CN104392245A - Multi-sensor fusion road surface type identification method and device - Google Patents

Abstract

The invention provides a multi-sensor fusion road surface type identification method and device. The multi-sensor fusion road surface type identification method comprises steps as follows: road surface type identification results of multiple sensors fixed on a vehicle are acquired; the road surface type identification results of all sensors are subjected to merge processing, and a merged road surface type identification result of each sensor is acquired; and the merged road surface type identification results of the multiple sensors are optimized according to a Markov random field model so as to generate a final road surface type identification result. Compared with the prior art, the multi-sensor fusion road surface type identification method and device improve the accuracy rate of road surface identification.

Description

The method for recognizing road surface types of Multi-sensor Fusion and device

Technical field

The present invention relates to the road surface types recognition technology in vehicle travel process, particularly relate to a kind of method for recognizing road surface types and device of Multi-sensor Fusion.

Background technology

What require security in car steering process, comfortableness along with people improves constantly, and road surface types recognition technology highlights particularly important function and significance.In prior art, the laser radar technique that adopts to improve scope and the accuracy rate of road surface identification more, by the laser radar being arranged on a certain position at the bottom of vehicle front, roof or car, the road surface that vehicle travels is scanned, extract road surface spatial frequency features through data processing again, can complete the identifying of road surface types.

State in realization in the process of invention, inventor finds that in prior art, at least there are the following problems: the mode of usual this installation laser radar requires that laser beam and road surface want shape in an angle.If the laser beam of laser radar can not vertical irradiation to ground, and form this angle, road surface profile measuring accuracy will be made to decline to some extent, thus affect the accuracy of road surface types identification.

Summary of the invention

Embodiments of the invention provide a kind of method for recognizing road surface types and device of Multi-sensor Fusion, to improve the accuracy rate that road surface identifies, can carry out road surface identification to vehicle front road surface simultaneously.

For achieving the above object, the invention provides a kind of method for recognizing road surface types of Multi-sensor Fusion, comprising: obtain road surface types recognition result step: the road surface types recognition result obtaining the multiple sensors be fixed on vehicle; Road surface types recognition result combining step: the road surface types recognition result of described each sensor is carried out merging treatment according to ballot method, obtains the road surface types recognition result after the merging of each sensor; Final road surface types recognition result generation step: the road surface types recognition result after the merging of described multiple sensor is optimized process according to Markov random field model, generates final road surface types recognition result.

Present invention also offers a kind of road surface types recognition device of Multi-sensor Fusion, comprising: obtaining road surface types recognition result module, for obtaining the road surface types recognition result of the multiple sensors be fixed on vehicle; Road surface types recognition result merges module, for the road surface types recognition result of described each sensor being carried out merging treatment according to ballot method, obtains the road surface types recognition result after the merging of each sensor; Final road surface types recognition result generation module, for the road surface types recognition result after the merging of described multiple sensor being optimized process according to Markov random field model, generates final road surface types recognition result.

The method for recognizing road surface types of the Multi-sensor Fusion that the embodiment of the present invention provides and device, by obtaining the road surface types recognition result of multiple sensor, and merging treatment is carried out to the road surface types recognition result of each sensor, utilize markov random file (Markov Radom Field, MRF) model energy equation is optimized process to the road surface types recognition result after the merging of multiple sensor, just can obtain final recognition result, thus improve the accuracy rate of road surface identification.

Accompanying drawing explanation

Fig. 1 is the schematic flow sheet of the method for recognizing road surface types of the Multi-sensor Fusion of the embodiment of the present invention one;

Fig. 2 is the MRF modular concept schematic diagram that vehicle of the present invention travels road surface ahead type identification;

Fig. 3 is the structural representation of the road surface types recognition device of the Multi-sensor Fusion of the embodiment of the present invention two.

Embodiment

Below in conjunction with accompanying drawing, the method for recognizing road surface types of embodiment of the present invention Multi-sensor Fusion and device are described in detail.

The know-why of the method for recognizing road surface types of Multi-sensor Fusion of the present invention is the road surface types recognition result by obtaining multiple sensor, and merging treatment is carried out to the road surface types recognition result of each sensor, utilize MRF model energy equation to be optimized process to the road surface types recognition result after the merging of multiple sensor, and then obtain final road surface types recognition result.

Embodiment one

As shown in Figure 1, it is the schematic flow sheet of the method for recognizing road surface types of the Multi-sensor Fusion of embodiment one, and it comprises:

Obtain road surface types recognition result step 101: the road surface types recognition result obtaining the multiple sensors be fixed on vehicle.

Road surface types recognition result combining step 102: the road surface types recognition result of each sensor is carried out merging treatment according to ballot method, obtains the road surface types recognition result after the merging of each sensor.Particularly, within a predetermined period of time, merging treatment is carried out to the road surface types recognition result of each sensor, generate the road surface types recognition result after merging, if in this predetermined amount of time, certain sensor only exports a road surface types recognition result, then direct using this road surface types recognition result as merge after road surface types recognition result, and for exporting the sensor of multiple road surface types recognition result within a predetermined period of time, with regard to needing, the multiple road surfaces recognition result exported is carried out merging treatment, thus obtain an Output rusults, for subsequent step provides data basis.

Such as, laser radar scans tens times (i.e. tens sweep traces) p.s., and all corresponding recognition result of every bar sweep trace, that is tens recognition results can be exported p.s., equally, imageing sensor imaging p.s. tens times, also tens recognition results can be exported, and acceleration transducer only exports a recognition result in same time, therefore multiple recognition results of laser radar and imageing sensor are needed to carry out merging treatment, so that and the number of the recognition result of acceleration transducer output matches, in order to subsequent treatment.

Final road surface types recognition result generation step 103: the road surface types recognition result after the merging of multiple sensor is optimized process according to Markov random field model, generates final road surface types recognition result.

The method for recognizing road surface types of Multi-sensor Fusion of the present invention, by obtaining the road surface types recognition result of multiple sensor, and merging treatment is carried out to the road surface types recognition result of each sensor, MRF model energy equation is utilized to be optimized process to the road surface types recognition result after the merging of multiple sensor, just can obtain final recognition result, compared with prior art, the accuracy rate of the road surface identification of pavement identification method of the present invention is higher.

Further, road surface types recognition result combining step 102 can specifically comprise: for often kind of road surface types gives numerical value respectively, the corresponding numerical value of each road surface types; Search the numerical value that often kind of road surface types is corresponding, the road surface types recognition result being divided into road surface types is treated to the road surface types recognition result be made up of numerical value; Add up the number of each numerical value in the road surface types recognition result be made up of numerical value, obtain the numerical value that number is maximum; Based on the road surface types recognition result after the merging of each sensor of numerical generation.

Particularly, suppose there are four kinds of road surface types, numerical value " 1 ", " 2 ", " 3 ", " 4 " can be given respectively for these four kinds of road surface types, the road surface types recognition result now represented by concrete road surface types has just been processed into the road surface types recognition result by numeric representation, and then add up the number of each numerical value, find out that numerical value that number is maximum, so this numerical value just can as the road surface types recognition result after the merging of sensor.By said method, more intuitively the Output rusults of the sensor exporting multiple road surface types recognition result is within a predetermined period of time carried out merging treatment.

Further, in order to improve the accuracy of road surface types identification, can choose a sensor as master reference from multiple sensor, other sensors are as from sensor, and final road surface types recognition result generation step 103 can specifically comprise:

The energy equation creating described Markov random field model is:

$\begin{array}{c}E(x,y,u,w,v)=\mathrm{\ρ}\underset{i}{\mathrm{\Σ}}\sqrt{|x_i^2-x_{i-1}^2|}+\underset{i}{\mathrm{\Σ}}{r}_i\sqrt{|x_i^2-y_i^2|}+\\ \underset{i}{\mathrm{\Σ}}{k1}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^1\right)}^2|}+\underset{i}{\mathrm{\Σ}}{k2}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^2\right)}^2|}+\underset{i}{\mathrm{\Σ}}{k3}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^3\right)}^2|}+...+\underset{i}{\mathrm{\Σ}}{\mathrm{kn}}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^n\right)}^2}\end{array}$

Wherein, i is the sequence number being identified section, and i is integer and i>=2, x _ibe the road surface types recognition result of i-th section master reference to be optimized, x _i-1be the final road surface types recognition result in the i-th-1 section, y _ibe the road surface types recognition result after the merging of i-th section master reference, be the i-th-1 section first from the road surface types recognition result after the merging of sensor, be the i-th-1 section second from the road surface types recognition result after the merging of sensor, be the i-th-1 section the 3rd from the road surface types recognition result after the merging of sensor, be the i-th-1 section n-th from the road surface types recognition result after the merging of sensor, n is the sequence number from sensor, and ρ is x _iwith x _i-1between link potential energy, r _ifor x _iwith y _ibetween link potential energy, k1 _i-1for x _iwith between link potential energy, k2 _i-1for x _iwith between link potential energy, k3 _i-1for x _iwith between link potential energy, kn _i-1for x _iwith between link potential energy;

By the road surface types recognition result x of master reference to be optimized for i-th section _icorresponding numerical value is brought energy equation respectively into and is calculated, and can make the x that the value of energy equation E (x, y, u, w, v) is minimum _icorresponding road surface types is as final road surface types recognition result.

Further, road surface types recognition result can comprise the combination of one or more road surface types in asphalt surface, cement pavement, road surface, meadow, stone road surface.

Further, in order to identify the road surface types of vehicle front accurately, thus improve the security of vehicle traveling, master reference can be forward laser light radar, can be imageing sensor, acceleration transducer and rearmounted laser radar from sensor.

Below in conjunction with concrete application scenarios, further illustrate the embody rule of the embodiment of the present invention.

Four sensors installed by vehicle, i.e. forward laser light radar, imageing sensor, acceleration transducer and rearmounted laser radar, wherein forward laser light radar is installed on vehicle front, the road surface of vehicle front is scanned, rearmounted laser radar is installed on rear view of vehicle, the road surface of rear view of vehicle is scanned, acceleration transducer can be installed on any position of vehicle, imageing sensor is installed on rear view of vehicle, shooting rear view of vehicle pavement image, above-mentioned mounting means is not limited thereto, such as imageing sensor also can be installed on the positions such as vehicle front, as long as can road pavement imaging, certainly different installation site possibility imaging effects is different.The target of this embody rule is optimized the road surface types recognition result of forward laser light radar, and reached the object identifying vehicle front road surface types, concrete identifying is as follows:

First, the road surface types recognition result of these four sensors is obtained, secondly, different numerical value is set respectively to be identified four kinds of road surface types, be " 1 " by asphalt surface assignment, cement pavement assignment is " 2 ", road surface, meadow assignment is " 3 ", stone road surface assignment is " 4 ", the road surface types recognition result represented by road surface types is processed into by the road surface types recognition result of numeric representation, be exactly add up in all road surface types recognition results exported in the given time " 1 " respectively to each sensor according to ballot method, " 2 ", " 3 ", these four numerical value number separately of " 4 " is how many respectively, find that numerical value that number is maximum, using this numerical value as the recognition result after the merging of this road section surface.For the sensor exporting multiple road surface types recognition result within this time period, just need to carry out above-mentioned merging treatment process, and for only exporting the sensor of a road surface types recognition result, then direct using this road surface types recognition result as the road surface types recognition result after merging, thus obtain the road surface types recognition result after the merging of forward laser light radar, imageing sensor, acceleration transducer and rearmounted laser radar respectively.

Then MRF model energy equation is created, Fig. 2 is the MRF modular concept schematic diagram that vehicle of the present invention travels road surface ahead type identification, with reference to Fig. 2, each MRF model is made up of 5 nodes, forward laser light radar y, imageing sensor u, acceleration transducer w respectively, road surface types recognition result after rearmounted laser radar v merges separately, middle node x is the road surface types recognition result of forward laser light radar to be optimized.The energy equation created is as follows:

$\begin{array}{c}E(x,y,u,w,v)=\mathrm{\ρ}\underset{i}{\mathrm{\Σ}}\sqrt{|x_i^2-x_{i-1}^2|}+\underset{i}{\mathrm{\Σ}}{r}_i\sqrt{|x_i^2-y_i^2|}+\\ \underset{i}{\mathrm{\Σ}}{k1}_{i-1}\sqrt{|x_i^2-u_{i-1}^2|}+\underset{i}{\mathrm{\Σ}}{k2}_{i-1}\sqrt{|x_i^2-w_{i-1}^2|}+\underset{i}{\mathrm{\Σ}}{k3}_{i-1}\sqrt{|x_i^2-v_{i-1}^2|}\end{array}$

Wherein, i is the sequence number being identified section, and i is integer and i>=2, x _ibe the road surface types recognition result of i-th section forward laser light radar to be optimized, x _i-1be the final road surface types recognition result in the i-th-1 section, y _ibe the road surface types recognition result after the merging of i-th section forward laser light radar, u _i-1be the road surface types recognition result after the merging of the i-th-1 section imageing sensor, w _i-1be the road surface types recognition result after the merging of the i-th-1 section acceleration transducer, v _i-1be the road surface types recognition result after the merging of the i-th-1 rearmounted laser radar in section, ρ is x _iwith x _i-1between link potential energy, r _ifor x _iwith y _ibetween link potential energy, k1 _i-1for x _iwith u _i-1between link potential energy, k2 _i-1for x _iwith w _i-1between link potential energy, k3 _i-1for x _iwith v _i-1between link potential energy.

By x _i=1, x _i=2, x _i=3, x _i=4 bring energy equation E (x, y, u, w, v) respectively into, can make that x that the value of energy equation E (x, y, u, w, v) is minimum _icorresponding road surface types is exactly final road surface types recognition result.By obtaining the road surface types recognition result of forward laser light radar, imageing sensor, acceleration transducer and rearmounted laser radar four sensors, and merging treatment is carried out to the road surface types recognition result of each sensor, utilize MRF model energy equation to be optimized process to the road surface types recognition result after the merging of four sensors and obtain final recognition result, thus vehicle front road surface types can be identified exactly, improve the security that vehicle travels.

In actual applications, multiple specific implementation can also be had, be not limited in the process of foregoing description, such as, more sensor can also be had to be installed on vehicle, or the road surface types recognition result of rearmounted laser radar is optimized, reach the object identifying rear view of vehicle road surface types.

Embodiment two

As shown in Figure 3, it is the structural representation of the road surface types recognition device of the Multi-sensor Fusion of the embodiment of the present invention two, it comprises: obtain road surface types recognition result module 201, for obtaining the road surface types recognition result of the multiple sensors be fixed on vehicle; Road surface types recognition result merges module 202, for the road surface types recognition result of each sensor being carried out merging treatment according to ballot method, obtains the road surface types recognition result after the merging of each sensor; Final road surface types recognition result generation module 203, for the road surface types recognition result after the merging of multiple sensor being optimized process according to Markov random field model, generates final road surface types recognition result.

The road surface types recognition device of Multi-sensor Fusion of the present invention, by obtaining the road surface types recognition result of multiple sensor, and merging treatment is carried out to the road surface types recognition result of each sensor, MRF model energy equation is utilized to be optimized process to the road surface types recognition result after the merging of multiple sensor, just can obtain final recognition result, compared with prior art, the accuracy rate of the road surface identification of road surface of the present invention recognition device is higher.

Further, because the type of sensor is different, within a predetermined period of time, some sensors only export a road surface types recognition result, and some sensors have exported multiple road surface types recognition result, in order to the Output rusults of the sensor exporting multiple road surface types recognition result is adjusted to one, road surface types recognition result merges module 202 and can comprise:

Road surface types assignment unit, for giving numerical value respectively for often kind of road surface types, the corresponding numerical value of each road surface types;

Road surface types recognition result processing unit, for searching numerical value corresponding to often kind of road surface types, is treated to the road surface types recognition result be made up of numerical value by the road surface types recognition result being divided into road surface types;

Road surface types recognition result statistic unit, for adding up the number of each numerical value in the road surface types recognition result that is made up of numerical value, obtains the numerical value that number is maximum;

The road surface types recognition result generation unit merged, for the road surface types recognition result after the merging based on each sensor of numerical generation.

Further, in order to improve the accuracy of road surface types identification, can choose a sensor as master reference from multiple sensor, other sensors are as from sensor, and final road surface types recognition result generation module 203 can comprise:

Energy equation creating unit, for creating the energy equation of Markov random field model is:

$\begin{array}{c}E(x,y,u,w,v)=\mathrm{\ρ}\underset{i}{\mathrm{\Σ}}\sqrt{|x_i^2-x_{i-1}^2|}+\underset{i}{\mathrm{\Σ}}{r}_i\sqrt{|x_i^2-y_i^2|}+\\ \underset{i}{\mathrm{\Σ}}{k1}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^1\right)}^2|}+\underset{i}{\mathrm{\Σ}}{k2}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^2\right)}^2|}+\underset{i}{\mathrm{\Σ}}{k3}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^3\right)}^2|}+...+\underset{i}{\mathrm{\Σ}}{\mathrm{kn}}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^n\right)}^2}\end{array}$

Wherein, i is the sequence number being identified section, and i is integer and i>=2, x _ibe the road surface types recognition result of i-th section master reference to be optimized, x _i-1be the final road surface types recognition result in the i-th-1 section, y _ibe the road surface types recognition result after the merging of i-th section master reference, be the i-th-1 section first from the road surface types recognition result after the merging of sensor, be the i-th-1 section second from the road surface types recognition result after the merging of sensor, be the i-th-1 section the 3rd from the road surface types recognition result after the merging of sensor, be the i-th-1 section n-th from the road surface types recognition result after the merging of sensor, n is the sequence number from sensor, and ρ is x _iwith x _i-1between link potential energy, r _ifor x _iwith y _ibetween link potential energy, k1 _i-1for x _iwith between link potential energy, k2 _i-1for x _iwith between link potential energy, k3 _i-1for x _iwith between link potential energy, kn _i-1for x _iwith between link potential energy;

Final road surface types recognition result generation unit, for the road surface types recognition result x by master reference to be optimized for i-th section _icorresponding numerical value is brought energy equation respectively into and is calculated, and can make the x that the value of energy equation E (x, y, u, w, v) is minimum _icorresponding road surface types is as final road surface types recognition result.

Further, master reference can be forward laser light radar, can be imageing sensor, acceleration transducer and rearmounted laser radar from sensor, like this laser radar being arranged on vehicle front is chosen for the mode of master reference, the road surface types of vehicle front can be identified accurately, thus improve the security of vehicle traveling.

Further, road surface types recognition result can comprise the combination of one or more road surface types in asphalt surface, cement pavement, road surface, meadow, stone road surface.

In several embodiment provided by the present invention, should be understood that, disclosed method and apparatus, can realize by another way.Such as, device embodiment described above is only schematic, and such as, the division of described module, is only a kind of logic function and divides, and actual can have other dividing mode when realizing.

In addition, each functional module in each embodiment of the present invention can be integrated in a processing module, also can be that the independent physics of modules exists, also can two or more module integrations in a module.Above-mentioned integrated module both can adopt the form of hardware to realize, and the form that hardware also can be adopted to add software function module realizes.

The above-mentioned integrated module realized with the form of software function module, can be stored in a computer read/write memory medium.Above-mentioned software function module is stored in a storage medium, comprising some instructions in order to make a computer equipment (can be personal computer, server, or the network equipment etc.) or processor (processor) perform the part steps of method described in each embodiment of the present invention.And aforesaid storage medium comprises: USB flash disk, portable hard drive, ROM (read-only memory) (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disc or CD etc. various can be program code stored medium.

The above; be only the specific embodiment of the present invention, but protection scope of the present invention is not limited thereto, is anyly familiar with those skilled in the art in the technical scope that the present invention discloses; change can be expected easily or replace, all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection domain of described claim.

Claims (10)

1. a method for recognizing road surface types for Multi-sensor Fusion, is characterized in that, described method comprises:

Obtain road surface types recognition result step: the road surface types recognition result obtaining the multiple sensors be fixed on vehicle;

Road surface types recognition result combining step: the road surface types recognition result of described each sensor is carried out merging treatment according to ballot method, obtains the road surface types recognition result after the merging of each sensor;

Final road surface types recognition result generation step: the road surface types recognition result after the merging of described multiple sensor is optimized process according to Markov random field model, generates final road surface types recognition result.

2. method according to claim 1, is characterized in that, described road surface types recognition result combining step specifically comprises:

For often kind of road surface types gives numerical value respectively, the corresponding numerical value of each road surface types;

Search the numerical value that often kind of road surface types is corresponding, the road surface types recognition result being divided into road surface types is treated to the road surface types recognition result be made up of numerical value;

In the road surface types recognition result be made up of numerical value described in statistics, the number of each numerical value, obtains the numerical value that number is maximum;

Based on the road surface types recognition result after the merging of each sensor of described numerical generation.

3. method according to claim 1, is characterized in that, chooses a sensor as master reference from described multiple sensor, and other sensors are as from sensor, and described final road surface types recognition result generation step specifically comprises:

The energy equation creating described Markov random field model is:

$\begin{array}{c}E(x,y,u,w,v)=\mathrm{\ρ}\underset{i}{\mathrm{\Σ}}\sqrt{|x_i^2-x_{i-1}^2|}+\underset{i}{\mathrm{\Σ}}{r}_i\sqrt{|x_i^2-y_i^2|}+\\ \underset{i}{\mathrm{\Σ}}k{1}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^1\right)}^2|}+\underset{i}{\mathrm{\Σ}}k{2}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^2\right)}^2|}+\underset{i}{\mathrm{\Σ}}k{3}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^3\right)}^2|}+...+\underset{i}{\mathrm{\Σ}}k{n}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^n\right)}^2|}\end{array}$

Wherein, i is the sequence number being identified section, and i is integer and i>=2, x _ibe the road surface types recognition result of i-th section master reference to be optimized, x _i-1be the final road surface types recognition result in the i-th-1 section, y _ibe the road surface types recognition result after the merging of i-th section master reference, be the i-th-1 section first from the road surface types recognition result after the merging of sensor, be the i-th-1 section second from the road surface types recognition result after the merging of sensor, be the i-th-1 section the 3rd from the road surface types recognition result after the merging of sensor, be the i-th-1 section n-th from the road surface types recognition result after the merging of sensor, n is the sequence number from sensor, and ρ is x _iwith x _i-1between link potential energy, r _ifor x _iwith y _ibetween link potential energy, k1 _i-1for x _iwith between link potential energy, k2 _i-1for x _iwith between link potential energy, k3 _i-1for x _iwith between link potential energy, kn _i-1for x _iwith between link potential energy;

By the road surface types recognition result x of master reference to be optimized for described i-th section _icorresponding numerical value is brought described energy equation respectively into and is calculated, and can make the x that the value of energy equation E (x, y, u, w, v) is minimum _icorresponding road surface types is as described final road surface types recognition result.

4. method according to claim 3, is characterized in that, described master reference is forward laser light radar, described from sensor be imageing sensor, acceleration transducer and rearmounted laser radar.

5. method according to claim 1, is characterized in that, described road surface types recognition result comprises the combination of one or more road surface types in asphalt surface, cement pavement, road surface, meadow, stone road surface.

6. a road surface types recognition device for Multi-sensor Fusion, is characterized in that, described device comprises:

Obtain road surface types recognition result module, for obtaining the road surface types recognition result of the multiple sensors be fixed on vehicle;

Road surface types recognition result merges module, for the road surface types recognition result of described each sensor being carried out merging treatment according to ballot method, obtains the road surface types recognition result after the merging of each sensor;

Final road surface types recognition result generation module, for the road surface types recognition result after the merging of described multiple sensor being optimized process according to Markov random field model, generates final road surface types recognition result.

7. device according to claim 6, is characterized in that, described road surface types recognition result merges module and comprises:

Road surface types assignment unit, for giving numerical value respectively for often kind of road surface types, the corresponding numerical value of each road surface types;

Road surface types recognition result processing unit, for searching numerical value corresponding to often kind of road surface types, is treated to the road surface types recognition result be made up of numerical value by the road surface types recognition result being divided into road surface types;

Road surface types recognition result statistic unit, for adding up the number of each numerical value in the described road surface types recognition result be made up of numerical value, obtains the numerical value that number is maximum;

The road surface types recognition result generation unit merged, for the road surface types recognition result after the merging based on each sensor of described numerical generation.

8. device according to claim 6, is characterized in that, chooses a sensor as master reference from described multiple sensor, and other sensors are as from sensor, and described final road surface types recognition result generation module comprises:

Energy equation creating unit, for creating the energy equation of described Markov random field model is:

$\begin{array}{c}E(x,y,u,w,v)=\mathrm{\ρ}\underset{i}{\mathrm{\Σ}}\sqrt{|x_i^2-x_{i-1}^2|}+\underset{i}{\mathrm{\Σ}}{r}_i\sqrt{|x_i^2-y_i^2|}+\\ \underset{i}{\mathrm{\Σ}}k{1}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^1\right)}^2|}+\underset{i}{\mathrm{\Σ}}k{2}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^2\right)}^2|}+\underset{i}{\mathrm{\Σ}}k{3}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^3\right)}^2|}+...+\underset{i}{\mathrm{\Σ}}k{n}_{i-1}\sqrt{|x_i^2-{\left(U_{i-1}^n\right)}^2|}\end{array}$

Wherein, i is the sequence number being identified section, and i is integer and i>=2, x _ibe the road surface types recognition result of i-th section master reference to be optimized, x _i-1be the final road surface types recognition result in the i-th-1 section, y _ibe the road surface types recognition result after the merging of i-th section master reference, be the i-th-1 section first from the road surface types recognition result after the merging of sensor, be the i-th-1 section second from the road surface types recognition result after the merging of sensor, be the i-th-1 section the 3rd from the road surface types recognition result after the merging of sensor, be the i-th-1 section n-th from the road surface types recognition result after the merging of sensor, n is the sequence number from sensor, and ρ is x _iwith x _i-1between link potential energy, r _ifor x _iwith y _ibetween link potential energy, k1 _i-1for x _iwith between link potential energy, k2 _i-1for x _iwith between link potential energy, k3 _i-1for x _iwith between link potential energy, kn _i-1for x _iwith between link potential energy;

Final road surface types recognition result generation unit, for the road surface types recognition result x by master reference to be optimized for described i-th section _icorresponding numerical value is brought described energy equation respectively into and is calculated, and can make the x that the value of energy equation E (x, y, u, w, v) is minimum _icorresponding road surface types is as described final road surface types recognition result.

9. device according to claim 8, is characterized in that, described master reference is forward laser light radar, described from sensor be imageing sensor, acceleration transducer and rearmounted laser radar.

10. device according to claim 6, is characterized in that, described road surface types recognition result comprises the combination of one or more road surface types in asphalt surface, cement pavement, road surface, meadow, stone road surface.