CN107301371A - A kind of unstructured road detection method and system based on image information fusion - Google Patents
A kind of unstructured road detection method and system based on image information fusion Download PDFInfo
- Publication number
- CN107301371A CN107301371A CN201710330557.5A CN201710330557A CN107301371A CN 107301371 A CN107301371 A CN 107301371A CN 201710330557 A CN201710330557 A CN 201710330557A CN 107301371 A CN107301371 A CN 107301371A
- Authority
- CN
- China
- Prior art keywords
- road
- video image
- little
- road video
- advance
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 36
- 230000004927 fusion Effects 0.000 title claims abstract description 35
- 238000012545 processing Methods 0.000 claims abstract description 26
- 238000012937 correction Methods 0.000 claims abstract description 20
- 238000000034 method Methods 0.000 claims abstract description 20
- 238000003708 edge detection Methods 0.000 claims abstract description 11
- 238000000638 solvent extraction Methods 0.000 claims abstract description 11
- 230000009466 transformation Effects 0.000 claims description 8
- 238000003064 k means clustering Methods 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 6
- 230000001629 suppression Effects 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 3
- 241001544487 Macromiidae Species 0.000 abstract description 7
- 230000006641 stabilisation Effects 0.000 abstract description 6
- 238000011105 stabilization Methods 0.000 abstract description 6
- 238000011160 research Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 10
- 238000005516 engineering process Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 5
- 238000003384 imaging method Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 230000004888 barrier function Effects 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 208000010877 cognitive disease Diseases 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 239000011800 void material Substances 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
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/48—Extraction of image or video features by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention discloses a kind of unstructured road detection method and system based on image information fusion, wherein, methods described includes step：Obtain road video image in real time by video camera, and the road video image is corrected；The road video image after correction is handled using unstructured road edge detection algorithm, first is obtained and takes aim at a little in advance；The road video image after correction is handled using unstructured road partitioning algorithm, second is obtained and takes aim at a little in advance；A little a little it is fitted with the second pre- take aim at using take aim at pre- to described first of least square method, obtains road virtual center line；The present invention can obtain stabilization, accurate reliable view data by above-mentioned calculating processing；Further, described image data application can be navigated in unmanned cruiser, it is possible to decrease research cost, with huge application value.
Description
Technical field
The present invention relates to road guide field, more particularly to a kind of unstructured road detection based on image information fusion
Method and system.
Background technology
The navigation system of view-based access control model is pattern-recognition, the research emphasis of artificial intelligence field, and it can be applied to nobody and drives
Sail intelligent vehicle.Lane detection technology is the key technology in auxiliary driving, unmanned technology, can be unmanned intelligent vehicle
Decision-making module provides necessary environmental information.Because navigation equipment has a limitation, vision sensor cheaper and with more
Big application potential, but the vision sensor is affected by environment larger.
Unstructured road refers to the road edge of the low road of structuring degree, typically no lane line and cleaning；By
Influenceed in by shade, water stain etc., unstructured road detection relative difficulty is still in conceptual phase at present.Unstructured road
Road detection method can be roughly divided into the detection based on roadway characteristic, the detection based on road model and the road based on machine learning
The three major types such as road detection.
Detection method based on roadway characteristic mainly detects road by extracting some features of road, and these features can
To be color, gray scale, texture, edge or frequency domain character of road etc., major advantage is, calculating speed insensitive to road shape
It hurry up, ensure that real-time；Have the disadvantage more sensitive to shadow, slight crack and water mark；K-means clusters based on SLIC super-pixel
Algorithm is, it is necessary to which priori, is divided into two major classes by adjusting parameter by road information：Wheeled region and non-running region；So
And because image information is influenceed than larger by illumination and external environment, image is carried out merely pretreatment obtain road edge or
Stabilization, accurate reliable view data can not be all obtained by splitting acquisition wheeled region.
Therefore, prior art has yet to be improved and developed.
The content of the invention
In view of above-mentioned the deficiencies in the prior art, it is an object of the invention to provide a kind of non-knot based on image information fusion
Structure Approach for road detection and system, it is intended to solve existing unstructured road detection method can not obtain stabilization, it is accurate reliable
View data the problem of.
Technical scheme is as follows：
A kind of unstructured road detection method based on image information fusion, wherein, including step：
A, obtain road video image in real time by video camera, and the road video image is corrected；
B, using unstructured road edge detection algorithm the road video image after correction is handled, obtain first and take aim in advance
Point；
C, using unstructured road partitioning algorithm the road video image after correction is handled, obtain second and take aim at a little in advance；
D, using least square method it is pre- to described first take aim at a little to take aim in advance with second be a little fitted, obtain road virtual center line.
The described unstructured road detection method based on image information fusion, wherein, the step B is specifically included：
B1, gray processing and noise suppression preprocessing are carried out to the road video image；
B2, road video image edge is detected using Canny algorithms and morphology amendment is carried out；
B3, road video image edge is extracted using Hough transformation and calculates first and takes aim at a little in advance.
The described unstructured road detection method based on image information fusion, wherein, the step C is specifically included：
C1, the color characteristic and space characteristics for extracting road video image in advance；
C2, using SLIC algorithms to road video image carry out segmentation obtain super-pixel data；
C3, road video image is clustered using K-means clustering algorithms and second is calculated and is taken aim at a little in advance.
The described unstructured road detection method based on image information fusion, wherein, the step C2 is specifically included：
C21, initialization clustering processing is carried out to the pixel in road video image obtain some initial seed points；
C22, in road video image 3*3 fields reselect seed point；
The nearest some pixels of C23, the seed point reselected described in detection range in the range of the 2S*2S, and to search
To pixel label be set and be classified as a class；
C24, when same pixel is classified into multiple seed points simultaneously, then calculate the pixel and multiple seed points it
Between distance, using apart from the corresponding seed point of minimum value as the pixel cluster centre；
C25, the iterative processing that step C22-C24 is carried out to the cluster centre, until error convergence, what is finally split is super
Pixel data.
The described unstructured road detection method based on image information fusion, wherein, the step C3 is specifically included：
C31, random several cluster center of mass point are chosen from the super-pixel data；
Each super-pixel data is divided into the center of mass point closest with it, formed by C32, the traversal super-pixel data
Cluster；
C33, the average value that each clusters of calculating and as new center of mass point；
C34, repeating said steps C32-C33, until center of mass point convergence, take aim at a little in advance so as to obtain second.
The described unstructured road detection method based on image information fusion, wherein, least square in the step D
The formula of method is：；
；Wherein, point
It is the function of many variables, and meets equationObtain minimum；K=0,1,
...n；WhenDuring linear independence,It is exactly required least square solution.
The described unstructured road detection method based on image information fusion, wherein, the gray scale in the step B1
Changing processing formula is：。
A kind of unstructured road detecting system based on image information fusion, wherein, including：
Correction module, is corrected for obtaining road video image in real time by video camera, and to the road video image；
First computing module, at using unstructured road edge detection algorithm to the road video image after correction
Reason, obtains first and takes aim at a little in advance；
Second computing module, for being handled using unstructured road partitioning algorithm the road video image after correction,
Second is obtained to take aim at a little in advance；
Fitting module, for being a little a little fitted with the second pre- take aim at using take aim at pre- to described first of least square method, obtains road
Imaginary center line.
The described unstructured road detecting system based on image information fusion, wherein, the first computing module tool
Body includes：
Pretreatment unit, for carrying out gray processing and noise suppression preprocessing to the road video image；
Amending unit, for detecting road video image edge using Canny algorithms and carrying out morphology amendment；
First computing unit, takes aim at a little in advance for being extracted road video image edge using Hough transformation and being calculated first.
The described unstructured road detecting system based on image information fusion, wherein, the second computing module tool
Body includes：
Extraction unit, color characteristic and space characteristics for extracting road video image in advance；
Cutting unit, super-pixel data are obtained for carrying out segmentation to road video image using SLIC algorithms；
Second computing unit, for being clustered and being calculated second to road video image using K-means clustering algorithms
Take aim at a little in advance.
Beneficial effect：The present invention passes through non-structural road-edge detection algorithm and unstructured road partitioning algorithm point first
The first pre- take aim at for not calculating road video image is a little taken aim at a little in advance with second, then pre- to described first by information fusion technology
Take aim at a little and second takes aim at the imaginary center line for being a little fitted and obtaining road video image in advance；The present invention can by above-mentioned calculating processing
Obtain stabilization, accurate reliable view data.
Brief description of the drawings
Fig. 1 is a kind of flow of the unstructured road detection method preferred embodiment based on image information fusion of the present invention
Figure；
Fig. 2 for the present invention in monocular-camera carry out calibration and usage chessboard schematic diagram；
Fig. 3 extracts road video image edge result schematic diagram for the present invention using Hough transformation；
Fig. 4 clusters the result schematic diagram of road video image for the present invention using the K-means based on SLIC；
Fig. 5 is steering wheel angle figure in the specific embodiment of the invention；
Fig. 6 is specific embodiment of the invention China and foreign countries side line road gps coordinate schematic diagram；
Fig. 7 is inner side road gps coordinate schematic diagram in the specific embodiment of the invention；
Fig. 8 is that angle schematic diagram is played on specific embodiment of the invention China and foreign countries side line road；
Fig. 9 plays angle schematic diagram for inner side road in the specific embodiment of the invention；
Figure 10 is road outside navigation driving trace and image driving trace contrast schematic diagram in the specific embodiment of the invention；
Figure 11 is road inner side navigation driving trace and image driving trace contrast schematic diagram in the specific embodiment of the invention；
Figure 12 is road inner side navigation path deviation map in the specific embodiment of the invention；
Figure 13 is navigation path deviation map on the outside of road in the specific embodiment of the invention；
Figure 14 is a kind of structural frames of the unstructured road detecting system preferred embodiment based on image information fusion of the present invention
Figure.
Embodiment
The present invention provides a kind of unstructured road detection method and system based on image information fusion, to make the present invention
Purpose, technical scheme and effect it is clearer, clear and definite, the present invention is described in more detail below.It should be appreciated that this place
The specific embodiment of description only to explain the present invention, is not intended to limit the present invention.
Referring to Fig. 1, Fig. 1 is preferably real for a kind of unstructured road detection method based on image information fusion of the present invention
The flow chart of example is applied, as illustrated, wherein, including step：
S100, obtain road video image in real time by video camera, and the road video image is corrected；
S200, using unstructured road edge detection algorithm the road video image after correction is handled, obtain first
Take aim at a little in advance；
S300, using unstructured road partitioning algorithm the road video image after correction is handled, obtain second and take aim in advance
Point；
S400, using least square method it is pre- to described first take aim at a little to take aim in advance with second be a little fitted, obtain road virtual center
Line.
Specifically, the unstructured road detection method based on image information fusion that the present invention is provided is in Windows
Developed on VS2010 platforms under system；Road video image is obtained by video camera first, then to the road video
Image is corrected, and calculates road respectively by non-structural road-edge detection algorithm and unstructured road partitioning algorithm
Pre- take aim at of the first of video image is a little taken aim at a little in advance with second, is taken aim at a little in advance to described first finally by information fusion technology and second is pre-
The imaginary center line for being a little fitted and obtaining road video image is taken aim at, stabilization can be obtained by above-mentioned calculating processing, accurate reliable
Road image data；The decision-making module that the road image data are sent into unmanned cruiser can complete vision guided navigation work(
Can, so as to realize the autonomous driving of unmanned cruiser.
In current intelligent driving field, the camera being applied to mainly has monocular cam and binocular camera two
Class, the range measurement principle of both cameras is entirely different, and monocular cam needs that target is first identified, that is to say, that monocular
It is car that camera needs first cognitive disorders thing before ranging, be people or other objects；Binocular camera is then directly by two
The disparity computation of width image determines distance, itself and to require no knowledge about barrier be what, the difficult point of binocular camera is to count
Amount amount is huge, and processing is slow and cost is high.
Based on above-mentioned difference, the present invention preferably monocular-camera goes to obtain road video image in real time, is imaged using monocular
Head can judge barrier in advance, and then ranging again can increase the security performance of intelligent driving.
Further, in the step S100, the present invention is entered using Zhang Zhengyou camera calibrations hair to the monocular-camera
Rower is determined, as shown in Fig. 2 being the plane that black and white that 10*10, the length of side are 0.05m is met each other to quasiconfiguaration by the monocular-camera
Chessboard, fixed monocular-camera, hand-held chessboard is rotated at various orientations obtains 18 checkerboard images, is obtained using Matlab tool boxes
The inside and outside parameter and distortion parameter of monocular-camera are obtained, is realized according to the parameter obtained and monocular-camera is demarcated；
Specifically, video camera generates each coordinate points under the plane of delineation, world coordinate system by national forest park in Xiaokeng and schemed
As there is corresponding pixel under pixel coordinate system, the process of camera calibration is video camera geometrical model parametric solution
Process, by the optimization of algorithm can improve camera correct accuracy.Camera parameter is divided into camera external parameter, in camera
Portion's parameter and camera distortion parameter, because radial distortion can be ignored on camera influence than other larger distortion.Therefore
The radial distortion parameter of camera is only calculated in MATLAB tool boxes.
If three-dimensional coordinate point is, corresponding two-dimensional camera plane pixel coordinates point is
,, gridiron pattern plane Z=0 is made,, wherein, k is
The Intrinsic Matrix of camera, s is scale factor, and R and t are the rotation and translation vector under camera coordinates system respectively, and H is single
Answering property matrix.
Three-dimensional coordinate point and image pixel coordinates point are, it is known that according to matrix solving method, when picture number is more than or equal to 3, k can
To obtain unique solution, Camera extrinsic can be obtained according to formula, camera radial distortion system can be drawn using radial distortion estimation
Number.
Further, after the monocular-camera obtains road video image, image can be carried out by geometric correction method
Correction, for example：Image picpointed coordinate can be pre-established（Row, column number）And object space（Or with reference to figure）Mapping between correspondence point coordinates
Relation, solution seeks the unknown parameter in mapping relations, and then according to mapping relations, to image, each pixel coordinate is corrected；Finally
Determine the gray value of each pixel.
Further, in the step S200, after road video image is corrected, using unstructured road rim detection
Algorithm is handled it, is obtained first and is taken aim at a little in advance, it specifically includes step：
S210, gray processing and noise suppression preprocessing are carried out to the road video image；
Specifically, the coloured image of the road video is changed into gray level image first, can be specifically entered using weighted mean method
Row is changed, and conversion formula is：, i.e., tri- components of R, G, B are carried out with different weights
Weighted average；It is minimum to the susceptibility of blueness due to susceptibility highest of the human eye to green, therefore, can be according to the formula pair
RGB three-components, which are weighted, averagely obtains rational gray level image；Further, bilateral filtering is carried out to the road video image
Processing, so as to reach the purpose for protecting side denoising.
S220, road video image edge is detected using Canny algorithms and morphology amendment is carried out；
Specifically, Edge-Detection Algorithm has Sobel, Prewitt, Laplace and Canny etc. a variety of, and the present invention is preferably
Canny algorithms detect road video image edge and carry out morphology amendment, specifically, first carry out expansion process to image and obtain
To continuous road edge, the noise in image is then removed by corrosion treatment.
S230, road video image edge is extracted using Hough transformation and calculates first and takes aim at a little in advance.
Specifically, the edge of road video image can be extracted by Hough transformation and calculate first it is pre- take aim at a little, after
And road axis is obtained, as shown in Figure 3.
Further, in the present invention, the step S300, using unstructured road partitioning algorithm to the road after correction
Video image is handled, and is obtained second and is taken aim at a little in advance, specifically includes step：
S310, the color characteristic and space characteristics for extracting road video image in advance；
Specifically, color characteristic is a kind of global characteristics, describes the superficiality of image or the scenery corresponding to image-region
Matter, goes to extract the color characteristic of image using color histogram；Space characteristics refer to multiple targets for being split in image it
Between mutual locus or relative direction relation, these relations can be divided into connection/syntople, overlap/overlapping relation and
Comprising/containment relationship etc., the present invention marks off the object or face included in image first by being split automatically to image
Color region, then according to these extracted region images features.
S320, using SLIC algorithms to road video image carry out segmentation obtain super-pixel data；
Specifically, the step S320 includes：
S321, initialization clustering processing is carried out to the pixel in road video image obtain some initial seed points；
S322, in road video image 3*3 fields reselect seed point；
The nearest some pixels of S323, the seed point reselected described in detection range in the range of the 2S*2S, and to search
To pixel label be set and be classified as a class；
S324, when same pixel simultaneously be classified into multiple seed points when, then calculate the pixel and multiple seed points
The distance between, using apart from the corresponding seed point of minimum value as the pixel cluster centre；
S325, the iterative processing that step S222-S224 is carried out to the cluster centre, until error convergence, are finally split
Super-pixel data.
Specifically, image is transformed into CIE-Lab color spaces from RGB color, each pixel of correspondence (L, a,
B) color value and（X, y）Coordinate constitutes a five dimensional vector V [L, a, b, x, y], the similitudes of two pixels can by they
Vector distance is measured, and distance is bigger, and similarity is smaller；
It is of the invention to carry out some initial seed points of initialization clustering processing acquisition to the pixel in road video image in advance, so
Seed point is reselected in road video image 3*3 pixel neighborhoods afterwards, it is then empty around each seed point reselected
Between in the nearest some pixels of the seed point that reselects described in detection range, the pixel searched is set into label and returned
For a class, all sort out until all pixels and finish；Specifically, a distance threshold can be set, when searching a pixel
When the distance between point and seed point for currently reselecting are less than the distance threshold, then the pixel is classified as the kind
Son one class of point；Assuming that the seed point reselected has K, road video image has N number of pixel, then each seed point is formed
Cluster can include N/K pixel, the length of side each clustered substantially S=[N/K]*0.5；It is preferred that in the seed reselected
The pixel closest with it is searched in the range of point surrounding 2S*2S；
Further, when same pixel is classified into multiple seed points simultaneously, then the pixel and multiple seeds are calculated
Point the distance between, using apart from the corresponding seed point of minimum value as the pixel cluster centre；Simultaneously to the cluster
Center is iterated processing, until error convergence, obtains last cluster, i.e. super-pixel data.
S330, road video image is clustered using K-means clustering algorithms and second is calculated and is taken aim at a little in advance；Tool
Body, the step S330 includes：
S331, random several cluster center of mass point are chosen from the super-pixel data；
Each super-pixel data is divided into the center of mass point closest with it, formed by S332, the traversal super-pixel data
Cluster；
Specifically, another distance threshold is set, the distance between the center of mass point and each super-pixel data are calculated, when this
When distance value is less than distance threshold, then the super-pixel data are divided into center of mass point formation cluster.
S333, the average value that each clusters of calculating and as new center of mass point；
Specifically, the average vector value of all pixels point in each cluster is calculated, new center of mass point is regained.
S334, the iterative processing for repeating step S332- S333, until center of mass point convergence, take aim in advance so as to obtain second
Point.
Specifically, road image is clustered by the K-means clustering algorithms based on SLIC, can be by road image
Data are divided into wheeled region and non-running region, as shown in Figure 4.
Further, in the step S400, ifFor
Optimal fitting curve, wherein,It is undetermined coefficient, is converted into the problem of solution optimal curve and seeks undetermined coefficient
Problem；The formula of the least square method is：；
；Wherein, pointIt is many
Meta-function, and meet equationObtain minimum；K=0,1 ... n；WhenDuring linear independence,It is exactly required least square solution；As imaginary center line.
Taken aim in advance to described first a little by using the least square method and second takes aim at a progress process of fitting treatment in advance, just can obtained
To the imaginary center line of road image.
Further, the present invention is verified by a specific embodiment to the road image data after processing；By the void
Intend center line angle information to send into the decision-making module of unmanned cruiser slave computer, provide the steering wheel of the unmanned cruiser
Angle as shown in figure 5, the unmanned cruiser in the process of moving, section position as shown in Figure 6 and Figure 7, send slave computer determine
Plan module beats angular data as shown in Figure 8 and Figure 9, and Figure 10 and Figure 11 are illustrated respectively in the navigation of position shown in section Fig. 6 and Fig. 7
The comparison diagram of driving trace and the driving trace obtained using imaging sensor；Figure 12 and Figure 13 are respectively on the outside of section and road
Section inner side uses imaging sensor and navigation traveling contrast trajector deviation figure.
Found by contrasting, the present invention utilizes monocular cam merely, and realizing road in less error range leads
Boat, and compared to imaging sensor is used, navigation mode of the invention is more accurate, therefore the present invention has huge application valency
Value.
Further, the present invention also provides a kind of unstructured road detecting system based on image information fusion, wherein, such as
Shown in Figure 14, including：
Correction module 100, school is carried out for obtaining road video image in real time by video camera, and to the road video image
Just；
First computing module 200, for being entered using unstructured road edge detection algorithm to the road video image after correction
Row processing, obtains first and takes aim at a little in advance；
Second computing module 300, at using unstructured road partitioning algorithm to the road video image after correction
Reason, obtains second and takes aim at a little in advance；
Fitting module 400, for being a little a little fitted with the second pre- take aim at using take aim at pre- to described first of least square method, is obtained
Road imaginary center line.
The described unstructured road detecting system based on image information fusion, wherein, first computing module 200
Specifically include：
Pretreatment unit, for carrying out gray processing and noise suppression preprocessing to the road video image；
Amending unit, for detecting road video image edge using Canny algorithms and carrying out morphology amendment；
First computing unit, takes aim at a little in advance for being extracted road video image edge using Hough transformation and being calculated first.
The described unstructured road detecting system based on image information fusion, wherein, second computing module 300
Specifically include：
Extraction unit, color characteristic and space characteristics for extracting road video image in advance；
Cutting unit, super-pixel data are obtained for carrying out segmentation to road video image using SLIC algorithms；
Second computing unit, for being clustered and being calculated second to road video image using K-means clustering algorithms
Take aim at a little in advance.
In summary, the present invention passes through non-structural road-edge detection algorithm and unstructured road partitioning algorithm point first
The first pre- take aim at for not calculating road video image is a little taken aim at a little in advance with second, then pre- to described first by information fusion technology
Take aim at a little and second takes aim at the imaginary center line for being a little fitted and obtaining road video image in advance；The present invention can by above-mentioned calculating processing
Obtain stabilization, accurate reliable view data；Further, described image data application can be navigated in unmanned cruiser,
Research cost can be reduced, with huge application value.
It should be appreciated that the application of the present invention is not limited to above-mentioned citing, for those of ordinary skills, can
To be improved or converted according to the above description, all these modifications and variations should all belong to the guarantor of appended claims of the present invention
Protect scope.
Claims (10)
1. a kind of unstructured road detection method based on image information fusion, it is characterised in that including step：
A, obtain road video image in real time by video camera, and the road video image is corrected；
B, using unstructured road edge detection algorithm the road video image after correction is handled, obtain first and take aim in advance
Point；
C, using unstructured road partitioning algorithm the road video image after correction is handled, obtain second and take aim at a little in advance；
D, using least square method it is pre- to described first take aim at a little to take aim in advance with second be a little fitted, obtain road virtual center line.
2. the unstructured road detection method according to claim 1 based on image information fusion, it is characterised in that institute
Step B is stated to specifically include：
B1, gray processing and noise suppression preprocessing are carried out to the road video image；
B2, road video image edge is detected using Canny algorithms and morphology amendment is carried out；
B3, road video image edge is extracted using Hough transformation and calculates first and takes aim at a little in advance.
3. the unstructured road detection method according to claim 1 based on image information fusion, it is characterised in that institute
Step C is stated to specifically include：
C1, the color characteristic and space characteristics for extracting road video image in advance；
C2, using SLIC algorithms to road video image carry out segmentation obtain super-pixel data；
C3, road video image is clustered using K-means clustering algorithms and second is calculated and is taken aim at a little in advance.
4. the unstructured road detection method according to claim 3 based on image information fusion, it is characterised in that institute
Step C2 is stated to specifically include：
C21, initialization clustering processing is carried out to the pixel in road video image obtain some initial seed points；
C22, in road video image 3*3 fields reselect seed point；
The nearest some pixels of C23, the seed point reselected described in detection range in the range of the 2S*2S, and to search
To pixel label be set and be classified as a class；
C24, when same pixel is classified into multiple seed points simultaneously, then calculate the pixel and multiple seed points it
Between distance, using apart from the corresponding seed point of minimum value as the pixel cluster centre；
C25, the iterative processing that step C22-C24 is carried out to the cluster centre, until error convergence, what is finally split is super
Pixel data.
5. the unstructured road detection method according to claim 3 based on image information fusion, it is characterised in that institute
Step C3 is stated to specifically include：
C31, random several cluster center of mass point are chosen from the super-pixel data；
Each super-pixel data is divided into the center of mass point closest with it, formed by C32, the traversal super-pixel data
Cluster；
C33, the average value that each clusters of calculating and as new center of mass point；
C34, repeating said steps C32-C33, until center of mass point convergence, take aim at a little in advance so as to obtain second.
6. the unstructured road detection method according to claim 1 based on image information fusion, it is characterised in that described
The formula of least square method is in step D：；
；Wherein, pointIt is
The function of many variables, and meet equationObtain minimum；K=0,1 ... n；WhenDuring linear independence,It is exactly required least square solution.
7. the unstructured road detection method according to claim 2 based on image information fusion, it is characterised in that institute
Stating the processing formula of the gray processing in step B1 is：。
8. a kind of unstructured road detecting system based on image information fusion, it is characterised in that including：
Correction module, is corrected for obtaining road video image in real time by video camera, and to the road video image；
First computing module, at using unstructured road edge detection algorithm to the road video image after correction
Reason, obtains first and takes aim at a little in advance；
Second computing module, for being handled using unstructured road partitioning algorithm the road video image after correction,
Second is obtained to take aim at a little in advance；
Fitting module, for being a little a little fitted with the second pre- take aim at using take aim at pre- to described first of least square method, obtains road
Imaginary center line.
9. the unstructured road detecting system according to claim 8 based on image information fusion, it is characterised in that institute
The first computing module is stated to specifically include：
Pretreatment unit, for carrying out gray processing and noise suppression preprocessing to the road video image；
Amending unit, for detecting road video image edge using Canny algorithms and carrying out morphology amendment；
First computing unit, takes aim at a little in advance for being extracted road video image edge using Hough transformation and being calculated first.
10. the unstructured road detecting system according to claim 8 based on image information fusion, it is characterised in that
Second computing module is specifically included：
Extraction unit, color characteristic and space characteristics for extracting road video image in advance；
Cutting unit, super-pixel data are obtained for carrying out segmentation to road video image using SLIC algorithms；
Second computing unit, for being clustered and being calculated second to road video image using K-means clustering algorithms
Take aim at a little in advance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710330557.5A CN107301371A (en) | 2017-05-11 | 2017-05-11 | A kind of unstructured road detection method and system based on image information fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710330557.5A CN107301371A (en) | 2017-05-11 | 2017-05-11 | A kind of unstructured road detection method and system based on image information fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107301371A true CN107301371A (en) | 2017-10-27 |
Family
ID=60137078
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710330557.5A Pending CN107301371A (en) | 2017-05-11 | 2017-05-11 | A kind of unstructured road detection method and system based on image information fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107301371A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108052559A (en) * | 2017-12-01 | 2018-05-18 | 国电南瑞科技股份有限公司 | Distribution terminal defect mining analysis method based on big data processing |
| CN110084190A (en) * | 2019-04-25 | 2019-08-02 | 南开大学 | Unstructured road detection method in real time under a kind of violent light environment based on ANN |
| CN111727437A (en) * | 2018-01-08 | 2020-09-29 | 远见汽车有限公司 | Multispectral system providing pre-crash warning |
| CN117648905A (en) * | 2024-01-30 | 2024-03-05 | 珠海芯烨电子科技有限公司 | Method and related device for analyzing label instruction of thermal printer |
-
2017
- 2017-05-11 CN CN201710330557.5A patent/CN107301371A/en active Pending
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108052559A (en) * | 2017-12-01 | 2018-05-18 | 国电南瑞科技股份有限公司 | Distribution terminal defect mining analysis method based on big data processing |
| CN111727437A (en) * | 2018-01-08 | 2020-09-29 | 远见汽车有限公司 | Multispectral system providing pre-crash warning |
| CN110084190A (en) * | 2019-04-25 | 2019-08-02 | 南开大学 | Unstructured road detection method in real time under a kind of violent light environment based on ANN |
| CN110084190B (en) * | 2019-04-25 | 2024-02-06 | 南开大学 | Real-time unstructured road detection method under severe illumination environment based on ANN |
| CN117648905A (en) * | 2024-01-30 | 2024-03-05 | 珠海芯烨电子科技有限公司 | Method and related device for analyzing label instruction of thermal printer |
| CN117648905B (en) * | 2024-01-30 | 2024-04-16 | 珠海芯烨电子科技有限公司 | Method and related device for analyzing label instruction of thermal printer |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110569704B (en) | Multi-strategy self-adaptive lane line detection method based on stereoscopic vision | |
| CN109460709A (en) | The method of RTG dysopia analyte detection based on the fusion of RGB and D information | |
| CN107844750A (en) | A kind of water surface panoramic picture target detection recognition methods | |
| CN107677274B (en) | Unmanned plane independent landing navigation information real-time resolving method based on binocular vision | |
| CN108369741A (en) | Method and system for registration data | |
| CN107301371A (en) | A kind of unstructured road detection method and system based on image information fusion | |
| CN108280401B (en) | Pavement detection method and device, cloud server and computer program product | |
| CN106183995B (en) | A kind of visual parking device method based on stereoscopic vision | |
| CN109993800A (en) | A kind of detection method of workpiece size, device and storage medium | |
| CN111738314A (en) | Deep learning method of multi-modal image visibility detection model based on shallow fusion | |
| CN104933434A (en) | Image matching method combining length between perpendiculars (LBP) feature extraction method and surf feature extraction method | |
| CN108597009A (en) | A method of objective detection is carried out based on direction angle information | |
| Oniga et al. | Polynomial curb detection based on dense stereovision for driving assistance | |
| CN110766720A (en) | Multi-camera vehicle tracking system based on deep learning | |
| CN110059683A (en) | A kind of license plate sloped antidote of wide-angle based on end-to-end neural network | |
| CN112164117A (en) | V-SLAM pose estimation method based on Kinect camera | |
| CN110428425B (en) | Sea-land separation method of SAR image based on coastline vector data | |
| CN113781562B (en) | Lane line virtual-real registration and self-vehicle positioning method based on road model | |
| CN112344923B (en) | Robot positioning method and positioning device thereof | |
| CN111998862A (en) | Dense binocular SLAM method based on BNN | |
| CN109164802A (en) | A kind of robot maze traveling method, device and robot | |
| CN107220632B (en) | Road surface image segmentation method based on normal characteristic | |
| CN116978009A (en) | Dynamic object filtering method based on 4D millimeter wave radar | |
| CN110197104B (en) | Distance measurement method and device based on vehicle | |
| CN117611525A (en) | Visual detection method and system for abrasion of pantograph slide plate |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| DD01 | Delivery of document by public notice |
Addressee: Xue Liantong Document name: Notice of commencement of preservation procedure |
|
| DD01 | Delivery of document by public notice |