CN106415602B - A kind of detection method and device of pairs of lane line - Google Patents
A kind of detection method and device of pairs of lane line Download PDFInfo
- Publication number
- CN106415602B CN106415602B CN201680000880.XA CN201680000880A CN106415602B CN 106415602 B CN106415602 B CN 106415602B CN 201680000880 A CN201680000880 A CN 201680000880A CN 106415602 B CN106415602 B CN 106415602B
- Authority
- CN
- China
- Prior art keywords
- lane line
- pairs
- artificial neural
- straight lines
- sampling point
- 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.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 24
- 239000013598 vector Substances 0.000 claims abstract description 69
- 238000005070 sampling Methods 0.000 claims abstract description 68
- 230000005284 excitation Effects 0.000 claims abstract description 64
- 238000013528 artificial neural network Methods 0.000 claims abstract description 57
- 238000000034 method Methods 0.000 claims abstract description 22
- 210000002569 neuron Anatomy 0.000 claims description 27
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000001537 neural effect Effects 0.000 claims 1
- 238000006467 substitution reaction Methods 0.000 claims 1
- 238000012549 training Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 7
- 230000008569 process Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Image Analysis (AREA)
Abstract
A kind of detection method of pairs of lane line, which comprises obtain two straight lines to be detected according to the sampling point that preset spacing selects respectively on two straight lines and obtain the distance between the sampling point and scheduled common point vector;The distance vector is substituted into preset artificial neural networks excitation value, wherein the weight vectors of the artificial neural network train acquisition according to pairs of lane line sample data gathered in advance;According to the excitation value that the artificial neural network exports, judge whether two straight lines are pairs of lane line.This method can effectively guarantee the real-time judged pairs of lane line, and can be improved the accuracy of judgement.
Description
Technical field
The invention belongs to automatic Pilot field more particularly to a kind of detection methods and device of pairs of lane line.
Background technique
Lane Departure Warning System be it is a kind of assisted by way of alarm driver reduce automobile sent out because of deviation
The auxiliary system of the car steering of raw traffic accident.When vehicle deviates traveling lane, pass through the Lane Departure Warning System
Early warning prompting can be issued, the early warning prompting may include alarm tone, vibration of steering wheel or automatic change steering etc..
In Lane Departure Warning System, in order to guarantee the accuracy of early warning, need correctly to extract lane line
And identification.Current pairs of method for detecting lane lines generally requires and consumes more system resource, when the higher accuracy of needs
When, then it needs to spend certain calculating time, not can guarantee real-time detection;Alternatively, then may to improve the real-time of detection
Leak detection is caused, false detection rate is caused to improve.
Summary of the invention
The purpose of the present invention is to provide a kind of detection methods of pairs of lane line, to solve the prior art in pairs of lane
When line detects, the problem of cannot effectively guaranteeing accuracy rate and real-time.
In a first aspect, the embodiment of the invention provides a kind of detection methods of pairs of lane line, which comprises
Two straight lines to be detected are obtained, the sample selected respectively on two straight lines according to preset spacing
Point obtains the distance between the sampling point and scheduled common point vector;
The distance vector is substituted into preset artificial neural networks excitation value, wherein the artificial neuron
The weight vectors of network train acquisition according to pairs of lane line sample data gathered in advance;
According to the excitation value that the artificial neural network exports, judge whether two straight lines are pairs of lane line.
With reference to first aspect, described according to preset spacing in the first possible implementation of first aspect
The sampling point selected on two straight lines, obtaining the distance between the sampling point and scheduled common point vector step includes:
According to preset spacing, sampling point is selected respectively on two straight lines;
Using the central point of image as common point, the distance between the sampling point and the common point vector are obtained.
With reference to first aspect, in second of possible implementation of first aspect, described by the distance vector generation
Before entering preset artificial neural networks excitation value step, the method also includes:
Acquisition a large amount of pairs of lane line sample and azygous lane line sample, according to the spacing in the lane
Sampling point is selected on line sample;
Calculate the distance between the sampling point and the common point;
The distance is substituted into the neuronal cell layers of artificial neural network, it is whether pairs of according to sample, calculate artificial neuron
The corresponding weight vectors of the neuronal cell layers of network.
With reference to first aspect, in the third possible implementation of first aspect, the sampling point is included in every lane
The N number of sampling point selected on line, the N are the natural number more than or equal to 2.
With reference to first aspect, described according to the artificial neural network in the 4th kind of possible implementation of first aspect
The excitation value of network output, judges whether two straight lines are that pairs of lane line step includes:
The excitation value for obtaining the artificial neural network output, the excitation value and preset threshold value are compared
Compared with;
If the excitation value is greater than the threshold value, it is determined that two straight lines are pairs of lane line, if described swash
Value is encouraged less than the threshold value, it is determined that two straight lines are not pairs of lane lines.
Second aspect, the embodiment of the invention provides a kind of detection device of pairs of lane line, described device includes:
Lane line acquiring unit, for obtaining two straight lines to be detected, according to preset spacing at described two
The sampling point selected respectively on straight line obtains the distance between the sampling point and scheduled common point vector;
Computing unit, for the distance vector to be substituted into preset artificial neural networks excitation value, wherein
The weight vectors of the artificial neural network train acquisition according to pairs of lane line sample data gathered in advance;
Judging unit, the excitation value for being exported according to the artificial neural network, judge two straight lines whether be
Pairs of lane line.
In conjunction with second aspect, in the first possible implementation of second aspect, the lane line acquiring unit includes:
Sampling point selects subelement, for selecting sampling point respectively on two straight lines according to preset spacing;
Common point obtains subelement, for using the central point of image as common point, obtain the sampling point with it is described public
The distance between point vector.
In conjunction with second aspect, in second of possible implementation of second aspect, described device further include:
Sample collection unit, for acquiring a large amount of pairs of lane line sample and azygous lane line sample, according to
The spacing selects sampling point on the lane line sample;
Metrics calculation unit, for calculating the distance between the sampling point and the common point;
Weight vector computation unit, for the distance to be substituted into the neuronal cell layers of artificial neural network, according to sample
Whether in pairs, the corresponding weight vectors of the neuronal cell layers of artificial neural network are calculated.
In conjunction with second aspect, in the third possible implementation of second aspect, the sampling point is included in every lane
The N number of sampling point selected on line, the N are the natural number more than or equal to 2.
In conjunction with second aspect, in the 4th kind of possible implementation of second aspect, the judging unit includes:
Comparing subunit has, for obtaining the excitation value of the artificial neural network output, by the excitation value and in advance
The threshold value of setting is compared;
Pairs of lane line determines subelement, if being greater than the threshold value for the excitation value, it is determined that described two straight
Line is pairs of lane line, if the excitation value is less than the threshold value, it is determined that two straight lines are not pairs of lane lines.
In the present invention, two straight lines to be detected are obtained to be selected on two straight lines according to preset spacing
Sampling point is selected, the distance between sampling point and common point vector are obtained, the distance vector is substituted into trained artificial neuron in advance
Network can be obtained the excitation value of artificial neural network output, judge whether two straight lines are pairs of according to the excitation value
Lane line.Using the method for the invention, it is only necessary to which the range data that will acquire substitutes into artificial neural network can quickly really
Whether fixed is pairs of lane line, can effectively guarantee the real-time judged pairs of lane line, and can be improved the standard of judgement
True property.
Detailed description of the invention
Fig. 1 is the implementation flow chart of the detection method of pairs of lane line provided in an embodiment of the present invention;
Fig. 2 is that distance vector provided in an embodiment of the present invention obtains example schematic;
Fig. 3 is that another distance vector provided in an embodiment of the present invention obtains example schematic;
Fig. 4 is artificial neural network schematic diagram provided in an embodiment of the present invention;
Fig. 5 is the implementation flow chart of artificial neural network training provided in an embodiment of the present invention;
Fig. 5 a, Fig. 5 b, Fig. 5 c are training sample schematic diagram provided in an embodiment of the present invention;
Fig. 6 is the structural schematic diagram of the detection device of pairs of lane line provided in an embodiment of the present invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
Pairs of method for detecting lane lines described in the embodiment of the present invention, it is therefore intended that overcome in the prior art with regard to pairs of lane line
In detection method, in order to improve the detection accuracy of pairs of lane line, generally require to cause using more complicated detection algorithm
Detection calculating process needs to consume certain duration, if under galloping state, it will cause testing result lag,
The lower defect of the real-time of detection.And if being easy to appear testing result error using simple lane line judgment method,
Influence user's judgement.With reference to the accompanying drawing, the present invention is further illustrated.
Fig. 1 shows the implementation process of the detection method of pairs of lane line provided in an embodiment of the present invention, and details are as follows:
In step s101, two straight lines to be detected are obtained, according to preset spacing on two straight lines
The sampling point selected respectively obtains the distance between the sampling point and scheduled common point vector.
Specifically, referring to the auxiliary line in the lane for limiting vehicle driving at lane line described in the embodiment of the present invention.
Due in vehicle travel process, other than lane line, it is also possible to including other tag lines, than as shown in figure 3, in addition to vehicle
It further include arrow logo other than diatom, the mark being made of arrow line and lane line should not then be identified as pairs of lane line.
Two straight lines to be detected, can be by carrying out identification acquisition to image.For example, the identification of the straight line,
It can be identified according to color in image, for example color is white in identification image or color is the straight line etc. of yellow.
The preset spacing can be set according to the size of image.Such as the width according to image, setting
1/3 screen width is the length of the spacing.It is, of course, also possible to select the spacing according to the number of required sampling point
Size sets the length of the spacing, so that the sampling point of selection includes the end position of the straight line.
The selection of the common point can flexibly be set according to the needs of users.For example top in image can be set
Midpoint as the common point, the midpoint of the lower part in image can also be set as the common point, figure can also be set
The central point of picture is as the common point.According to the difference of the selection mode of common point, corresponding to the artificial neural network
Weight vectors are not also identical.And the position for the common point selected in the training process of weight vectors, and it is described to be detected
The position of the common point of two line correspondences is identical.
The distance between the sampling point and the common point vector, can be by measuring the distance between sampling point and common point
It obtains.Such as two straight lines in Fig. 2, (one of implementation selected for example of two sampling points is selected on every straight line
Mode), the sampling point of selection includes four, and the common point is the center of image, then the distance of upper left corner line segment is
7.7cm, the distance of lower left corner line segment are 10.5cm, and the distance of upper right corner line segment is 8.5cm, and the distance of lower right corner line segment is
12cm, constitute distance vector be<7.7,10.5,8.5,12>.
Two straight lines as shown in Figure 3, two sampling points are selected on every straight line, and (one of which selected for example is real
Apply mode), selected sampling point includes four, and the common point is the center of image, then the distance of upper left corner line segment is
2.7cm, the distance of lower left corner line segment are 8.2cm, and the distance of upper right corner line segment is 6.2cm, and the distance of lower right corner line segment is
4.3cm, constitute distance vector be<2.7,8.2,6.2,4.3>.
In step s 102, the distance vector is substituted into preset artificial neural networks excitation value, wherein
The weight vectors of the artificial neural network train acquisition according to pairs of lane line sample data gathered in advance.
It, can be according to setting in advance specifically, the calculating of the weight vectors of artificial neural network described in the embodiment of the present invention
Fixed multiple sample trainings obtain, wherein the training method of the weight vectors may include following step shown in fig. 5:
In step S501, acquisition a large amount of pairs of lane line sample and azygous lane line sample, according to described
Spacing selects sampling point on the lane line sample;
In step S02, the distance between the sampling point and the common point are calculated;
In step S503, the distance is substituted into the neuronal cell layers of artificial neural network, it is whether pairs of according to sample,
Calculate the corresponding weight vectors of the neuronal cell layers of artificial neural network.
Specifically, the artificial neural network may include neuronal cell layers, output layer.The neuronal cell layers with it is defeated
It can also include hidden layer out between layer.If Fig. 4 is a kind of structural representation of artificial neural network provided in an embodiment of the present invention
Figure.As shown in figure 4, the artificial neural network includes input layer X1, X2, X3 and X4, neuronal cell layers Y1, Y2, Z1 and Z2,
In, Y1 and Y2 constitute hidden layer, and Z1 and Z2 constitute output layer.
Wherein, the input number of the input layer is set according to the number of input vector.Such as in Fig. 5 a, Fig. 5 b and figure
Sampling point in 5c is 4, and the input number of corresponding input layer is also 4.
Assuming that Fig. 5 a, Fig. 5 b and Fig. 5 c are three in a large amount of all of lane in pairs, and for Fig. 5 a, Fig. 5 b and
The distance vector of Fig. 5 c is successively are as follows:
Two straight lines as shown in Figure 5 a select two sampling points (one of which selected for example on every straight line
Embodiment), selected sampling point includes four, and the common point is the center of image, then the distance of upper left corner line segment
For 5cm, the distance of lower left corner line segment is 14cm, and the distance of upper right corner line segment is 11cm, and the distance of lower right corner line segment is 9.5cm,
Constituting distance vector is<5,14,11,9.5>.
Two straight lines as shown in Figure 5 b select two sampling points (one of which selected for example on every straight line
Embodiment), selected sampling point includes four, and the common point is the center of image, then the distance of upper left corner line segment
For 5cm, the distance of lower left corner line segment is 6.5cm, and the distance of upper right corner line segment is 10cm, and the distance of lower right corner line segment is 16cm,
Constituting distance vector is<5,6.5,10,16>.
Two straight lines as shown in Figure 5 c select two sampling points (one of which selected for example on every straight line
Embodiment), selected sampling point includes four, and the common point is the center of image, then the distance of upper left corner line segment
For 7cm, the distance of lower left corner line segment is 11cm, and the distance of upper right corner line segment is 7.7cm, and the distance of lower right corner line segment is 12cm,
Constituting distance vector is<7,11,7.7,12>.
Fig. 5 a, Fig. 5 b, Fig. 5 c output result be " pairs of lane line ", the artificial neural network in conjunction with described in Fig. 4,
W11, W13, W15 and W17 are the corresponding weights of four inputs of nerve cell Y1;W12, W14, W16 and W18 are nerve cell Y2
Four corresponding weights of input;W21 and W23 is the corresponding weight of two inputs of nerve cell Z1;W22 and W24 is nerve
The corresponding weight of two inputs of cell Z2.
Wherein, the excitation value of nerve cell is the sum of products of input with weight, such as in Fig. 4: the excitation value that Y1 is obtained=
X1*W11+X2*W13+X3*W15+X4*W17。
In the training process, excitation function can be set for nerve cell, such as: the excitation value of output layer is more than certain threshold
Value, output 1;Otherwise 0 is exported.In this instance, the excitation function setting of the nerve cell Y1 and Y2 of hidden layer are as follows: output==swashs
Encourage value.
The input of nerve cell Z1 and Z2 are the output of Y1 and Y2, it may be assumed that excitation value=Y1*W21+Y2* that Z1 is obtained
W23.Wherein, the present invention can be with for Z1 and Z2 setting excitation function are as follows: if (excitation value >=0), 1, otherwise of output is defeated
Out 0.
According to above-mentioned excitation function, each weight of artificial neural network, the i.e. training of artificial neural network can be determined.
In the present invention, each weighted value of artificial neural network is initialized as any random decimal between [- 1,1], so
Afterwards the sample in training set is inputted artificial neural network one by one, adjust each weighted value, makes to own " just " sample in Z1 generation 1
Output, Z2 generate 0 output;And all " negative samples " generates 0 output in Z1, generates 1 output in Z2.By Fig. 5 a, figure
Three vectors in 5b and Fig. 5 c are used as positive sample to input one by one, repetition training, adjust weighted value.It is, of course, also possible to input negative
Sample is trained.Finally obtain nerve cell Y1 weight vectors be<0.8, -0.2,0.65, -0.3>, the input weight of Y2
Vector is<0.7, -0.3,0.9, -0.4>, the input weight vector of Z1 is<1, -1>, the input weight vector of Z2 is<-1,1>.
In step s 103, according to the artificial neural network export excitation value, judge two straight lines whether be
Pairs of lane line.
The distance vector obtained according to fig. 2 with Fig. 3 substitutes into the artificial neural network respectively, can calculate artificial neuron
The excitation value of the output layer of network:
For Fig. 2, input vector are as follows:<7.7,10.5,8.5,12>, the weight vectors of Y1 are<0.8, -0.2,0.65, -
0.3 >, obtain excitation value=5.985 Y1;The weight vectors of Y2 be<0.7, -0.3,0.9, -0.4>, obtain Y2 excitation value=
5.09。
The excitation function of Y1 and Y2 is set as " output==excitation value ", then the excitation value of Z1=<5.985,5.09>*<
1, -1 >=0.985;Z2 excitation value=<5.985,5.09>*<-1,1>=- 0.985.
The excitation function of Z1 and Z2 is set as that " if (excitation value >=0), 1, otherwise of output output 0 ", then Z1 is exported
1, Z2 output 0, obtains the judgement of " two straight lines are pairs of lane lines ".
For two lines between in Fig. 4, the vector value of acquirement is<2.7,8.2,6.2,4.3>, the corresponding weight vectors of Y1
Are as follows:<0.8, -0.2,0.65, -0.3>, obtain excitation value=3.26 Y1;The excitation vector of Y2 be<0.7, -0.3,0.9, -0.4>,
Obtain excitation value=3.29 Y2.
The excitation value of Z1=<3.26,3.29>*<1, -1>=- 0.03;Z2 excitation value=<3.26,3.29>*<-1,1>=
0.03.Z1 exports 0, Z2 output 1, obtains the judgement of " two non-paired lane lines of straight line ".
The present invention obtains two straight lines to be detected and selects sample on two straight lines according to preset spacing
Point obtains the distance between sampling point and common point vector, and the distance vector is substituted into trained artificial neural network in advance,
The excitation value that artificial neural network output can be obtained, judges whether two straight lines are pairs of lane according to the excitation value
Line.Using the method for the invention, it is only necessary to which the range data that will acquire, which substitutes into artificial neural network and can quickly determine, is
No is pairs of lane line, it can effectively guarantees the real-time judged pairs of lane line, and can be improved the accurate of judgement
Property.
Fig. 6 show the structural schematic diagram of the detection device of pairs of lane line provided in an embodiment of the present invention, and details are as follows:
The detection device of pairs of lane line described in the embodiment of the present invention, comprising:
Lane line acquiring unit 601, for obtaining two straight lines to be detected, according to preset spacing described two
The sampling point selected respectively on straight line, obtains the distance between the sampling point and scheduled common point vector;
Computing unit 602, for the distance vector to be substituted into preset artificial neural networks excitation value,
In, the weight vectors of the artificial neural network train acquisition according to pairs of lane line sample data gathered in advance;
Whether judging unit 603, the excitation value for being exported according to the artificial neural network judge two straight lines
For pairs of lane line.
Preferably, the lane line acquiring unit includes:
Sampling point selects subelement, for selecting sampling point respectively on two straight lines according to preset spacing;
Common point obtains subelement, for using the central point of image as common point, obtain the sampling point with it is described public
The distance between point vector.
Preferably, described device further include:
Sample collection unit, for acquiring a large amount of pairs of lane line sample and azygous lane line sample, according to
The spacing selects sampling point on the lane line sample;
Metrics calculation unit, for calculating the distance between the sampling point and the common point;
Weight vector computation unit, for the distance to be substituted into the neuronal cell layers of artificial neural network, according to sample
Whether in pairs, the corresponding weight vectors of the neuronal cell layers of artificial neural network are calculated.
Preferably, the sampling point includes the N number of sampling point selected on every lane line, and the N is oneself more than or equal to 2
So number.
Preferably, the judging unit includes:
Comparing subunit has, for obtaining the excitation value of the artificial neural network output, by the excitation value and in advance
The threshold value of setting is compared;
Pairs of lane line determines subelement, if being greater than the threshold value for the excitation value, it is determined that described two straight
Line is pairs of lane line, if the excitation value is less than the threshold value, it is determined that two straight lines are not pairs of lane lines.
The detection device of pairs of lane line, corresponding with the detection method of above-mentioned pairs of lane line described in the embodiment of the present invention,
So here is no more repetition.
In several embodiments provided by the present invention, it should be understood that disclosed device and method can pass through it
Its mode is realized.For example, the apparatus embodiments described above are merely exemplary, for example, the division of the unit, only
Only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components can be tied
Another system is closed or is desirably integrated into, or some features can be ignored or not executed.Another point, it is shown or discussed
Mutual coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or logical of device or unit
Letter connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially
The all or part of the part that contributes to existing technology or the technical solution can be in the form of software products in other words
It embodies, which is stored in a storage medium, including some instructions are used so that a computer
Equipment (can be personal computer, server or the network equipment etc.) executes the complete of each embodiment the method for the present invention
Portion or part.And storage medium above-mentioned include: USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory),
Random access memory (RAM, Random Access Memory), magnetic or disk etc. be various to can store program code
Medium.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (10)
1. a kind of detection method of pairs of lane line, which is characterized in that the described method includes:
Two straight lines to be detected are obtained, according to the sampling point that preset spacing selects respectively on two straight lines, are obtained
The distance between the sampling point and scheduled common point vector are taken, the common point includes the midpoint on the top in image, image
In lower part midpoint or image central point;
The distance vector is substituted into preset artificial neural networks excitation value, wherein the artificial neural network
Weight vectors acquisition is trained according to pairs of lane line sample data gathered in advance and non-paired lane line sample data;
According to the excitation value that the artificial neural network exports, judge whether two straight lines are pairs of lane line.
2. method according to claim 1, which is characterized in that it is described according to preset spacing on two straight lines
The sampling point selected respectively, obtaining the distance between the sampling point and scheduled common point vector step includes:
According to preset spacing, sampling point is selected respectively on two straight lines;
Using the central point of image as common point, the distance between the sampling point and the common point vector are obtained.
3. method according to claim 1, which is characterized in that described that distance vector substitution is preset artificial
Before neural computing excitation value step, the method also includes:
Acquisition a large amount of pairs of lane line sample and azygous lane line sample, according to the spacing in the lane line sample
Sampling point is selected on this;
Calculate the distance between the sampling point and the common point;
The distance is substituted into the neuronal cell layers of artificial neural network, it is whether pairs of according to sample, calculate artificial neural network
Neuronal cell layers corresponding weight vectors.
4. method according to claim 1, which is characterized in that the sampling point includes the N number of sample selected on every lane line
Point, the N are the natural number more than or equal to 2.
5. method according to claim 1, which is characterized in that the excitation value exported according to the artificial neural network,
Judge whether two straight lines are that pairs of lane line step includes:
The excitation value for obtaining the artificial neural network output, the excitation value is compared with preset threshold value;
If the excitation value is greater than the threshold value, it is determined that two straight lines are pairs of lane line, if the excitation value
Less than the threshold value, it is determined that two straight lines are not pairs of lane lines.
6. a kind of detection device of pairs of lane line, which is characterized in that described device includes:
Lane line acquiring unit, for obtaining two straight lines to be detected, according to preset spacing in two straight lines
The upper sampling point selected respectively, obtains the distance between the sampling point and scheduled common point vector, and the common point includes image
In top midpoint, the midpoint of the lower part in image or the central point of image;
Computing unit, for the distance vector to be substituted into preset artificial neural networks excitation value, wherein described
The weight vectors of artificial neural network are according to pairs of lane line sample data gathered in advance and non-paired lane line sample data
Train acquisition;
Judging unit, the excitation value for being exported according to the artificial neural network judge whether two straight lines are pairs of
Lane line.
7. device according to claim 6, which is characterized in that the lane line acquiring unit includes:
Sampling point selects subelement, for selecting sampling point respectively on two straight lines according to preset spacing;
Common point obtains subelement, for regarding the central point of image as common point, obtain the sampling point and the common point it
Between distance vector.
8. device according to claim 6, which is characterized in that described device further include:
Sample collection unit, for acquiring a large amount of pairs of lane line sample and azygous lane line sample, according to described
Spacing selects sampling point on the lane line sample;
Metrics calculation unit, for calculating the distance between the sampling point and the common point;
Weight vector computation unit, for by it is described distance substitute into artificial neural network neuronal cell layers, according to sample whether
In pairs, the corresponding weight vectors of the neuronal cell layers of artificial neural network are calculated.
9. device according to claim 6, which is characterized in that the sampling point includes the N number of sample selected on every lane line
Point, the N are the natural number more than or equal to 2.
10. device according to claim 6, which is characterized in that the judging unit includes:
Comparing subunit, for obtaining the excitation value of artificial neural network output, by the excitation value with it is preset
Threshold value is compared;
Pairs of lane line determines subelement, if being greater than the threshold value for the excitation value, it is determined that two straight lines are
Pairs of lane line, if the excitation value is less than the threshold value, it is determined that two straight lines are not pairs of lane lines.
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/CN2016/096761 WO2018035815A1 (en) | 2016-08-25 | 2016-08-25 | Method and device for detecting paired lane lines |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106415602A CN106415602A (en) | 2017-02-15 |
CN106415602B true CN106415602B (en) | 2019-12-03 |
Family
ID=58087907
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201680000880.XA Active CN106415602B (en) | 2016-08-25 | 2016-08-25 | A kind of detection method and device of pairs of lane line |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN106415602B (en) |
WO (1) | WO2018035815A1 (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106462755B (en) * | 2016-09-26 | 2019-05-28 | 深圳市锐明技术股份有限公司 | Pairs of method for detecting lane lines and device |
CN106778791A (en) * | 2017-03-01 | 2017-05-31 | 成都天衡电科科技有限公司 | A kind of timber visual identity method based on multiple perceptron |
Family Cites Families (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR2848935B1 (en) * | 2002-12-20 | 2005-04-29 | Valeo Vision | METHOD FOR DETECTING TURNS ON A ROAD AND SYSTEM FOR IMPLEMENTING SAME |
CN102201167B (en) * | 2010-04-07 | 2013-03-06 | 宫宁生 | Video-based driveway automatic identification method |
KR101261409B1 (en) * | 2012-04-24 | 2013-05-10 | 이엔지정보기술 주식회사 | System for recognizing road markings of image |
CN102750825B (en) * | 2012-06-19 | 2014-07-23 | 银江股份有限公司 | Urban road traffic condition detection method based on neural network classifier cascade fusion |
JP6193819B2 (en) * | 2014-07-11 | 2017-09-06 | 株式会社Soken | Traveling line recognition device |
CN104063877B (en) * | 2014-07-16 | 2017-05-24 | 中电海康集团有限公司 | Hybrid judgment identification method for candidate lane lines |
CN104102905B (en) * | 2014-07-16 | 2018-03-16 | 中电海康集团有限公司 | A kind of adaptive detection method of lane line |
CN105069415B (en) * | 2015-07-24 | 2018-09-11 | 深圳市佳信捷技术股份有限公司 | Method for detecting lane lines and device |
CN105046235B (en) * | 2015-08-03 | 2018-09-07 | 百度在线网络技术(北京)有限公司 | The identification modeling method and device of lane line, recognition methods and device |
CN105260699B (en) * | 2015-09-10 | 2018-06-26 | 百度在线网络技术(北京)有限公司 | A kind of processing method and processing device of lane line data |
CN105260713B (en) * | 2015-10-09 | 2019-06-28 | 东方网力科技股份有限公司 | A kind of method for detecting lane lines and device |
CN105608429B (en) * | 2015-12-21 | 2019-05-14 | 重庆大学 | Robust method for detecting lane lines based on difference excitation |
CN105718916A (en) * | 2016-01-27 | 2016-06-29 | 大连楼兰科技股份有限公司 | Lane line detection method based on Hough transform |
-
2016
- 2016-08-25 CN CN201680000880.XA patent/CN106415602B/en active Active
- 2016-08-25 WO PCT/CN2016/096761 patent/WO2018035815A1/en active Application Filing
Also Published As
Publication number | Publication date |
---|---|
CN106415602A (en) | 2017-02-15 |
WO2018035815A1 (en) | 2018-03-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106462757B (en) | A kind of rapid detection method and device of pairs of lane line | |
CN108121984B (en) | Character recognition method and device | |
CN107368787A (en) | A kind of Traffic Sign Recognition algorithm that application is driven towards depth intelligence | |
CN110070029B (en) | Gait recognition method and device | |
CN106485951A (en) | Abnormal driving person's classification and report | |
CN109034632B (en) | A kind of deep learning model safety methods of risk assessment based on to resisting sample | |
CN112466298A (en) | Voice detection method and device, electronic equipment and storage medium | |
CN108537326A (en) | For the method for automatic driving vehicle, medium and system | |
Xing et al. | Driver workload estimation using a novel hybrid method of error reduction ratio causality and support vector machine | |
KR20210048187A (en) | Method and apparatus for training model for object classification and detection | |
CN109977980A (en) | A kind of method for recognizing verification code and device | |
CN101216884A (en) | A method and system for face authentication | |
CN108460427A (en) | A kind of disaggregated model training method, device and sorting technique and device | |
CN104867225A (en) | Banknote face orientation identification method and apparatus | |
CN104573680B (en) | Image detecting method, image detection device and traffic offence detecting system | |
CN106415602B (en) | A kind of detection method and device of pairs of lane line | |
CN113688957A (en) | Target detection method, device, equipment and medium based on multi-model fusion | |
CN105678381A (en) | Gender classification network training method, gender classification method and related device | |
CN107153811A (en) | Know method for distinguishing, apparatus and system for multi-modal biological characteristic | |
CN111881706B (en) | Living body detection, image classification and model training method, device, equipment and medium | |
CN117271768A (en) | False news detection method and device based on large language model analysis and guidance | |
CN105991290A (en) | Pseudo random voiceprint cipher text generation method and system | |
Hernández Sánchez et al. | Deep neural networks for driver identification using accelerometer signals from smartphones | |
CN110378353A (en) | A kind of tongue picture feature extracting method, system and computer readable storage medium | |
CN113192602A (en) | Test system for student mental health 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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |