CN102156860A - Method and device for detecting vehicle - Google Patents
Method and device for detecting vehicle Download PDFInfo
- Publication number
- CN102156860A CN102156860A CN 201110103886 CN201110103886A CN102156860A CN 102156860 A CN102156860 A CN 102156860A CN 201110103886 CN201110103886 CN 201110103886 CN 201110103886 A CN201110103886 A CN 201110103886A CN 102156860 A CN102156860 A CN 102156860A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- characteristic attribute
- local binary
- binary pattern
- numerical value
- 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.)
- Pending
Links
Images
Landscapes
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention relates to the technical field of image detection, in particular to a method and a device for detecting a vehicle. The method provided by the invention comprises the following steps of: scanning an acquired interest region of a current image frame; dividing an interest region image which is obtained by scanning into a plurality of detection regions, and extracting a local binary patterns (LBP) characteristic attribute of each detection region respectively; calculating dot products for a vector set which corresponds to the LBP characteristic attribute of each detection region and a pre-set hyper-plane respectively to obtain a to-be-detected value which corresponds to each detection region; and determining that the vehicle exists in the current image frame if a maximum value in the to-be-detected values which correspond to the detection regions is greater than a pre-set threshold value. The device provided by the invention comprises a scanning module, a characteristic attribute extraction and determination module, a to-be-detected value acquisition module and a vehicle determination module. By the method and the device provided by the invention, a real road condition can be detected conveniently and correctly, and vehicle traffic management can be performed conveniently.
Description
Technical field
The present invention relates to technical field of image detection, particularly a kind of vehicle checking method and device.
Background technology
Along with The development in society and economy, automobile is played the part of an indispensable role in people's daily life, yet automobile quantity is increasing, also brings huge challenge to urban traffic control.Intelligent transportation has become the important means of urban traffic control nowadays, and the important techniques basis that vehicle detection detects as the vehicle monitoring in the intelligent transportation, wagon flow statistics, vehicle peccancy, for high-definition electronic police system, path ambiguity system etc. at present the important management system of intelligent transportation field basic guarantee is provided, its detection efficiency is directly connected to the performance of vehicle snapshot and car plate identification.
Traditional vehicle detection means are to bury ground induction coil below the road surface underground, and through out-of-date, the inductance value of coil changes as vehicle, make the electric signal of output also change thereupon, form vehicle detection signal, and then obtain transport information such as vehicle flowrate.In actual applications, this method construction maintenance expense is higher, also wants road pavement to destroy, and the cabling complexity is unfavorable for anti-tampering and lightning protection, also exists detection information not comprehensive simultaneously, is subject to shortcomings such as external environment influence.
In order to replace this vehicle checking method, in recent years, to release a kind of video that utilizes on the market and trigger the technology that replaces coil to trigger, majority adopts the video detecting method based on car plate to come positioned vehicle.This just makes and can't realize for the vehicle detection that does not have car plate, and does not have the vehicle of car plate to need the part of processing emphatically just in management systems such as vehicle peccancy detections, path ambiguity.
Summary of the invention
A kind of vehicle checking method and device that the embodiment of the invention provides can make things convenient for, detect actual road conditions accurately, are convenient to vehicular traffic control.
A kind of vehicle checking method that the embodiment of the invention provides comprises:
Region-of-interest to the current image frame of gathering scans;
The region-of-interest image division that scanning is obtained is a plurality of surveyed areas, extracts the local binary pattern characteristic attribute of each surveyed area respectively;
The vector set of the local binary pattern characteristic attribute correspondence of each surveyed area is calculated dot product with the lineoid that presets respectively, obtain the measured value to be checked of each surveyed area correspondence;
If greater than preset threshold value, there is vehicle in the maximal value in the measured value to be checked of each surveyed area correspondence in the then described current image frame.
Accordingly, the embodiment of the invention also provides a kind of vehicle detection apparatus, comprising:
Scan module is used for the region-of-interest of the current image frame of gathering is scanned;
The characteristic attribute extraction module, the region-of-interest image division that is used for that scanning is obtained is a plurality of surveyed areas, extracts the local binary pattern characteristic attribute of each surveyed area respectively;
Measured value acquisition module to be checked is used for the vector set of the local binary pattern characteristic attribute correspondence of each surveyed area is calculated dot product with the lineoid that presets respectively, obtains the measured value to be checked of each surveyed area correspondence;
Whether vehicle determination module, the measured value to be checked that is used for judging described a plurality of measured value numerical value maximums to be checked greater than preset threshold value, when selected measured value to be checked during greater than described preset threshold value, determines to have vehicle in the described current image frame.
The vehicle checking method and the device that use the embodiment of the invention to provide, by obtaining the local binary pattern characteristic attribute of local binary pattern, and the lineoid that obtains in conjunction with the training of linear support vector machine, determine whether there is vehicle in the current image frame, thereby remedied the situation of vehicle omission in the prior art, and reduced the rate of false alarm in the vehicle monitoring, for electronic police project, path ambiguity project etc. provides strong basis.
Description of drawings
Fig. 1 is a vehicle checking method schematic flow sheet of the present invention;
Fig. 2 is the schematic flow sheet of basic LBP characteristic attribute in the vehicle checking method of the present invention;
Fig. 3 is a schematic flow sheet of determining lineoid in the vehicle checking method of the present invention;
Fig. 4 is the schematic flow sheet that extracts the LBP characteristic attribute in the specific embodiment of the invention in the vehicle checking method;
Fig. 5 is a vehicle detection apparatus structural representation in the embodiment of the invention.
Embodiment
Below in conjunction with Figure of description the embodiment of the invention is described in further detail.
In order to solve the problem that prior art exists, the embodiment of the invention provides a kind of vehicle checking method, as shown in Figure 1, may further comprise the steps:
Particularly, behind the camera acquisition image, the region-of-interest of the current image frame chosen is scanned.Wherein, can determine the position of region-of-interest according to number of lanes, the factors such as position of generalized case under body in picture frame at real road crossing.
Particularly, extract the LBP binary sequence of each pixel of image in each surveyed area respectively; The LBP binary sequence of each pixel is converted into decimal system numerical value as the LBP characteristic attribute; Filter out the decimal system numerical value of the basic LBP characteristic attribute that belongs to predetermined; In belonging to the decimal system numerical value of predetermined basic LBP characteristic attribute, obtain the maximum decimal system numerical value of occurrence number and its occurrence number; To belong to number of times that the decimal system numerical value of described predetermined basic LBP characteristic attribute occurs, with the maximum decimal system numerical value of described occurrence number and its occurrence number as vector element, form initial array vector; Initial array vector is carried out standardization, obtain the array vector of local binary pattern LBP characteristic attribute correspondence.
Preferable situation, vector element in the above-mentioned initial array vector, also comprise: after the LBP binary sequence of each pixel of image is converted into decimal system numerical value in the surveyed area, do not belong to the number of times summation of the decimal system numerical value appearance of described basic LBP characteristic attribute in all decimal system numerical value.Can emphasize the otherness between vehicle image and the non-vehicle image so more.
Wherein, can obtain predetermined basic LBP characteristic attribute according to the picture frame that presets with vehicle; And, the conventional textural characteristics of the corresponding vehicle of basic LBP characteristic attribute that this is predetermined; Conventional textural characteristics generally comprises: the edge textural characteristics of the straight line textural characteristics of vehicle, the turning textural characteristics of vehicle, vehicle or the planar grains feature of vehicle etc.
Wherein, can carry out standardization to the initial array vector of predetermined basic LBP characteristic attribute correspondence; Use training classifier then, the array vector that obtains after the described standardization is trained, obtain the above-mentioned lineoid that presets.
This preset threshold value is an empirical value, and generally under the well-lighted condition of vehicle intersection, this value is 0; If vehicle intersection light is inadequate, cause the image definition of gathering relatively poor easily, this moment can be according to actual conditions and this threshold value of experience adjustments.
When the measured value of choosing to be checked during, determine to exist in the current image frame vehicle greater than preset threshold value.When measured value to be checked is not more than preset threshold value, determine not exist in the current image frame vehicle, detect next picture frame of current image frame.Preferable, after having vehicle in definite current image frame, can trigger relevant otherwise vehicle detection such as identification of vehicle tracking and/or car plate and/or vehicle peccancy detection.
The vehicle checking method that the embodiment of the invention provides is mainly used in each crossing of urban road, by gathering the crossing image, determine its LBP characteristic attribute, and calculate dot product with the lineoid that presets, the measured value to be checked and the preset threshold value that obtain are compared, thereby determine whether current crossing exists vehicle.
, be elaborated to vehicle checking method provided by the invention below by specific embodiment, specifically may further comprise the steps:
Before the vehicle image frame is carried out vehicle detection, also need to pre-determine basic LBP characteristic attribute and lineoid.
At first, determine basic LBP characteristic attribute, detailed process comprises as shown in Figure 2:
Particularly, the region-of-interest of this picture frame can be one or more, can determine the position of region-of-interest according to number of lanes, the factors such as position of generalized case under body in picture frame at real road crossing.For example, in picture frame, be start line with the position of the tailstock, the vehicle body medium line place straight line of vehicle is a terminated line, and the zone between this start line and the terminated line is made as region-of-interest.
Particularly, can be a plurality of surveyed areas with this region-of-interest image division according to pre-defined rule, can be whole region-of-interest after these a plurality of surveyed areas combinations, also can be the part region-of-interest.This pre-defined rule can be set according to actual needs by the user, for example is 3 * 4 zone with whole area dividing, promptly laterally is divided into 3, vertically is divided into 4, perhaps can divide surveyed area according to the preset detection region template.
Particularly, with each gray values of pixel points of image in the surveyed area and around it all gray values of pixel points in symmetric neighborhood compare, if some gray values of pixel points are greater than the gray values of pixel points that is positioned at the center in the neighborhood, then the LBP value with this pixel in the neighborhood is changed to 1; Otherwise the LBP value is changed to 0.Generally, get the neighborhood that radius is 1 pixel, like this in the neighborhood on every side of central pixel point, have 8 pixels, the LBP value of each pixel is changed to 1 or 0, from the neighbor pixel of the front-right of this central pixel point, according to clockwise direction, obtain the value of each neighbor pixel successively, produce one 8 binary sequence thus.Adopt in the same way equal corresponding one 8 binary sequence of each pixel in the surveyed area.Certainly, can choose the size of symmetric neighborhood arbitrarily, for example choose with the central pixel point is the center, radius is that n(n is a positive integer) symmetric neighborhood of pixel, determine m number pixel (m is the positive integer corresponding with the n logic) thus, determine the LBP value of each pixel again, produce a m position binary sequence.Need to prove, can be initial pixel with arbitrary neighbor pixel when specifically obtaining binary sequence, and according to obtaining in proper order clockwise or counterclockwise, the mode of still obtaining the binary sequence of each pixel must be consistent.
After above-mentioned binary sequence is converted into decimal system numerical value, wherein, learn according to statistics, wherein the number of times that occurs of the decimal system numerical value of the AD HOC of some and, greater than 90% of all decimal system numerical value quantity summations, then the decimal system numerical value with the AD HOC of this some is defined as basic LBP characteristic attribute.Basic process is as follows, add up the number of times of appearance of the AD HOC of described some, and the decimal system numerical value and the occurrence number thereof of maximum LBP correspondences therefrom extract to appear, the decimal system numerical value of the LBP correspondence that occurrence number is maximum in the number of times that the corresponding numerical value of pattern of the AD HOC of some is occurred, the AD HOC of described some and occurrence number thereof are as LBP characteristic attribute substantially.Preferably, basic LBP characteristic attribute also comprises the number of times summation that the decimal system numerical value of the LBP correspondence that does not belong to described AD HOC occurs.For example, use the integer between 0 ~ 255 to represent 256 kinds of characteristic attributes, choose the number of times (corresponding 58 numerical value) of the wherein decimal system numerical value appearance of 58 kinds of AD HOC, and extract in these 58 kinds of AD HOC decimal system numerical value and occurrence number (corresponding 2 numerical value) thereof that maximum LBP occurs, calculate the number of times summation (corresponding 1 numerical value) that the decimal system numerical value of the LBP correspondence that does not belong to these 58 kinds of AD HOC occurs at last, will above-mentioned 61 numerical value (58+2+1) as LBP characteristic attribute substantially.This basic LBP characteristic attribute is represented the conventional textural characteristics of vehicle; Conventional textural characteristics generally comprises: the edge textural characteristics of the straight line textural characteristics of vehicle, the turning textural characteristics of vehicle, vehicle or the planar grains feature of vehicle etc.
Through after the above-mentioned steps, the basic LBP characteristic attribute that obtains is trained, and then whether be identified for distinguishing be the lineoid of vehicle that concrete steps comprise as shown in Figure 3:
Ordered series of numbers n for a basic LBP characteristic attribute occurrence number of expression, if this ordered series of numbers have a limit absolute value | N|, so with each element in this ordered series of numbers divided by this ultimate value n/|N|, that obtain is a positive and negative number percent sequence n%, like this value of all elements all normalization in positive and negative 1 scope, the absolute value that can prevent some basic LBP characteristic attributes like this is excessive, and other basic LBP characteristic attributes are exerted an influence.
Closed test is meant that test sample book belongs to the training sample set.By closed test, determine penalty coefficient, so that adjust of the influence of certain training sample to whole training sample.For example, above-mentioned the 26th kind of attribute, the 35th kind of attribute, the 107th kind of attribute and the 203rd kind of attribute are divided into two equal portions, carry out features training, determine penalty coefficient.
Particularly, can select SVM(support vector machine, support vector machine) as training classifier, this SVM is by a Nonlinear Mapping, in the feature space of training sample spatial mappings to a higher-dimension, make non-linear problem of dividing in original training sample space, be converted into the problem of the linear separability in feature space.Briefly, carry out rising the peacekeeping linearization exactly.Rise dimension and be meant that a training sample does mapping to higher dimensional space, concerning problems such as classification, recurrence, probably the low-dimensional sample space can't linear process training sample set, in high-dimensional feature space, but can realize linear division by a linear lineoid.The general dimension that rises all can be brought the complicated of calculating, and the SVM method has solved this difficult problem dexterously, and it uses the expansion theorem of kernel function, does not just need to know the explicit expression of Nonlinear Mapping.Select different kernel functions, can generate different SVM, kernel function commonly used has following 4 kinds:
(1) linear kernel function K (x, y)=xy;
(2) polynomial kernel function K (x, y)=[(xy)+1] d;
(3) radial basis function K (x, y)=exp (| x-y|^2/d^2);
(4) two layers of neural network kernel function (Sigmoid kernel function) K (x, y)=tanh (a (xy)+b).
Because linear kernel function is simple in structure, computation complexity is low, and then saves computing time, therefore under the condition that detects real-time road, selects for use linear kernel function comparatively suitable.
Wherein, lineoid is the linear subspaces that the codimension degree equals in the n dimension Euclidean space, by LBP characteristic attribute and this lineoid are calculated dot product, compare with preset threshold value again, can in the component-bar chart picture frame whether vehicle be arranged, even result of calculation is greater than threshold value, and then there is vehicle in the image in the picture frame, otherwise, do not have vehicle.
Through definite process of above-mentioned basic LBP characteristic attribute and lineoid, ready for vehicle detection work, concrete testing process is as follows:
The method of using the embodiment of the invention to provide in practice, according to shown in Figure 1, after the road cross images acquired, the specific implementation step is as follows:
Step 101 is divided the region-of-interest of the current image frame of gathering.
Particularly, behind the camera acquisition image, the region-of-interest of the current image frame chosen is scanned.Wherein, can determine the position of region-of-interest according to number of lanes, the factors such as position of generalized case under body in picture frame at real road crossing.For example, in picture frame, be start line with the position of the tailstock, the vehicle body medium line place straight line of vehicle is a terminated line, and the zone between this start line and the terminated line is made as region-of-interest.
The detailed process of this step comprises as shown in Figure 4:
Can be a plurality of surveyed areas with this region-of-interest image division according to pre-defined rule, can be whole region-of-interest after these a plurality of surveyed areas combinations, also can be the part region-of-interest.This pre-defined rule can be set according to actual needs by the user, for example is 3 * 4 zone with whole area dividing, promptly laterally is divided into 3, vertically is divided into 4, perhaps can divide surveyed area according to the preset detection region template.
Particularly, with each gray values of pixel points of image in the surveyed area and around it all gray values of pixel points in symmetric neighborhood compare, if some gray values of pixel points are greater than the gray values of pixel points that is positioned at the center in the neighborhood, then the LBP value with this pixel in the neighborhood is changed to 1; Otherwise the LBP value is changed to 0.Neighbor pixel from the front-right of this central pixel point according to clockwise direction, obtains the value of each neighbor pixel successively, produces one 8 binary sequence thus.Adopt in the same way equal corresponding one 8 binary sequence of each pixel in the surveyed area.Need to prove, can be initial pixel with arbitrary neighbor pixel when specifically obtaining binary sequence, and according to obtaining in proper order clockwise or counterclockwise, the mode of still obtaining the binary sequence of each pixel must be consistent.
Particularly, the decimal system numerical value according to basic LBP characteristic attribute correspondence screens the decimal system numerical value that the LBP binary sequence transforms.For example, if a numerical value in the decimal system numerical value that the LBP binary sequence transforms equals one in the numerical value of basic LBP characteristic attribute, then this decimal value is kept, as a kind of LBP characteristic attribute.
Preferable situation, above-mentioned initial array vector also comprises: after the LBP binary sequence of each pixel of image is converted into decimal system numerical value in the surveyed area, do not belong to the number of times summation of the decimal system numerical value appearance of basic LBP characteristic attribute.
The process of the initial vector collection being carried out standardization is identical with set of eigenvectors course of standardization process in the basic LBP characteristic attribute acquisition process, does not repeat them here.
Step 103 is calculated dot product with the lineoid that presets respectively with the vector set of the LBP characteristic attribute correspondence of each surveyed area, obtains the measured value to be checked of each surveyed area correspondence;
Step 104 is if greater than preset threshold value, there is vehicle in the maximal value in the measured value to be checked of each surveyed area correspondence in the then described current image frame.
Wherein, this preset threshold value is an empirical value, and generally under the well-lighted condition of vehicle intersection, this value is 0; If vehicle intersection light is inadequate, cause the image definition of gathering relatively poor easily, this moment can be according to actual conditions and this threshold value of experience adjustments.
Vehicle checking method of the present invention also comprises after the maximal value in the measured value to be checked of definite each surveyed area correspondence is greater than preset threshold value:
Step 105, the testing process of associated vehicles such as triggering vehicle tracking and/or car plate identification and/or vehicle peccancy detection.
If the maximal value in the measured value to be checked of each surveyed area correspondence is not more than preset threshold value, then determines not have vehicle in the described current image frame, and continue to detect next picture frame of current image frame.
By foregoing description as can be known, the vehicle checking method that the application of the invention embodiment provides, by obtaining local binary pattern LBP characteristic attribute, and the lineoid that obtains in conjunction with the training of linear support vector machine, determine whether there is vehicle in the current image frame, thereby remedied the situation of vehicle omission in the prior art, and reduced the rate of false alarm in the vehicle monitoring, for electronic police project, path ambiguity project etc. provides strong basis.
Based on same inventive concept, a kind of vehicle detection apparatus also is provided in the embodiment of the invention, the principle of this device solves problem is similar to vehicle checking method, as shown in Figure 5, comprising:
Scan module 501 is used for the region-of-interest of the current image frame of gathering is scanned;
Characteristic attribute extraction module 502, the region-of-interest image division that is used for that scanning is obtained is a plurality of surveyed areas, extracts the LBP characteristic attribute of each surveyed area respectively;
Measured value acquisition module 503 to be checked is used for the vector set of the LBP characteristic attribute correspondence of each surveyed area is calculated dot product with the lineoid that presets respectively, obtains the measured value to be checked of each surveyed area correspondence;
Whether vehicle determination module 504, the measured value to be checked that is used for judging described a plurality of measured value numerical value maximums to be checked greater than preset threshold value, when selected measured value to be checked during greater than described preset threshold value, determines to have vehicle in the described current image frame.
Preferable, characteristic attribute extraction module 502 specifically is used for extracting respectively the LBP binary sequence of each pixel of each surveyed area image; The LBP binary sequence of described each pixel is converted into decimal system numerical value; From the decimal system numerical value that the LBP binary sequence transforms, filter out the decimal system numerical value that belongs to basic LBP characteristic attribute; In the described decimal system numerical value that belongs to basic LBP characteristic attribute that filters out, obtain the maximum decimal system numerical value of occurrence number and its occurrence number; The decimal system numerical value that number of times, the described occurrence number that the decimal system numerical value that belongs to basic LBP characteristic attribute that filters out is occurred is maximum and its occurrence number are formed the initial vector collection as vector element; Described initial vector collection is carried out standardization, obtain the LBP characteristic attribute.
Preferably, described LBP characteristic attribute also comprises the number of times summation of the decimal system numerical value appearance that does not belong to basic LBP characteristic attribute.
Preferably, this device also comprises trigger module 505, is used for when described vehicle determination module determines that there is vehicle in current image frame, triggers identification of vehicle tracking and/or car plate and/or vehicle peccancy and detects.
By foregoing description as can be known, the method and apparatus that uses the embodiment of the invention to provide, by obtaining local binary pattern LBP characteristic attribute, and the lineoid that obtains in conjunction with the training of linear support vector machine, determine whether there is vehicle in the current image frame, thereby remedied the situation of vehicle omission in the prior art, and reduced the rate of false alarm in the vehicle monitoring, for electronic police project, path ambiguity project etc. provides strong basis.
Those skilled in the art should understand that embodiments of the invention can be provided as method, system or computer program.Therefore, the present invention can adopt complete hardware embodiment, complete software implementation example or in conjunction with the form of the embodiment of software and hardware aspect.And the present invention can adopt the form that goes up the computer program of implementing in one or more computer-usable storage medium (including but not limited to magnetic disk memory, CD-ROM, optical memory etc.) that wherein include computer usable program code.
The present invention is that reference is described according to the process flow diagram and/or the block scheme of method, equipment (system) and the computer program of the embodiment of the invention.Should understand can be by the flow process in each flow process in computer program instructions realization flow figure and/or the block scheme and/or square frame and process flow diagram and/or the block scheme and/or the combination of square frame.Can provide these computer program instructions to the processor of multi-purpose computer, special purpose computer, Embedded Processor or other programmable data processing device to produce a machine, make the instruction of carrying out by the processor of computing machine or other programmable data processing device produce to be used for the device of the function that is implemented in flow process of process flow diagram or a plurality of flow process and/or square frame of block scheme or a plurality of square frame appointments.
These computer program instructions also can be stored in energy vectoring computer or the computer-readable memory of other programmable data processing device with ad hoc fashion work, make the instruction that is stored in this computer-readable memory produce the manufacture that comprises command device, this command device is implemented in the function of appointment in flow process of process flow diagram or a plurality of flow process and/or square frame of block scheme or a plurality of square frame.
These computer program instructions also can be loaded on computing machine or other programmable data processing device, make on computing machine or other programmable devices and to carry out the sequence of operations step producing computer implemented processing, thereby the instruction of carrying out on computing machine or other programmable devices is provided for being implemented in the step of the function of appointment in flow process of process flow diagram or a plurality of flow process and/or square frame of block scheme or a plurality of square frame.
Although described the preferred embodiments of the present invention, in a single day those skilled in the art get the basic creative notion of cicada, then can make other change and modification to these embodiment.So claims are intended to all changes and the modification that are interpreted as comprising preferred embodiment and fall into the scope of the invention.
Obviously, those skilled in the art can carry out various changes and modification to the present invention and not break away from the spirit and scope of the present invention.Like this, if of the present invention these are revised and modification belongs within the scope of claim of the present invention and equivalent technologies thereof, then the present invention also is intended to comprise these changes and modification interior.
Claims (13)
1. a vehicle checking method is characterized in that, this method comprises:
Region-of-interest to the current image frame of gathering scans;
The region-of-interest image division that scanning is obtained is a plurality of surveyed areas, extracts the local binary pattern characteristic attribute of each surveyed area respectively;
The vector set of the local binary pattern characteristic attribute correspondence of each surveyed area is calculated dot product with the lineoid that presets respectively, obtain the measured value to be checked of each surveyed area correspondence;
If greater than preset threshold value, there is vehicle in the maximal value in the measured value to be checked of each surveyed area correspondence in the then described current image frame.
2. vehicle checking method according to claim 1 is characterized in that, before the region-of-interest to the current image frame of gathering scans, also comprises step: determine basic local binary pattern characteristic attribute and definite lineoid.
3. vehicle checking method according to claim 2 is characterized in that, described definite basic local binary pattern characteristic attribute comprises:
Region-of-interest to the picture frame with vehicle that presets scans;
The region-of-interest image division that scanning is obtained is a plurality of surveyed areas;
Extract the local binary pattern binary sequence of each surveyed area respectively;
The local binary pattern binary sequence of described each pixel is converted into decimal system numerical value;
The decimal system numerical value that extracts AD HOC wherein is as basic local binary pattern characteristic attribute.
4. vehicle checking method according to claim 3 is characterized in that, extracting wherein, the decimal system numerical value of AD HOC comprises as basic local binary pattern characteristic attribute:
The decimal system numerical value and the occurrence number thereof of the local binary pattern correspondence that occurrence number is maximum in the number of times that the corresponding decimal system numerical value of the pattern of the AD HOC of some is occurred, the AD HOC of described some.
5. vehicle checking method according to claim 4 is characterized in that, the characteristic attribute of described basic local binary pattern also comprises:
Do not belong to the number of times summation that the decimal system numerical value of the local binary pattern correspondence of described AD HOC occurs.
6. vehicle checking method according to claim 4 is characterized in that,
Described predetermined basic local binary pattern characteristic attribute is represented the conventional textural characteristics of vehicle;
Described conventional textural characteristics comprises: the edge textural characteristics of the straight line textural characteristics of vehicle, the turning textural characteristics of vehicle, vehicle, the planar grains feature of vehicle.
7. vehicle checking method according to claim 3, it is characterized in that the vector set to described basic local binary pattern characteristic attribute correspondence carries out standardization, the vector set that obtains after the described standardization is trained, obtain the described lineoid that presets.
8. vehicle checking method according to claim 2 is characterized in that, the described local binary pattern characteristic attribute that extracts each surveyed area respectively comprises:
Extract the local binary pattern binary sequence of each pixel of image in each surveyed area respectively;
The local binary pattern binary sequence of described each pixel is converted into decimal system numerical value;
From the decimal system numerical value that local binary pattern binary sequence transforms, filter out the decimal system numerical value that belongs to basic local binary pattern characteristic attribute;
In the described decimal system numerical value that belongs to basic local binary pattern characteristic attribute that filters out, obtain the maximum decimal system numerical value of occurrence number and its occurrence number;
Decimal system numerical value that number of times, the described occurrence number that the decimal system numerical value that belongs to basic local binary pattern characteristic attribute that filters out is occurred is maximum and occurrence number thereof are formed the initial vector collection as vector element;
Described initial vector collection is carried out standardization, obtain local binary pattern characteristic attribute.
9. method as claimed in claim 8 is characterized in that, the vector element in the described initial array vector also comprises:
After the local binary pattern binary sequence of each pixel of image is converted into decimal system numerical value in the surveyed area, do not belong to the number of times summation of the decimal system numerical value appearance of described basic local binary pattern characteristic attribute.
10. vehicle checking method according to claim 1 is characterized in that, when described measured value to be checked is not more than described preset threshold value, determines not exist in the described current image frame vehicle, detects next picture frame of current image frame.
11. vehicle checking method according to claim 1 is characterized in that, determines to exist after the vehicle in the described current image frame, also comprises:
Triggering identification of vehicle tracking and/or car plate and/or vehicle peccancy detects.
12. a vehicle detection apparatus is characterized in that, comprising:
Scan module is used for the region-of-interest of the current image frame of gathering is scanned;
The characteristic attribute extraction module, the region-of-interest image division that is used for that scanning is obtained is a plurality of surveyed areas, extracts the local binary pattern characteristic attribute of each surveyed area respectively;
Measured value acquisition module to be checked is used for the vector set of the local binary pattern characteristic attribute correspondence of each surveyed area is calculated dot product with the lineoid that presets respectively, obtains the measured value to be checked of each surveyed area correspondence;
Whether vehicle determination module, the measured value to be checked that is used for judging described a plurality of measured value numerical value maximums to be checked greater than preset threshold value, when selected measured value to be checked during greater than described preset threshold value, determines to have vehicle in the described current image frame.
13. vehicle detection apparatus according to claim 12 is characterized in that, also comprises:
Trigger module is used for when described vehicle determination module determines that there is vehicle in current image frame, triggers identification of vehicle tracking and/or car plate and/or vehicle peccancy and detects.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110103886 CN102156860A (en) | 2011-04-25 | 2011-04-25 | Method and device for detecting vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110103886 CN102156860A (en) | 2011-04-25 | 2011-04-25 | Method and device for detecting vehicle |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102156860A true CN102156860A (en) | 2011-08-17 |
Family
ID=44438351
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201110103886 Pending CN102156860A (en) | 2011-04-25 | 2011-04-25 | Method and device for detecting vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102156860A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110246134A (en) * | 2019-06-24 | 2019-09-17 | 株洲时代电子技术有限公司 | A kind of rail defects and failures sorter |
| CN110264398A (en) * | 2019-07-16 | 2019-09-20 | 北京市商汤科技开发有限公司 | Image processing method and device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101251927A (en) * | 2008-04-01 | 2008-08-27 | 东南大学 | Vehicle detecting and tracing method based on video technique |
| CN101650783A (en) * | 2008-08-13 | 2010-02-17 | 株式会社Ntt都科摩 | Image identification method and imaging apparatus |
| CN101968924A (en) * | 2010-10-19 | 2011-02-09 | 北方工业大学 | High-resolution image-based real-time traffic flow detection method |
-
2011
- 2011-04-25 CN CN 201110103886 patent/CN102156860A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101251927A (en) * | 2008-04-01 | 2008-08-27 | 东南大学 | Vehicle detecting and tracing method based on video technique |
| CN101650783A (en) * | 2008-08-13 | 2010-02-17 | 株式会社Ntt都科摩 | Image identification method and imaging apparatus |
| CN101968924A (en) * | 2010-10-19 | 2011-02-09 | 北方工业大学 | High-resolution image-based real-time traffic flow detection method |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110246134A (en) * | 2019-06-24 | 2019-09-17 | 株洲时代电子技术有限公司 | A kind of rail defects and failures sorter |
| CN110264398A (en) * | 2019-07-16 | 2019-09-20 | 北京市商汤科技开发有限公司 | Image processing method and device |
| CN110264398B (en) * | 2019-07-16 | 2021-05-28 | 北京市商汤科技开发有限公司 | Image processing method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen | Automatic License Plate Recognition via sliding-window darknet-YOLO deep learning | |
| Amato et al. | Deep learning for decentralized parking lot occupancy detection | |
| CN109711264B (en) | Method and device for detecting occupation of bus lane | |
| Soon et al. | PCANet-based convolutional neural network architecture for a vehicle model recognition system | |
| CN109711437A (en) | A kind of transformer part recognition methods based on YOLO network model | |
| CN101859382A (en) | License plate detection and identification method based on maximum stable extremal region | |
| Yu et al. | Vehicle logo recognition based on bag-of-words | |
| CN103258213A (en) | Vehicle model dynamic identification method used in intelligent transportation system | |
| Chen et al. | Multi-shape descriptor vehicle classification for urban traffic | |
| CN108090459B (en) | Traffic sign detection and identification method suitable for vehicle-mounted vision system | |
| CN103971097A (en) | Vehicle license plate recognition method and system based on multiscale stroke models | |
| CN105740886A (en) | Machine learning based vehicle logo identification method | |
| CN105224945B (en) | A kind of automobile logo identification method based on joint-detection and identification algorithm | |
| Zhang et al. | Vehicle classification with confidence by classified vector quantization | |
| Mao et al. | Rapid vehicle logo region detection based on information theory | |
| CN111915583A (en) | Vehicle and pedestrian detection method based on vehicle-mounted thermal infrared imager in complex scene | |
| Tian et al. | An algorithm combined with color differential models for license-plate location | |
| Thubsaeng et al. | Vehicle logo detection using convolutional neural network and pyramid of histogram of oriented gradients | |
| CN111062319B (en) | Driver call detection method based on active infrared image | |
| CN102156860A (en) | Method and device for detecting vehicle | |
| Vashisth et al. | Histogram of Oriented Gradients based reduced feature for traffic sign recognition | |
| US10026014B2 (en) | Method and apparatus for data set classification based on generator features | |
| Dong et al. | Nighttime pedestrian detection with near infrared using cascaded classifiers | |
| KY et al. | Road pothole detection using yolov2 model | |
| Kim et al. | Stereo-based region of interest generation for real-time pedestrian detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20110817 |