CN108022429A - A kind of method and device of vehicle detection - Google Patents
A kind of method and device of vehicle detection Download PDFInfo
- Publication number
- CN108022429A CN108022429A CN201610971878.9A CN201610971878A CN108022429A CN 108022429 A CN108022429 A CN 108022429A CN 201610971878 A CN201610971878 A CN 201610971878A CN 108022429 A CN108022429 A CN 108022429A
- Authority
- CN
- China
- Prior art keywords
- pixel
- image
- target vehicle
- benchmark image
- statistic unit
- 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
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
- G08G1/0175—Detecting movement of traffic to be counted or controlled identifying vehicles by photographing vehicles, e.g. when violating traffic rules
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
-
- 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
Abstract
The invention discloses a kind of method and device of vehicle detection, for solving when being blocked between car and car, or when vehicle is close with ground color, generation flase drop, it is difficult to which the problem of ensureing accuracy of detection, method include：Using Binocular Stereo Vision System, obtain image pair, an image is chosen as benchmark image from image pair, according to benchmark image, Preliminary detection is carried out to target vehicle, the first bounding box of the Partial Feature comprising target vehicle is determined in benchmark image, according to image pair, determine the D coordinates value for each pixel that the image in the first bounding box includes, according to the D coordinates value determined, determine the pixel that target vehicle includes.Combine in benchmark image and Preliminary detection is carried out to target vehicle, and the D coordinates value of each pixel that the first image in bounding box includes, to determine pixel that the target vehicle includes, the pixel for the target vehicle determined is more accurate, improves vehicle detection precision.
Description
Technical field
The present invention relates to intelligent transportation field, more particularly to a kind of method and device of vehicle detection.
Background technology
, it is necessary to obtain the information of driving vehicle on road in road traffic, wherein, the information of vehicle is believed including car plate
Breath, car body information etc., it is therefore desirable to be detected to vehicle.
Current most of vehicle detections are all based on the vehicle checking method of video analysis, are obtained and schemed by single camera
Picture, then carries out vehicle detection in accessed image, finally obtains position of the vehicle in accessed image.It is existing
Having vehicle checking method, there are the following problems：Due to just with overall profile information, when being blocked between car and car,
Or vehicle it is close with ground color when, it may occur that flase drop, it is difficult to ensure accuracy of detection.
The content of the invention
The object of the present invention is to provide a kind of method and device of vehicle detection, is blocked with solving to work as between car and car
When, or when vehicle is close with ground color, it may occur that flase drop, it is difficult to the problem of ensureing accuracy of detection.
The purpose of the present invention is what is be achieved through the following technical solutions：
A kind of method of vehicle detection, including：
Using Binocular Stereo Vision System, image pair is obtained；
Choose an image as benchmark image from described image centering, according to the benchmark image, to target vehicle into
Row Preliminary detection, determines the first bounding box of the Partial Feature comprising the target vehicle in the benchmark image；
According to described image pair, the three-dimensional coordinate for each pixel that the image in first bounding box includes is determined
Value；
According to the D coordinates value determined, the pixel that target vehicle includes described in the benchmark image is determined.
Optionally, it is described that Part-base vehicle checking methods are used to target vehicle progress Preliminary detection.
Optionally, the D coordinates value that the basis is determined, determines that target vehicle includes described in the benchmark image
Pixel, including：
The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, removes
The absolute value of coordinate value on y-axis direction is less than the pixel of first threshold.
Optionally, the D coordinates value that the basis is determined, determines that target vehicle includes described in the benchmark image
Pixel, further include：
For each the first bounding box, at least one change in coordinate axis direction in x-axis and z-axis, according to setpoint distance
M statistic unit is divided, and from the M statistic unit, determines the total quantity of included pixel more than setting the
The ratio of the total quantity for the pixel that two threshold values and the total quantity of the pixel included are included with the M statistic unit is big
In the N number of statistic unit for setting the 3rd threshold value, M is the integer more than or equal to 1, and N is the integer less than or equal to M；
The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, determines
Go out the minimum range of each statistic unit and character pixel point set in N number of statistic unit, and from N number of statistic unit
In, determine that minimum range is less than P statistic unit of the 4th threshold value of setting, wherein, the minimum range is in statistic unit
Each pixel and the character pixel point set in each pixel at least one change in coordinate axis direction away from
Minimum value from；The character pixel point set is the pixel point set of any feature of the fixed target vehicle；
The pixel that will be included in the P statistic unit, is determined as target vehicle described in the benchmark image and includes
Pixel.
Optionally, after determining the pixel that target vehicle includes described in the benchmark image, including：
The pixel that target vehicle includes according to the benchmark image, determines to include institute in the benchmark image
State the second bounding box of whole pixels of target vehicle.
Based on the inventive concept same with method, an embodiment of the present invention provides a kind of device of vehicle detection, including：
Acquisition module, for obtaining image pair；
First determining module, for choosing an image as benchmark image from described image centering, according to the benchmark
Image, carries out Preliminary detection to target vehicle, determines to include the Partial Feature of the target vehicle in the benchmark image
The first bounding box；
Second determining module, for according to described image pair, it is every to determine that the image in first bounding box includes
The D coordinates value of a pixel；
3rd determining module, for according to the D coordinates value determined, determining target carriage described in the benchmark image
The pixel included.
Optionally, it is described that Part-base vehicle checking methods are used to target vehicle progress Preliminary detection.
Optionally, the 3rd determining module is specifically used for：
The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, removes
The absolute value of coordinate value on y-axis direction is less than the pixel of first threshold.
Optionally, the 3rd determining module is specifically used for：
For each the first bounding box, at least one change in coordinate axis direction in x-axis and z-axis, respectively according to setting
Distance M statistic unit of division, and from the M statistic unit, determine that the total quantity of included pixel is more than and set
The ratio of the total quantity for the pixel that the total quantity for the pixel determined second threshold and included is included with the M statistic unit
Value is more than N number of statistic unit of the 3rd threshold value of setting, and M is the integer more than or equal to 1, and N is the integer less than or equal to M；
The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, determines
Go out the minimum range of each statistic unit and character pixel point set in N number of statistic unit, and from N number of statistic unit
In, determine that minimum range is less than P statistic unit of the 4th threshold value of setting, wherein, the minimum range is in statistic unit
Each pixel and the character pixel point set in each pixel at least one change in coordinate axis direction away from
Minimum value from；The character pixel point set is the pixel point set of any feature of the fixed target vehicle；
The pixel that will be included in the P statistic unit, is determined as target vehicle described in the benchmark image and includes
Pixel.
Optionally, the 3rd determining module is additionally operable to：
The pixel that target vehicle includes according to the benchmark image, determines to include institute in the benchmark image
State the second bounding box of whole pixels of target vehicle.
A kind of method and device of vehicle detection provided in an embodiment of the present invention, using Binocular Stereo Vision System, obtains
Image pair, chooses an image as benchmark image from described image centering, according to the benchmark image, target vehicle is carried out
Preliminary detection, determines the first bounding box of the Partial Feature comprising the target vehicle, according to institute in the benchmark image
Image pair is stated, determines the D coordinates value for each pixel that the image in first bounding box includes, according to determining
D coordinates value, determine the pixel that target vehicle includes described in the benchmark image.Due to combining in benchmark image
In the D coordinates value of each pixel that the image in Preliminary detection, and the first bounding box includes is carried out to target vehicle,
To determine target vehicle includes described in the benchmark image pixel so that the pixel for the target vehicle determined is more
Accurately, so as to improve vehicle detection precision.
Brief description of the drawings
Fig. 1 is a kind of method flow diagram of vehicle detection provided in an embodiment of the present invention；
Fig. 2 is one group of image pair provided in an embodiment of the present invention；
Fig. 3 is a kind of vehicle Preliminary detection figure provided in an embodiment of the present invention；
Fig. 4 is a kind of vehicle depth information figure provided in an embodiment of the present invention；
Fig. 5 is a kind of vehicle detection result figure provided in an embodiment of the present invention；
Fig. 6 is the method flow diagram of another vehicle detection provided in an embodiment of the present invention；
Fig. 7 is a kind of schematic device of vehicle detection provided in an embodiment of the present invention.
Embodiment
Below in conjunction with attached drawing, technical solution provided in an embodiment of the present invention is described in detail.
An embodiment of the present invention provides a kind of method of vehicle detection, as shown in Figure 1, including following operation：
Step 100, using Binocular Stereo Vision System, obtain image pair.
Step 110, from described image centering choose an image as benchmark image, according to the benchmark image, to mesh
Mark vehicle and carry out Preliminary detection, determine that first of the Partial Feature comprising the target vehicle surrounds in the benchmark image
Box.
Step 120, according to described image pair, determine each pixel that the image in first bounding box includes
D coordinates value.
Wherein, the quantity of first bounding box is at least two, its quantity is not limited in the embodiment of the present invention.
The D coordinates value that step 130, basis are determined, determines the picture that target vehicle includes described in the benchmark image
Vegetarian refreshments.
A kind of method of vehicle detection provided in an embodiment of the present invention, using Binocular Stereo Vision System, obtains image pair,
An image is chosen as benchmark image from described image centering, and according to the benchmark image, target vehicle is tentatively examined
Survey, the first bounding box of the Partial Feature comprising the target vehicle is determined in the benchmark image, according to described image
It is right, the D coordinates value for each pixel that the image in first bounding box includes is determined, according to the three-dimensional determined
Coordinate value, determines the pixel that target vehicle includes described in the benchmark image.Due to combining in benchmark image to mesh
Mark vehicle carries out the D coordinates value for each pixel that the image in Preliminary detection, and the first bounding box includes, to determine
The pixel that target vehicle includes described in the benchmark image so that the pixel for the target vehicle determined is more accurate,
So as to improve vehicle detection precision.
It is described that Part-base vehicle detections are used to target vehicle progress Preliminary detection in a kind of optional implementation
Method.
In a kind of optional implementation, an image is chosen as benchmark image from described image centering in step 100,
Including：
To described image to carrying out EP point correction, the image pair after EP point correction, chooses an image
As the benchmark image.
To described image to carrying out EP point correction, same three dimensions point can be made to open the projection on image in left and right two
The ordinate of point is equal, search cost during reducing Stereo matching, improves the accuracy rate of Stereo matching.
In the embodiment of the present invention, D coordinates value that the basis is determined determines target described in the benchmark image
The pixel that vehicle includes, including following optional implementation：
Mode one, remove y-axis direction (i.e. the short transverse of target vehicle) on non-targeted vehicle pixel, specifically such as
Under：
The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, removes
The absolute value of coordinate value on y-axis direction is less than the pixel of first threshold, so that in the short transverse of target vehicle, effectively goes
Except the interference of the pixel of non-targeted vehicle so that the pixel that the target vehicle determined includes is more accurate.
Further, it is possible to by first bounding box, except the absolute value of the coordinate value on y-axis direction is less than first threshold
Pixel outside residual pixel point be determined as the pixel that target vehicle includes.
Mode two, remove x-axis direction (i.e. the width of target vehicle) on non-targeted vehicle pixel, specifically such as
Under：
For each the first bounding box, M statistic unit is divided according to setpoint distance in x-axis direction, and from the M
In statistic unit, the total of the pixel that the total quantity of included pixel is more than setting second threshold and is included is determined
The ratio of the total quantity for the pixel that quantity is included with the M statistic unit is more than N number of statistic unit of the 3rd threshold value of setting,
M is the integer more than or equal to 1, and N is the integer less than or equal to M；
The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, determines
Go out the minimum range of each statistic unit and character pixel point set in N number of statistic unit, and from N number of statistic unit
In, determine that minimum range is less than P statistic unit of the 4th threshold value of setting, wherein, the minimum range is in statistic unit
Each pixel and the character pixel point set in each pixel distance in the direction of the x axis in minimum value；
The pixel point set of a certain feature for the target vehicle that the character pixel point set is to determine out；By described in
The pixel included in P statistic unit, is determined as the pixel that target vehicle includes described in the benchmark image.
Under which, in the width of target vehicle, the interference of the pixel of non-targeted vehicle is effectively eliminated so that
The pixel that the target vehicle determined includes is more accurate.
Mode three, remove z-axis direction (i.e. the length direction of target vehicle) on non-targeted vehicle pixel, specifically such as
Under：
For each the first bounding box, M statistic unit is divided according to setpoint distance in z-axis direction, and from the M
In statistic unit, the total of the pixel that the total quantity of included pixel is more than setting second threshold and is included is determined
The ratio of the total quantity for the pixel that quantity is included with the M statistic unit is more than N number of statistic unit of the 3rd threshold value of setting,
M is the integer more than or equal to 1, and N is the integer less than or equal to M；
The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, determines
Go out the minimum range of each statistic unit and character pixel point set in N number of statistic unit, and from N number of statistic unit
In, determine that minimum range is less than P statistic unit of the 4th threshold value of setting, wherein, the minimum range is in statistic unit
Each pixel and the character pixel point set in each pixel distance in the z-axis direction in minimum value；
The pixel point set of a certain feature for the target vehicle that the character pixel point set is to determine out；By described in
The pixel included in P statistic unit, is determined as the pixel that target vehicle includes described in the benchmark image.
Under which, in the length direction of target vehicle, the interference of the pixel of non-targeted vehicle is effectively eliminated so that
The pixel that the target vehicle determined includes is more accurate.
Above-mentioned three kinds of modes can be used alone, and can also be used in combination, such as：
1st, the mode that employing mode one is combined with mode two, determines that target vehicle includes described in the benchmark image
Pixel；So as to eliminate the interference of the pixel of non-targeted vehicle on y-axis direction and x-axis direction.Specifically：Can be first
Employing mode one removes the pixel of non-targeted vehicle on y-axis direction, then employing mode two removes non-targeted vehicle on x-axis direction
Pixel, determine the pixel that target vehicle includes described in the benchmark image；Can also the first removal of employing mode two x
The pixel of non-targeted vehicle on direction of principal axis, then employing mode one remove the pixel of non-targeted vehicle on y-axis direction, so that really
Make the pixel that target vehicle includes described in the benchmark image.
2nd, the mode that employing mode one is combined with mode three, determines that target vehicle includes described in the benchmark image
Pixel；So as to eliminate the interference of the pixel of non-targeted vehicle on y-axis direction and z-axis direction.Specifically：Can be first
Employing mode one removes the pixel of non-targeted vehicle on y-axis direction, then employing mode three removes non-targeted vehicle on z-axis direction
Pixel, determine the pixel that target vehicle includes described in the benchmark image；Can also the first removal of employing mode three z
The pixel of non-targeted vehicle on direction of principal axis, then employing mode one remove the pixel of non-targeted vehicle on y-axis direction, so that really
Make the pixel that target vehicle includes described in the benchmark image.
3rd, the mode that employing mode two is combined with mode three, determines that target vehicle includes described in the benchmark image
Pixel；So as to eliminate the interference of the pixel of non-targeted vehicle on x-axis direction and z-axis direction.Specifically：Can be first
Employing mode two removes the pixel of non-targeted vehicle on x-axis direction, then employing mode three removes non-targeted vehicle on z-axis direction
Pixel, determine the pixel that target vehicle includes described in the benchmark image；Can also the first removal of employing mode three x
The pixel of non-targeted vehicle on direction of principal axis, then employing mode two remove the pixel of non-targeted vehicle on z-axis direction, so that really
Make the pixel that target vehicle includes described in the benchmark image.
4th, the mode that employing mode one, mode two are combined with mode three, determines target described in the benchmark image
The pixel that vehicle includes；So as to eliminate the dry of the pixel of non-targeted vehicle on y-axis direction, x-axis direction and z-axis direction
Disturb.Specifically, using the order of three kinds of modes, the embodiment of the present invention is not construed as limiting it.
Based on any of the above-described embodiment, after determining the pixel that target vehicle includes described in the benchmark image,
Including：
The pixel that target vehicle includes according to the benchmark image, determines to include institute in the benchmark image
State the second bounding box of whole pixels of target vehicle.
Below by two specific embodiments, a kind of method of vehicle detection of the present invention is described in detail.
Embodiment one, by Binocular Stereo Vision System obtain image pair, to the image to carry out EP point correction, obtain
Image pair after correction, as shown in Figure 2.
To select left-side images to carry out auto model Preliminary detection as exemplified by benchmark image, after Preliminary detection,
Multiple bounding boxs are determined in benchmark image, the bounding box is the rectangle frame for including target vehicle Partial Feature, and difference is surrounded
There can be overlapping part between box, auto model Preliminary detection can use Part-base scheduling algorithms to realize, the present invention is real
Apply example not limit, below using exemplified by Part-base, to illustrate the method flow of vehicle detection：
First, the Part-base models of training vehicle；Prepare the positive negative sample of vehicle, wherein, preferably instruct in order to obtain
Practice effect, prepare the picture 2000 containing cart, trolley feature and open, as positive sample, while prepare not containing vehicle characteristics
Picture 1000 is opened, as negative sample；Using positive negative sample, using Part-base methods, it is trained, obtains vehicle detection mould
Type.
Then, Preliminary detection is carried out to target vehicle using obtained vehicle detection model, obtains preliminary information of vehicles, i.e.,
The first bounding box of the Partial Feature comprising the target vehicle is determined in the benchmark image, as shown in Figure 3.
Secondly, the image pair after being corrected according to EP point, obtains the depth information of the benchmark image, that is, obtains reference map
D coordinates value (the x of each pixel included as inwp,ywp,zwp), according to the three-dimensional coordinate of each pixel be worth to as
Depth information figure shown in Fig. 4.
Finally, for each first bounding box, using the mapping function of EP point timing, by each first bounding box area
Domain mapping, according to the D coordinates value of the pixel in each first bounding box region, removes target on the benchmark image
The pixel of chaff interferent outside vehicle.
Wherein, according to the D coordinates value of the pixel in each first bounding box region, remove outside target vehicle
The pixel of chaff interferent, detailed process are as follows：
Using y-axis direction as vehicle-height direction, z-axis direction is vehicle lengthwise direction, and x-axis direction is that vehicle-width direction is
Example.
, can be according to y-axis side in D coordinates value when removing the pixel of the chaff interferent on y-axis direction outside target vehicle
To value remove ground region interference, that is, remove ywpLess than given threshold thgPixel.
When removing the pixel of the chaff interferent on z-axis direction outside target vehicle, in z-axis direction per Δ SzDistance divides one into
A statistic unit (bin), is divided into M altogetherZA bin, each statistic unit are denoted as Bk, each B counted in listkPixel quantity
It is denoted as nk, k ∈ [0, MZ] in positive integer.The total quantity for removing included pixel is less than or equal to second threshold thnzAnd
Comprising pixel total quantity and the MZThe ratio of the total quantity for the pixel that a statistic unit includes is less than or equal to
3rd threshold value thnrzStatistic unit, obtain target vehicle pixel point set in each statistic unitIt is shown below：
First, the statistic unit where target vehicle vehicle license plate characteristic is determined, as target vehicle pixel set's
Initial pixel point set, it is then determined that going out MZEach statistic unit and the most narrow spacing of initial pixel point set in a statistic unit
From, and from the MZIn a statistic unit, determine that minimum range is less than given threshold thBzP statistic unit, by the P
The pixel of target vehicle is added in target vehicle pixel set in a statistic unit, obtains whole pictures of the target vehicle
Vegetarian refreshments, finally according to whole pixels of the target vehicle, determines the target vehicle included in the benchmark image
Whole pixels the second bounding box, as shown in Figure 5.
When removing the pixel of the chaff interferent on x-axis direction outside target vehicle, with remove on z-axis direction target vehicle it
The processing procedure during pixel of outer chaff interferent is identical, is repeated no more in the embodiment of the present invention.
Embodiment two, vehicle detection process provided in this embodiment as shown in fig. 6, including：
Step 601, using Binocular Stereo Vision System, obtain image pair.
Step 602, to described image to carry out EP point correction.
Step 603, from after correction image pair choose an image as benchmark image.
Step 604, carry out Preliminary detection in benchmark image to target vehicle, determines to include the portion of the target vehicle
First bounding box of dtex sign.
Step 605, using the image pair after correction, obtain the three-dimensional coordinate of each pixel included in benchmark image
Value.
Step 606, for each first bounding box, using the mapping function of EP point timing, each first is surrounded
On benchmark image after box area maps to EP point correction.
Step 607, go according to the D coordinates value of the pixel on the benchmark image in each first bounding box region
Pixel in addition to target vehicle.
Step 608, will remove remaining pixel after pixel on the benchmark image outside target vehicle, be determined as
The pixel of target vehicle described in the benchmark image.
Step 609, the pixel according to the target vehicle, determine the target carriage included in the benchmark image
Whole pixels the second bounding box.
Based on the inventive concept same with method, the embodiment of the present invention also provides a kind of device of vehicle detection, such as Fig. 7 institutes
Show, including：
Acquisition module 701, for obtaining image pair；
First determining module 702, for choosing an image as benchmark image from described image centering, according to the base
Quasi- image, Preliminary detection is carried out to target vehicle, determines that the part comprising the target vehicle is special in the benchmark image
First bounding box of sign；
Second determining module 703, for according to described image pair, determining what the image in first bounding box included
The D coordinates value of each pixel；
3rd determining module 704, for according to the D coordinates value determined, determining target described in the benchmark image
The pixel that vehicle includes.
A kind of device of vehicle detection provided in an embodiment of the present invention, using Binocular Stereo Vision System, obtains image pair,
An image is chosen as benchmark image from described image centering, and according to the benchmark image, target vehicle is tentatively examined
Survey, the first bounding box of the Partial Feature comprising the target vehicle is determined in the benchmark image, according to described image
It is right, the D coordinates value for each pixel that the image in first bounding box includes is determined, according to the three-dimensional determined
Coordinate value, determines the pixel that target vehicle includes described in the benchmark image.Due to combining in benchmark image to mesh
Mark vehicle carries out the D coordinates value for each pixel that the image in Preliminary detection, and the first bounding box includes, to determine
The pixel that target vehicle includes described in the benchmark image so that the pixel for the target vehicle determined is more accurate,
So as to improve vehicle detection precision.
Optionally, it is described that Part-base vehicle checking methods are used to target vehicle progress Preliminary detection.
Optionally, the 3rd determining module is specifically used for：
The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, removes
The absolute value of coordinate value on y-axis direction is less than the pixel of first threshold.
Optionally, the 3rd determining module is specifically used for：
For each the first bounding box, at least one change in coordinate axis direction in x-axis and z-axis, according to setpoint distance
M statistic unit is divided, and from the M statistic unit, determines the total quantity of included pixel more than setting the
The ratio of the total quantity for the pixel that two threshold values and the total quantity of the pixel included are included with the M statistic unit is big
In the N number of statistic unit for setting the 3rd threshold value, M is the integer more than or equal to 1, and N is the integer less than or equal to M；
The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, determines
Go out the minimum range of each statistic unit and character pixel point set in N number of statistic unit, and from N number of statistic unit
In, determine that minimum range is less than P statistic unit of the 4th threshold value of setting, wherein, the minimum range is in statistic unit
Each pixel and the character pixel point set in each pixel at least one change in coordinate axis direction away from
Minimum value from；The character pixel point set is the pixel point set of any feature of the fixed target vehicle；
The pixel that will be included in the P statistic unit, is determined as target vehicle described in the benchmark image and includes
Pixel.
Optionally, the 3rd determining module is additionally operable to：
The pixel that target vehicle includes according to the benchmark image, determines to include institute in the benchmark image
State the second bounding box of whole pixels of target vehicle.
A kind of method and device of vehicle detection provided in an embodiment of the present invention, using Binocular Stereo Vision System, obtains
Image pair, chooses an image as benchmark image from described image centering, according to the benchmark image, target vehicle is carried out
Preliminary detection, determines the first bounding box of the Partial Feature comprising the target vehicle, according to institute in the benchmark image
Image pair is stated, determines the D coordinates value for each pixel that the image in first bounding box includes, according to determining
D coordinates value, determine the pixel that target vehicle includes described in the benchmark image.Due to being determined in benchmark image
The pixel that target vehicle contains, therefore improve vehicle detection precision.
It should be understood by those skilled in the art that, the embodiment of the present invention can be provided as method, system or computer program
Product.Therefore, the present invention can use the reality in terms of complete hardware embodiment, complete software embodiment or combination software and hardware
Apply the form of example.Moreover, the present invention can use the computer for wherein including computer usable program code in one or more
The computer program production that usable storage medium is implemented on (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
The form of product.
The present invention be with reference to according to the method for the embodiment of the present invention, the flow of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that it can be realized by computer program instructions every first-class in flowchart and/or the block diagram
The combination of flow and/or square frame in journey and/or square frame and flowchart and/or the block diagram.These computer programs can be provided
The processors of all-purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices is instructed to produce
A raw machine so that the instruction performed by computer or the processor of other programmable data processing devices, which produces, to be used in fact
The device for the function of being specified in present one flow of flow chart or one square frame of multiple flows and/or block diagram or multiple square frames.
These computer program instructions, which may also be stored in, can guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works so that the instruction being stored in the computer-readable memory, which produces, to be included referring to
Make the manufacture of device, the command device realize in one flow of flow chart or multiple flows and/or one square frame of block diagram or
The function of being specified in multiple square frames.
These computer program instructions can be also loaded into computer or other programmable data processing devices so that counted
Series of operation steps is performed on calculation machine or other programmable devices to produce computer implemented processing, thus in computer or
The instruction performed on other programmable devices is provided and is used for realization in one flow of flow chart or multiple flows and/or block diagram one
The step of function of being specified in a square frame or multiple square frames.
Although preferred embodiments of the present invention have been described, but those skilled in the art once know basic creation
Property concept, then can make these embodiments other change and modification.So appended claims be intended to be construed to include it is excellent
Select embodiment and fall into all change and modification of the scope of the invention.
Obviously, various changes and modifications can be made to the invention without departing from essence of the invention by those skilled in the art
God and scope.In this way, if these modifications and changes of the present invention belongs to the scope of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to comprising including these modification and variations.
Claims (10)
- A kind of 1. method of vehicle detection, it is characterised in that this method includes：Using Binocular Stereo Vision System, image pair is obtained；An image is chosen as benchmark image from described image centering, and according to the benchmark image, target vehicle is carried out just Step detection, determines the first bounding box of the Partial Feature comprising the target vehicle in the benchmark image；According to described image pair, the D coordinates value for each pixel that the image in first bounding box includes is determined；According to the D coordinates value determined, the pixel that target vehicle includes described in the benchmark image is determined.
- 2. according to the method described in claim 1, it is characterized in that, described use Part- to target vehicle progress Preliminary detection Base vehicle checking methods.
- 3. according to the method described in claim 1, it is characterized in that, the D coordinates value that the basis is determined, determines described The pixel that target vehicle described in benchmark image includes, including：The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, removes y-axis The absolute value of coordinate value on direction is less than the pixel of first threshold.
- 4. according to method according to any one of claims 1 to 3, it is characterised in that the three-dimensional coordinate that the basis is determined Value, determines the pixel that target vehicle includes described in the benchmark image, further includes：For each the first bounding box, at least one change in coordinate axis direction in x-axis and z-axis, M is divided according to setpoint distance A statistic unit, and from the M statistic unit, determine that the total quantity of included pixel is more than the second threshold of setting The ratio of the total quantity for the pixel that value and the total quantity of the pixel included are included with the M statistic unit, which is more than, to be set N number of statistic unit of fixed 3rd threshold value, M are the integer more than or equal to 1, and N is the integer less than or equal to M；The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, determines institute Each statistic unit and the minimum range of character pixel point set in N number of statistic unit are stated, and from N number of statistic unit, Determine that minimum range is less than P statistic unit of the 4th threshold value of setting, wherein, the minimum range is every in statistic unit In distance of each pixel at least one change in coordinate axis direction in a pixel and the character pixel point set Minimum value；The character pixel point set is the pixel point set of any feature of the fixed target vehicle；The pixel that will be included in the P statistic unit, is determined as the picture that target vehicle includes described in the benchmark image Vegetarian refreshments.
- 5. according to the method described in claim 1, it is characterized in that, determine that target vehicle includes described in the benchmark image Pixel after, including：The pixel that target vehicle includes according to the benchmark image, is determined comprising mesh described in the benchmark image Mark the second bounding box of whole pixels of vehicle.
- A kind of 6. device of vehicle detection, it is characterised in that including：Acquisition module, for obtaining image pair；First determining module, for choosing an image as benchmark image from described image centering, according to the benchmark image, Preliminary detection is carried out to target vehicle, first of the Partial Feature comprising the target vehicle is determined in the benchmark image Bounding box；Second determining module, for according to described image pair, determining each picture that the image in first bounding box includes The D coordinates value of vegetarian refreshments；3rd determining module, for according to the D coordinates value determined, determining target vehicle bag described in the benchmark image The pixel contained.
- 7. device according to claim 6, it is characterised in that described that Part- is used to target vehicle progress Preliminary detection Base vehicle checking methods.
- 8. device according to claim 6, it is characterised in that the 3rd determining module is specifically used for：The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, removes y-axis The absolute value of coordinate value on direction is less than the pixel of first threshold.
- 9. the device according to any one of claim 6~8, it is characterised in that the 3rd determining module is specifically used for：For each the first bounding box, at least one change in coordinate axis direction in x-axis and z-axis, M is divided according to setpoint distance A statistic unit, and from the M statistic unit, determine that the total quantity of included pixel is more than the second threshold of setting The ratio of the total quantity for the pixel that value and the total quantity of the pixel included are included with the M statistic unit, which is more than, to be set N number of statistic unit of fixed 3rd threshold value, M are the integer more than or equal to 1, and N is the integer less than or equal to M；The D coordinates value of the pixel included according to the image in the benchmark image in each first bounding box, determines institute Each statistic unit and the minimum range of character pixel point set in N number of statistic unit are stated, and from N number of statistic unit, Determine that minimum range is less than P statistic unit of the 4th threshold value of setting, wherein, the minimum range is every in statistic unit In distance of each pixel at least one change in coordinate axis direction in a pixel and the character pixel point set Minimum value；The character pixel point set is the pixel point set of any feature of the fixed target vehicle；The pixel that will be included in the P statistic unit, is determined as the picture that target vehicle includes described in the benchmark image Vegetarian refreshments.
- 10. device according to claim 6, it is characterised in that the 3rd determining module is additionally operable to：The pixel that target vehicle includes according to the benchmark image, is determined comprising mesh described in the benchmark image Mark the second bounding box of whole pixels of vehicle.
Priority Applications (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610971878.9A CN108022429B (en) | 2016-11-04 | 2016-11-04 | Vehicle detection method and device |
| EP16863627.2A EP3374967B1 (en) | 2015-11-11 | 2016-11-09 | Methods and systems for binocular stereo vision |
| US15/775,800 US10861177B2 (en) | 2015-11-11 | 2016-11-09 | Methods and systems for binocular stereo vision |
| PCT/CN2016/105174 WO2017080451A1 (en) | 2015-11-11 | 2016-11-09 | Methods and systems for binocular stereo vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610971878.9A CN108022429B (en) | 2016-11-04 | 2016-11-04 | Vehicle detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108022429A true CN108022429A (en) | 2018-05-11 |
| CN108022429B CN108022429B (en) | 2021-08-27 |
Family
ID=62084679
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610971878.9A Active CN108022429B (en) | 2015-11-11 | 2016-11-04 | Vehicle detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108022429B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11188776B2 (en) | 2019-10-26 | 2021-11-30 | Genetec Inc. | Automated license plate recognition system and related method |
| US11367267B2 (en) | 2018-02-08 | 2022-06-21 | Genetec Inc. | Systems and methods for locating a retroreflective object in a digital image |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB1217223A (en) * | 1967-05-23 | 1970-12-31 | Cesare Morais | Rear-view optical system |
| CN102521587A (en) * | 2011-11-24 | 2012-06-27 | 华中科技大学 | License plate location method |
| CN102999759A (en) * | 2012-11-07 | 2013-03-27 | 东南大学 | Light stream based vehicle motion state estimating method |
| CN103231708A (en) * | 2013-04-12 | 2013-08-07 | 安徽工业大学 | Intelligent vehicle obstacle avoiding method based on binocular vision |
| CN103985114A (en) * | 2014-03-21 | 2014-08-13 | 南京大学 | Surveillance video person foreground segmentation and classification method |
| CN104167109A (en) * | 2014-07-17 | 2014-11-26 | 浙江宇视科技有限公司 | Detection method and detection apparatus for vehicle position |
| CN105096591A (en) * | 2014-05-14 | 2015-11-25 | 杭州海康威视数字技术股份有限公司 | Intelligent road monitoring system and method |
| CN105469124A (en) * | 2015-11-20 | 2016-04-06 | 厦门雅迅网络股份有限公司 | Traffic sign classification method |
| CN105989572A (en) * | 2015-02-10 | 2016-10-05 | 腾讯科技（深圳）有限公司 | Picture processing method and apparatus thereof |
| CN106022329A (en) * | 2016-05-12 | 2016-10-12 | 福建南威软件有限公司 | Character noise reduction method in character recognition process |
| CN106056086A (en) * | 2016-06-02 | 2016-10-26 | 江苏华通晟云科技有限公司 | Vehicle brand and model identification method based on fast learning framework |
| CN106650708A (en) * | 2017-01-19 | 2017-05-10 | 南京航空航天大学 | Visual detection method and system for automatic driving obstacles |
| CN107315095A (en) * | 2017-06-19 | 2017-11-03 | 哈尔滨工业大学 | Many vehicle automatic speed-measuring methods with illumination adaptability based on Video processing |
| JP6425990B2 (en) * | 2014-12-18 | 2018-11-21 | 東日本旅客鉄道株式会社 | Track shape measurement method |
-
2016
- 2016-11-04 CN CN201610971878.9A patent/CN108022429B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB1217223A (en) * | 1967-05-23 | 1970-12-31 | Cesare Morais | Rear-view optical system |
| CN102521587A (en) * | 2011-11-24 | 2012-06-27 | 华中科技大学 | License plate location method |
| CN102999759A (en) * | 2012-11-07 | 2013-03-27 | 东南大学 | Light stream based vehicle motion state estimating method |
| CN103231708A (en) * | 2013-04-12 | 2013-08-07 | 安徽工业大学 | Intelligent vehicle obstacle avoiding method based on binocular vision |
| CN103985114A (en) * | 2014-03-21 | 2014-08-13 | 南京大学 | Surveillance video person foreground segmentation and classification method |
| CN105096591A (en) * | 2014-05-14 | 2015-11-25 | 杭州海康威视数字技术股份有限公司 | Intelligent road monitoring system and method |
| CN104167109A (en) * | 2014-07-17 | 2014-11-26 | 浙江宇视科技有限公司 | Detection method and detection apparatus for vehicle position |
| JP6425990B2 (en) * | 2014-12-18 | 2018-11-21 | 東日本旅客鉄道株式会社 | Track shape measurement method |
| CN105989572A (en) * | 2015-02-10 | 2016-10-05 | 腾讯科技（深圳）有限公司 | Picture processing method and apparatus thereof |
| CN105469124A (en) * | 2015-11-20 | 2016-04-06 | 厦门雅迅网络股份有限公司 | Traffic sign classification method |
| CN106022329A (en) * | 2016-05-12 | 2016-10-12 | 福建南威软件有限公司 | Character noise reduction method in character recognition process |
| CN106056086A (en) * | 2016-06-02 | 2016-10-26 | 江苏华通晟云科技有限公司 | Vehicle brand and model identification method based on fast learning framework |
| CN106650708A (en) * | 2017-01-19 | 2017-05-10 | 南京航空航天大学 | Visual detection method and system for automatic driving obstacles |
| CN107315095A (en) * | 2017-06-19 | 2017-11-03 | 哈尔滨工业大学 | Many vehicle automatic speed-measuring methods with illumination adaptability based on Video processing |
Non-Patent Citations (2)
| Title |
|---|
| 孔婷: "基于双目立体视觉的集装箱卡车定位方法研究", 《计算机应用与软件》 * |
| 王荣本: "基于分形盒子维数的车辆定位和识别方法", 《吉林大学学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11367267B2 (en) | 2018-02-08 | 2022-06-21 | Genetec Inc. | Systems and methods for locating a retroreflective object in a digital image |
| US11830256B2 (en) | 2018-02-08 | 2023-11-28 | Genetec Inc. | Systems and methods for locating a retroreflective object in a digital image |
| US11188776B2 (en) | 2019-10-26 | 2021-11-30 | Genetec Inc. | Automated license plate recognition system and related method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108022429B (en) | 2021-08-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106462968B (en) | Method and device for calibrating a camera system of a motor vehicle | |
| CN106056570B (en) | Road gradient detection device | |
| CN102646343B (en) | Vehicle detection apparatus | |
| CN110203210A (en) | A kind of lane departure warning method, terminal device and storage medium | |
| US8406472B2 (en) | Method and system for processing image data | |
| CN113421289B (en) | High-precision vehicle track data extraction method for overcoming unmanned aerial vehicle shooting disturbance | |
| CN108805934A (en) | A kind of method for calibrating external parameters and device of vehicle-mounted vidicon | |
| DE102013211930A1 (en) | Binocular broad base line object comparison method using a minimal cost flownet | |
| CN108073943A (en) | Image data extraction device and image data extraction method | |
| CN111105452B (en) | Binocular vision-based high-low resolution fusion stereo matching method | |
| CN108072385A (en) | Space coordinates localization method, device and the electronic equipment of mobile target | |
| WO2018030010A1 (en) | Road surface estimation device, vehicle control device, road surface estimation method, and program | |
| CN112037159A (en) | Cross-camera road space fusion and vehicle target detection tracking method and system | |
| CN111539484A (en) | Method and device for training neural network | |
| CN109074653A (en) | The method of the object beside road for detecting motor vehicles calculates equipment, driver assistance system and motor vehicles | |
| CN107451976A (en) | A kind of image processing method and device | |
| CN112489106A (en) | Video-based vehicle size measuring method and device, terminal and storage medium | |
| CN110619674A (en) | Three-dimensional augmented reality equipment and method for accident and alarm scene restoration | |
| CN112614192A (en) | On-line calibration method of vehicle-mounted camera and vehicle-mounted information entertainment system | |
| CN109115232B (en) | Navigation method and device | |
| CN108022429A (en) | A kind of method and device of vehicle detection | |
| CN108256470A (en) | A kind of lane shift judgment method and automobile | |
| CN111476798A (en) | Vehicle space form recognition method and system based on contour constraint | |
| CN109300154A (en) | A kind of distance measuring method and device based on binocular solid | |
| CN111881752B (en) | Guardrail detection classification method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |