R2 is underbody shadow region, is, is then obtained on minimum enclosed rectangle R2
Rectangle frame region R1 that is square and connecting therewith has vehicle region, the size of rectangle frame region R1 to meet W (R1)=W for be determined
(R2), H (R1)=W (R2) * 3/4, wherein W (R1) and H (R1), W (R2) indicate respectively rectangle frame region R1 width and height,
The width of minimum enclosed rectangle R2；Otherwise frame detection terminates, into the detection process of next frame image.
The present invention is further arranged to：The full figure to current frame image carries out binarization segmentation, specifically, first to complete
Figure is sampled, and the pixel obtained by sampling represents the overall situation, and Otsu algorithm is recycled to seek full figure binarization segmentation threshold value.
The present invention is further arranged to：It is described to judge whether minimum enclosed rectangle R2 is that underbody shadow region is specifically sentenced
The central point of disconnected profile makees ratio whether in the detection zone of selection, by the foreground of shade in minimum enclosed rectangle R2 and its area
Shade ratio is obtained, is calculated using the quantitative relationship that calibration line obtains and obtains the corresponding actual sizes of minimum enclosed rectangle R2, with
And the positions minimum enclosed rectangle R2 and the ratio of width to height；If meeting fractional threshold, size threshold, locality condition and the ratio of width to height item
Part then judges minimum enclosed rectangle R2 for underbody shadow region.
The present invention is further arranged to：Judge in rectangle frame region R1 whether to be to have examined vehicle in the step 4), specifically
It is by rectangle frame region R1 and to have examined vehicle and carry out grey level histogram and compare to obtain gray scale ratio, while having compared rectangle frame region R1
Scale error is obtained with the positions and dimensions for having examined vehicle, if meeting gray threshold, locality condition and size relative error threshold value,
Then judge in rectangle frame region R1 to have examined vehicle.
The present invention is further arranged to：Judge in the step 5) in rectangle frame region R1 whether be red doubtful vehicle
, specifically, new car detection is carried out to tracking failed vehicle, rectangle frame region R1 is converted to HSV space, tone is utilized
H, the triple channel Threshold segmentation of saturation degree S and brightness V goes out red area, and using red area labeled as white point as foreground, its
Remaining label is as background；Judge rectangle frame region according to whether the sum of foreground pixel point meets red vehicle threshold value
Whether it is red in the color region R4 to be determined of R1, is to judge that rectangle frame region R1 is interior for red vehicle, into car plate spy
Sign detection；Otherwise judge to detect into tail-light feature for non-red vehicle in rectangle frame region R1.
The present invention is further arranged to：It is detected using vehicle license plate characteristic in the step 5), specifically, to rectangle frame area
Domain R1 obtains segmentation threshold using Otsu algorithm, and carries out binary segmentation and obtain binary map, using linear structure element to segmentation
Obtained binary map carries out closed operation, then compares gray value and segmentation threshold extraction profile, seeks the external square of minimum of profile
Shape R3, if minimum enclosed rectangle R3 meets size condition, the ratio of width to height condition, foreground in locality condition and minimum enclosed rectangle R3
With its area ratio value, then judge that minimum enclosed rectangle R3 for license plate area, that is, is detected successfully.
The present invention is further arranged to：It is detected using tail-light feature in the step 5), specifically, by rectangle frame
The red area being partitioned into the R1 of region carries out Symmetry Detection, meets symmetric condition, then finds the car light in red area, i.e.,
It detects successfully.
The present invention is further arranged to：The Symmetry Detection, specifically, the car light area being likely to occur in red area
It is sampled in the R5 of domain, symmetry axis is set and is calculated in each position for the symmetry axis region R6 that traversal symmetry axis is likely to occur and is adopted
Ratio, the mean value of ratio calculated shared by symmetrical foreground white point take the upper limit value of mean value corresponding on every sampling line in sample
Symmetry axis Z is as the symmetrical shaft position of detection；If the symmetry axis Z location acquired meets position threshold, and before every sampling line
Scene vegetarian refreshments total number meets car light threshold value, then judges vehicle lamp area R5 for the new car car light that is found in rectangle frame region R1,
It detects successfully.
Compared with prior art, the invention has the advantages that：
So that the installation site of video camera is no longer restricted by proving operation, may be mounted at the arbitrary position of vehicle front
It sets, it is easy to operate, applied widely, convenient for flexibly installation and use replacement；It is unnecessary to reduce part by the judgement of same vehicle
It is follow-up to judge, not only save operation time, but also it is high with accuracy of detection, detection accuracy is strong, calculation amount is smaller, processing cost
Low advantage；Wherein, binarization segmentation is carried out to the full figure of current frame image in being detected using underbody shadow character, is utilized
It improves after Otsu algorithm carries out calculation processing and seeks shadow region to be determined, detection is accurate quick；Non- same vehicle is made whether
It handles for the judgement of red doubtful vehicle, and is detected using vehicle license plate characteristic for screw oil expeller, vehicle is utilized for non-screw oil expeller
Taillight feature is detected, and is overcome caused by the prior art using various features is detected each frame in detection vehicle
The problems such as processing speed is slow and detection accuracy is low, while flase drop can be effectively prevent.
The above is only the general introduction of technical solution of the present invention, in order to be better understood upon the technological means of the present invention, under
In conjunction with attached drawing, the invention will be further described in face.
Description of the drawings
Fig. 1 is the schematic diagram of proving operation in a kind of front vehicles automatic testing method of view-based access control model of the present invention；
Fig. 2 is a kind of flow chart of the front vehicles automatic testing method of view-based access control model of the present invention.
Specific implementation mode
With reference to the accompanying drawings of the specification, the present invention is further illustrated.
Continuous image sequence is obtained by vehicle-mounted video camera, and is transmitted to computer and is handled；In detection process
Advanced row proving operation carries out drafting mark as shown in Figure 1, according to the installation site of video camera to image to be detected 10 before starting
The proving operation of alignment 1 draws 2 calibration lines altogether, and respectively upper calibration line and lower calibration line are distributed in image to be detected 10
In, and the beginning and end of 2 calibration lines is respectively the both sides of front lane line 2.
According to 1 manual setting detection zone 3 of front lane line 2 and calibration line；The detection zone 3 is according to mapping to be checked
As the trapezoid area being made of front lane line 2 and calibration line 1 where 10 selection detection vehicle dimension ranges, for detection zone
Existing vehicle does not make detection process other than domain 3.And obtain 3 size of detection zone and actual size according to by calibration line 1
Quantitative relationship, to obtain the actual size of doubtful vehicle；It is initialized to having examined vehicle, the gray scale for having examined vehicle will be contained
Figure and the rectangle frame zero setting for indicating to have examined vehicle location.
The testing process of each frame is as shown in Fig. 2, include the following steps：
1, it carries out binarization segmentation to the full figure of current frame image specifically first to sample full figure, be obtained by sampling
Pixel represent the overall situation, recycle Otsu algorithm to seek full figure binarization segmentation threshold value POS and obtain full figure binary map.It will
Point of the gray value less than full figure binarization segmentation threshold value POS labeled as white point as foreground, remaining be labeled as black as background,
And profile is extracted to full figure binary map, the minimum enclosed rectangle R2 of profile is sought, internal is shadow region to be determined.
2, judge whether minimum enclosed rectangle R2 is underbody shadow region, specifically, judge profile central point whether
In the detection zone of selection, the foreground of shade in minimum enclosed rectangle R2 and its area are made, than obtaining shade ratio X, to utilize mark
The quantitative relationship that alignment obtains, which calculates, obtains the corresponding actual size Y of minimum enclosed rectangle R2 and minimum enclosed rectangle R2 institutes
In position and the ratio of width to height；If the shade ratio X and actual size Y of minimum enclosed rectangle R2 meet fractional threshold and size threshold
(during first time shade judges in this embodiment, X=0.7, Y=2.5 can be selected, both in the threshold range of setting),
And minimum enclosed rectangle R2 meets locality condition and the ratio of width to height condition, then judges minimum enclosed rectangle R2 for underbody shadow region
Domain.The minimum enclosed rectangle R2 of the condition of satisfaction is not found, then frame detection terminates, into the detection process of next frame image.
3, it is the areas You Che to be determined to obtain the rectangle frame region R1 for being located at and connecting above minimum enclosed rectangle R2 and therewith
The size in domain, rectangle frame region R1 meets W (R1)=W (R2), H (R1)=W (R2) * 3/4, wherein W (R1) and H (R1), W (R2)
The width of the width and height, minimum enclosed rectangle R2 of rectangle frame region R1 is indicated respectively.
Judge in rectangle frame region R1 whether to be to have examined vehicle again, by rectangle frame region R1 and examined vehicle carry out gray scale it is straight
Square figure, which compares, obtains gray scale ratio A, while comparing rectangle frame region R1 and having examined the positions and dimensions acquisition scale error B of vehicle,
If it is (adjacent in the first time of this embodiment that rectangle frame region R1 meets gray threshold, locality condition and size relative error threshold value
Frame relatively in, can be selected A=0.85, B=0.3, both in the threshold range of setting), then judge in rectangle frame region R1
To have examined vehicle, then success is tracked, adds one with vehicle frame number, preserves rectangle frame region R1 as the position for having examined vehicle, preservation is worked as
The gray scale map combining of previous frame contains the gray-scale map for having examined vehicle, and frame detection terminates, into the detection process of next frame image；It is no
Then tracking enters new car and detects not successfully.
4, new car detection is carried out to tracking failed vehicle, rectangle frame region R1 is converted to HSV space, tone is utilized
H, the triple channel Threshold segmentation of saturation degree S and brightness V goes out red area, and using red area labeled as white point as foreground, its
It is remaining label for as background (this embodiment first time screw oil expeller judge in, triple channel threshold value can be selected as respectively H=340,
S=0.4, V=100, three is in the threshold range of setting)；Whether meet red vehicle according to the sum of foreground pixel point
Whether be red in color region R4 to be determined of the threshold value to judge rectangle frame region R1, be judge be in rectangle frame region R1
Red vehicle is detected into vehicle license plate characteristic；Otherwise judge to examine into tail-light feature for non-red vehicle in rectangle frame region R1
It surveys；Wherein, color region R4 to be determined is selected as W (R4)=W (R1), H (R1)/3<H(R4)<H (R1) * 2/3, wherein W (R4) and H
(R4) width range and altitude range, H (R1) for indicating color region R4 to be determined respectively indicate rectangle frame region R1 height.
5, it is detected using vehicle license plate characteristic, specifically, segmentation threshold is obtained using Otsu algorithm to rectangle frame region R1,
And carry out binary segmentation and obtain binary map, closed operation is carried out to the binary map that segmentation obtains using linear structure element, is then compared
Profile is extracted compared with gray value and segmentation threshold, the minimum enclosed rectangle R3 of profile is sought, if minimum enclosed rectangle R3 meets size
Foreground and its area ratio value in condition, the ratio of width to height condition, locality condition and minimum enclosed rectangle R3 then judge minimum external
Rectangle R3 is license plate area, that is, is detected successfully, then finds new car, preserves rectangle frame region R1 as the position for having examined vehicle, protects
The gray-scale map of present frame is deposited as containing the gray-scale map for having examined vehicle, frame detection terminates；Otherwise frame detection is detected unsuccessfully
Terminate, into the detection process of next frame image.
Wherein, minimum enclosed rectangle R3 is in the first time car plate detection of this embodiment, meet condition W (R3)=W (R1)/
6, W (R3)=H (R3) * 5.41, X (R3)=W (R1)/2, Y (R3)=H (R1)/2, are selected in the threshold range of setting,
Middle W (R1) and H (R1) indicate that the width and height of rectangle frame region R1, X (R3) and Y (R3) indicate minimum external square respectively respectively
Abscissa and ordinate of the center of shape R3 in rectangle frame region R1, W (R3) and H (R3) indicate minimum enclosed rectangle R3 respectively
Width and height.
6, it is detected using tail-light feature, specifically, the red area being partitioned into rectangle frame region R1 is carried out
Symmetry Detection, meets symmetric condition, then finds the car light in red area, that is, detects successfully, then finds new car, preserves rectangle
Frame region R1 preserves the gray-scale map of present frame as containing the gray-scale map for having examined vehicle, frame inspection as the position for having examined vehicle
Survey terminates；Otherwise it detects unsuccessfully frame detection also to terminate, into the detection process of next frame image.
Wherein, the Symmetry Detection is specifically adopted in the vehicle lamp area R5 being likely to occur in red area
Sample is arranged symmetry axis in each position for the symmetry axis region R6 that traversal symmetry axis is likely to occur and calculates every pumping in sampling
Ratio, the mean value of ratio calculated shared by symmetrical foreground white point on line-transect take the corresponding symmetry axis Z of the upper limit value of mean value as inspection
Survey symmetrical shaft position；If the symmetry axis Z location acquired meets position threshold, and the foreground pixel point total number of every sampling line
Meet car light threshold value, then judges that vehicle lamp area R5 for the new car car light found in rectangle frame region R1, that is, is detected successfully；Car light
Region R5 meets condition W (R5)=W (R1), H (R5)=H (R1)/3, X (R5) in the first time car light detection of this embodiment
Symmetry axis in=W (R1)/2 and Y (R5)=H (R1)/2, symmetry axis region R6 meets Z=W (R1) * 0.45, in setting
Selection in threshold range, wherein W (R1) and H (R1) indicate the width and height of rectangle frame region R1, W (R5) and H (R5) respectively
The width and height of vehicle lamp area R5 are indicated respectively, and X (R5) and Y (R5) indicate the center of car light detection zone R5 in rectangle respectively
Abscissa in frame region R1 and ordinate.
7, during being detected processing by frame successively, it is continuous set frame number such as 5 frames and do not find examined vehicle and be judged as with losing,
It is initialized to having examined vehicle, the rectangle frame region R1 that vehicle location has been examined containing the gray-scale map and expression of having examined vehicle is clear
Zero；To not with losing the case where, share 5 frames with vehicle frame number and find and examined vehicle and then judge the vehicle for front vehicles, to it into the ranks
Every candid photograph.
The basic principles and main features and advantage of the present invention have been shown and described above.The technical staff of the industry should
Understand, the present invention is not limited to the above embodiments, and the above embodiments and description only describe the originals of the present invention
Reason, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes and improvements
It all fall within the protetion scope of the claimed invention.The claimed scope of the invention is by appended claims and its equivalent circle
It is fixed.
Claims (8)
1. a kind of front vehicles automatic testing method of view-based access control model, which is characterized in that include the following steps：
1) continuous image sequence is obtained by vehicle-mounted video camera, and is transmitted to computer and is handled；
2) before detection process starts, according to the installation site of video camera, to image to be detected draw the mark of calibration line
Fixed operation, and according to front lane line and calibration line manual setting detection zone；Detection zone size and reality are obtained by calibration line
The quantitative relationship of border size, to obtain the actual size of doubtful vehicle；
3) computer is detected processing by frame successively, is detected using underbody shadow character, primarily determines that detection zone can
The rectangle frame region R1 that can have front vehicles, has, enters step 4)；Without the detection process for then entering next frame image；
4) judge in rectangle frame region R1 whether to be to have examined vehicle, be to track success, add one with vehicle frame number, preserve rectangle frame area
For domain R1 as the position for having examined vehicle, the gray scale map combining for preserving present frame contains the gray-scale map for having examined vehicle, frame detection knot
Beam, into the detection process of next frame image；Otherwise it enters step 5)；
5) judge in rectangle frame region R1 whether to be red doubtful vehicle, be detected using vehicle license plate characteristic, it is otherwise sharp
It is detected with tail-light feature；
Using vehicle license plate characteristic or tail-light feature is utilized, detects and successfully then finds new car, preservation rectangle frame region R1, which is used as, have been examined
The position of vehicle preserves the gray-scale map of present frame as containing the gray-scale map for having examined vehicle, and frame detection terminates, into next frame
The detection process of image；Using vehicle license plate characteristic or tail-light feature is utilized, unsuccessfully frame detection also terminates for detection, into next frame
The detection process of image；
6) during being detected processing by frame successively, it is continuous set frame number and do not find examined vehicle and be judged as with losing, to having examined vehicle
It is initialized, containing the gray-scale map for having examined vehicle and will indicate that the rectangle frame region R1 for having examined vehicle location is reset；To not
The case where with losing, with vehicle frame number it is shared set frame number and find examined vehicle and then judge that the vehicle as front vehicles, is captured it；
The calibration line includes upper calibration line and lower calibration line totally 2, is distributed in image to be detected；
The detection zone be according to image to be detected select detection vehicle dimension range where by front lane line and calibration
The trapezoid area that line is constituted does not make detection process for vehicle existing other than detection zone；
It is detected using underbody shadow character in the step 3), specifically, binaryzation is carried out to the full figure of current frame image
Segmentation obtains full figure binarization segmentation threshold value and full figure binary map using Otsu algorithm；By gray value less than full figure binaryzation point
Cut the point of threshold value labeled as white point as foreground, remaining, as background, and to full figure binary map extraction profile, asks labeled as black
The minimum enclosed rectangle R2 of contouring, internal is shadow region to be determined；
Judge whether minimum enclosed rectangle R2 is underbody shadow region, is, is then obtained above minimum enclosed rectangle R2 simultaneously
The rectangle frame region R1 to connect therewith has vehicle region, the size of rectangle frame region R1 to meet W (R1)=W (R2), H for be determined
(R1)=W (R2) * 3/4, wherein W (R1) and H (R1), W (R2) indicate that the width of rectangle frame region R1 and height, minimum are outer respectively
Connect the width of rectangle R2；Otherwise frame detection terminates, into the detection process of next frame image.
2. a kind of front vehicles automatic testing method of view-based access control model according to claim 1, it is characterised in that：It is described right
The full figure of current frame image carries out binarization segmentation and is specifically first sampled to full figure, is represented by the pixel that sampling obtains
The overall situation recycles Otsu algorithm to seek full figure binarization segmentation threshold value.
3. a kind of front vehicles automatic testing method of view-based access control model according to claim 1, it is characterised in that：It is described to sentence
Whether disconnected minimum enclosed rectangle R2 is underbody shadow region, specifically, judge profile central point whether selection detection zone
In domain, the foreground of shade in minimum enclosed rectangle R2 and its area are made, than obtaining shade ratio, to utilize the amount of calibration line acquisition
Change relationship, which calculates, obtains the corresponding actual sizes of minimum enclosed rectangle R2 and the positions minimum enclosed rectangle R2 and width height
Than；
If meeting fractional threshold, size threshold, locality condition and the ratio of width to height condition, judge minimum enclosed rectangle R2 for underbody the moon
Shadow zone domain.
4. a kind of front vehicles automatic testing method of view-based access control model according to claim 1, it is characterised in that：The step
It is rapid 4) in judge in rectangle frame region R1 whether to be to have examined vehicle, specifically, by rectangle frame region R1 and to have examined vehicle progress grey
Degree histogram, which compares, obtains gray scale ratio, while comparing rectangle frame region R1 and having examined the positions and dimensions acquisition size mistake of vehicle
Difference judges in rectangle frame region R1 if meeting gray threshold, locality condition and size relative error threshold value to have examined vehicle.
5. a kind of front vehicles automatic testing method of view-based access control model according to claim 1, it is characterised in that：The step
It is rapid 5) in judge in rectangle frame region R1 whether to be that red doubtful vehicle specifically carries out newly to tracking failed vehicle
Car test is surveyed, and rectangle frame region R1 is converted to HSV space, is gone out using the triple channel Threshold segmentation of tone H, saturation degree S and brightness V
Red area, and using red area labeled as white point as foreground, remaining be labeled as black as background；
Judge the color area to be determined of rectangle frame region R1 according to whether the sum of foreground pixel point meets red vehicle threshold value
Whether it is red in the R4 of domain, is to judge to detect into vehicle license plate characteristic for red vehicle in rectangle frame region R1；Otherwise judge square
It is non-red vehicle in shape frame region R1, is detected into tail-light feature.
6. the front vehicles automatic testing method of a kind of view-based access control model according to claim 1 or 5, it is characterised in that：Institute
It states in step 5) and is detected using vehicle license plate characteristic, specifically, segmentation threshold is obtained using Otsu algorithm to rectangle frame region R1,
And carry out binary segmentation and obtain binary map, closed operation is carried out to the binary map that segmentation obtains using linear structure element, is then compared
Profile is extracted compared with gray value and segmentation threshold, the minimum enclosed rectangle R3 of profile is sought, if minimum enclosed rectangle R3 meets size
Foreground and its area ratio value in condition, the ratio of width to height condition, locality condition and minimum enclosed rectangle R3 then judge minimum external
Rectangle R3 is license plate area, that is, is detected successfully.
7. a kind of front vehicles automatic testing method of view-based access control model according to claim 5, it is characterised in that：The step
Rapid 5) middle utilization tail-light feature is detected, and specifically, the red area being partitioned into rectangle frame region R1 is carried out symmetrical
Property detection, meet symmetric condition, then find the car light in red area, that is, detect successfully.
8. a kind of front vehicles automatic testing method of view-based access control model according to claim 7, it is characterised in that：It is described right
The detection of title property, specifically, is sampled in the vehicle lamp area R5 being likely to occur in red area, may be gone out in traversal symmetry axis
Each position setting symmetry axis of existing symmetry axis region R6 simultaneously calculates symmetrical foreground white point on every sampling line in sampling
Shared ratio, the mean value of ratio calculated take the corresponding symmetry axis Z of the upper limit value of mean value as the symmetrical shaft position of detection；
If the symmetry axis Z location acquired meets position threshold, and the foreground pixel point total number of every sampling line meets car light threshold
Value, then judge that vehicle lamp area R5 for the new car car light found in rectangle frame region R1, that is, is detected successfully.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510296813.4A CN104866838B (en) | 2015-06-02 | 2015-06-02 | A kind of front vehicles automatic testing method of view-based access control model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510296813.4A CN104866838B (en) | 2015-06-02 | 2015-06-02 | A kind of front vehicles automatic testing method of view-based access control model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104866838A CN104866838A (en) | 2015-08-26 |
| CN104866838B true CN104866838B (en) | 2018-08-03 |
Family
ID=53912659
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510296813.4A Expired - Fee Related CN104866838B (en) | 2015-06-02 | 2015-06-02 | A kind of front vehicles automatic testing method of view-based access control model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104866838B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108021856B (en) * | 2016-10-31 | 2020-09-15 | 比亚迪股份有限公司 | Vehicle tail lamp identification method and device and vehicle |
| CN108229248A (en) * | 2016-12-14 | 2018-06-29 | 贵港市瑞成科技有限公司 | Vehicle checking method based on underbody shade |
| CN107154045B (en) * | 2017-03-31 | 2020-06-09 | 南京航空航天大学 | Accurate positioning method for vehicles on flow guide area illegal rolling line based on computer vision |
| CN111414857B (en) * | 2020-03-20 | 2023-04-18 | 辽宁工业大学 | Front vehicle detection method based on vision multi-feature fusion |
| CN112528056B (en) * | 2020-11-29 | 2021-09-07 | 枞阳县中邦科技信息咨询有限公司 | Double-index field data retrieval system and method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102610104A (en) * | 2012-03-16 | 2012-07-25 | 南京航空航天大学 | Onboard front vehicle detection method |
| CN103150898A (en) * | 2013-01-25 | 2013-06-12 | 大唐移动通信设备有限公司 | Method and device for detection of vehicle at night and method and device for tracking of vehicle at night |
| CN103473534A (en) * | 2013-09-10 | 2013-12-25 | 西安翔迅科技有限责任公司 | Vehicle detecting method based on video |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101458814B (en) * | 2007-12-13 | 2012-02-01 | 东软集团股份有限公司 | Method and apparatus for separating objects interesting region from image |
-
2015
- 2015-06-02 CN CN201510296813.4A patent/CN104866838B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102610104A (en) * | 2012-03-16 | 2012-07-25 | 南京航空航天大学 | Onboard front vehicle detection method |
| CN103150898A (en) * | 2013-01-25 | 2013-06-12 | 大唐移动通信设备有限公司 | Method and device for detection of vehicle at night and method and device for tracking of vehicle at night |
| CN103473534A (en) * | 2013-09-10 | 2013-12-25 | 西安翔迅科技有限责任公司 | Vehicle detecting method based on video |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104866838A (en) | 2015-08-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104866838B (en) | A kind of front vehicles automatic testing method of view-based access control model | |
| CN105913041B (en) | It is a kind of based on the signal lamp recognition methods demarcated in advance | |
| CN105488454B (en) | Front vehicles detection and ranging based on monocular vision | |
| CN105160691A (en) | Color histogram based vehicle body color identification method | |
| CN108764352B (en) | Method and device for detecting repeated page content | |
| CN108052904B (en) | Method and device for acquiring lane line | |
| CN111753692A (en) | Target object extraction method, product detection method, device, computer and medium | |
| Chen | Nighttime vehicle light detection on a moving vehicle using image segmentation and analysis techniques | |
| CN102393902A (en) | Vehicle color detection method based on H_S two-dimensional histogram and regional color matching | |
| CN107909012B (en) | Real-time vehicle tracking detection method and device based on disparity map | |
| CN115082466B (en) | PCB surface welding spot defect detection method and system | |
| CN107273838A (en) | Traffic lights capture the processing method and processing device of picture | |
| CN106709412B (en) | Traffic sign detection method and device | |
| CN104766344B (en) | Vehicle checking method based on movement edge extractor | |
| CN110175556B (en) | Remote sensing image cloud detection method based on Sobel operator | |
| CN103440503A (en) | Vehicle body color detection and identification method | |
| Naufal et al. | Preprocessed mask RCNN for parking space detection in smart parking systems | |
| CN103530622A (en) | Car logo identification method | |
| CN104598907A (en) | Stroke width figure based method for extracting Chinese character data from image | |
| CN106127124A (en) | The automatic testing method of the abnormal image signal in region, taxi front row | |
| CN108765456B (en) | Target tracking method and system based on linear edge characteristics | |
| Mahersatillah et al. | Unstructured road detection and steering assist based on hsv color space segmentation for autonomous car | |
| CN108734113A (en) | Vehicle automatic marking method, storage medium, electronic equipment, system | |
| CN107066929B (en) | Hierarchical recognition method for parking events of expressway tunnel integrating multiple characteristics | |
| CN112598674A (en) | Image processing method and device for vehicle and vehicle |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| 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: 20180803 Termination date: 20210602 |