The content of the invention
For drawbacks described above present in prior art, the present invention provide a kind of method for identifying surface deformation class disease andDevice.
An aspect of of the present present invention provides a kind of method for identifying surface deformation class disease, including：S1, according to what is obtained in advanceThe three-dimensional data on road surface to be detected, obtain the deforming depth information on the road surface to be detected；S2, believed according to the deforming depthBreath, doubtful disease edge is extracted by edge detection algorithm；S3, according to the deforming depth information and the doubtful disease edgeJust positioning disease region is obtained, and disease region is positioned by region growing reduction treatment；S4, obtain the disease in the disease regionEvil attribute set, and according to the disease attribute set, identify the Damage Types in the disease region.
Wherein, the step S1 further comprises：S11, abnormality value removing processing, data mark are carried out to the three-dimensional dataFixed processing and filtering process, obtain control of section profile；S12, the contour feature of the control of section profile is analyzed, obtainedTake section nominal contour corresponding with the control of section profile；S13, by the control of section profile and the section standard wheelsIt is poor that exterior feature is made, and obtains the actual elevation of section on the road surface to be detected and the measuring height of section difference information of standard elevation；S14, according to instituteMeasuring height of section difference information is stated, continuous multiple sections are subjected to splicing, obtain the deforming depth information on the road surface to be detected.
Wherein, the step S3 further comprises：S31, the deforming depth information is divided into multiple images sub-block, obtainedTake depth attribute information corresponding to the multiple image subblock difference；Wherein, the depth attribute information include depth-averaged value,Depth value variance and depth value distributed intelligence；S32, it is corresponding according to the multiple image subblock with reference to deformation class disease knowledge baseDepth attribute information and the doubtful disease edge, the confidence level of the multiple image subblock is assessed, will be higherThe image subblock of confidence level is as the just positioning disease region；S33, connected region mark is carried out to the just positioning disease regionNote, obtain the region parameter of each connected region；Wherein, the region parameter include zone length, region area, regional location andRegion direction；S34, meet that the connected region of preparatory condition carries out edge extension to the region parameter, obtain the diseaseEvil region.
Wherein, the step S4 further comprises：S41, obtain effective length, effective width, the length in the disease regionWidth than and disease position, and obtain according to the effective length and the effective width depth capacity, flat in the disease regionEqual depth and disease area；The effective length of the disease attribute set including the disease region, effective width, length-width ratio,Disease position, depth capacity, mean depth and disease area；S42, according to the disease attribute set, based on deformation class diseaseKnowledge base, obtain the Damage Types that the disease region includes disease.
Wherein, the step S34 further comprises：S341, the of predetermined threshold value is more than to zone length and region areaOne connected region, edge extension is carried out along the region direction of first connected region；S342, if having second in elongated areaConnected region, and the distance between first connected region and second connected region are less than pre-determined distance, then obtain instituteState the line of centres of the first connected region and second connected region；S343, along the direction of the line of centres, by describedOne connected region and second connected region merge, and the region for merging to obtain is as new disease region.
Wherein, the step S2 further comprises：S21, believed according to default edge detection operator and the deforming depthBreath, obtain the gradient component of each pixel；S22, the pixel that gradient component is more than predetermined gradient threshold value is obtained, as doubtfulDisease marginal point, the doubtful disease edge is obtained according to the doubtful disease marginal point.
Wherein, the step S42 further comprises：S421, grader is obtained according to the deformation class disease knowledge base；S422, according to the disease attribute set, based on disease linear character, depth characteristic, face battle array feature, position feature and deformation journeyDegree, the Damage Types of disease are included by the grader acquisition disease region；Wherein, the linear character includesImitate length, effective width, length-width ratio；The depth characteristic includes depth capacity and mean depth；The face battle array feature includes diseaseEvil area；The position feature includes disease position；The deformation extent includes slight deformation, moderate deformation and severe deformation.
Another aspect of the present invention provides a kind of device for identifying surface deformation class disease, including：Acquisition module, for rootAccording to the three-dimensional data on the road surface to be detected obtained in advance, the deforming depth information on the acquisition road surface to be detected；Processing module, useAccording to the deforming depth information, doubtful disease edge is extracted by edge detection algorithm；Locating module, for according toDeforming depth information and the doubtful disease edge obtain just positioning disease region, and position disease by region growing reduction treatmentEvil region；Identification module, for obtaining the disease attribute set in the disease region, and according to the disease attribute set, knowThe Damage Types in not described disease region.
Another aspect of the present invention provides a kind of equipment for identifying surface deformation class disease, including：At least one processor；And at least one memory being connected with the processor communication, wherein：The memory storage has can be by the processorThe programmed instruction of execution, the processor call described program instruction to be able to carry out the identification road surface that the above-mentioned aspect of the present invention providesThe method for deforming class disease, such as including：S1, according to the three-dimensional data on the road surface to be detected obtained in advance, obtain described to be checkedSurvey the deforming depth information on road surface；S2, according to the deforming depth information, doubtful disease side is extracted by edge detection algorithmEdge；S3, just positioning disease region, and pass through region growing is obtained according to the deforming depth information and the doubtful disease edgeReduction treatment positioning disease region；S4, the disease attribute set in the disease region is obtained, and according to the disease property setClose, identify the Damage Types in the disease region.
Another aspect of the present invention provides a kind of non-transient computer readable storage medium storing program for executing, and the non-transient computer is readableStorage medium stores computer instruction, and the computer instruction makes the computer perform the identification that the above-mentioned aspect of the present invention providesThe method of surface deformation class disease, such as including：S1, according to the three-dimensional data on the road surface to be detected obtained in advance, described in acquisitionThe deforming depth information on road surface to be detected；S2, according to the deforming depth information, doubtful disease is extracted by edge detection algorithmEdge；S3, just positioning disease region, and being given birth to by region is obtained according to the deforming depth information and the doubtful disease edgeLong reduction treatment positioning disease region；S4, the disease attribute set in the disease region is obtained, and according to the disease property setClose, identify the Damage Types in the disease region.
The method and apparatus of identification surface deformation class disease provided by the invention, by obtaining disease depth information and deformationDepth image, the disease region in road pavement are accurately extracted, and ensure that the integrality in disease region；Also pass through diseaseAttribute is identified exactly to the Damage Types in disease region, is realized and is efficiently and accurately detected and identify surface deformationClass disease.
Embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present inventionIn accompanying drawing, the technical scheme in the embodiment of the present invention is explicitly described, it is clear that described embodiment be the present inventionPart of the embodiment, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art are not havingThe every other embodiment obtained under the premise of creative work is made, belongs to the scope of protection of the invention.
At present, dimensional Modeling Technology has been widely used for every field, from Land resources investigation, three-dimensional visualization, threeDimension animation, high-precision three-dimensional modeling are used widely to 3 D-printing.According to laser triangulation principle, based on cable architectureThe method of light combination vision sensor measurement realizes same posture, the synchro measure of synchronization, that is, requires that one-shot measurement samplesOne complete section, ensure that a section completes measurement under same posture, surveyed based on line-structured light combination vision sensorThe three dimensional point cloud that amount obtains can accurately obtain high-precision road surface section contoured three-dimensional information, while also contains road surface diseaseHarmful two-dimensional signal, so as to which three dimensional point cloud relatively can directly and easily obtain deformation class disease complete information, including deformationPosition, Damage Types and degree of disease etc..The embodiment of the present invention is that the three-dimensional data road pavement disease based on road surface is identified.
Fig. 1 is the schematic flow sheet of the method for identification surface deformation class disease provided in an embodiment of the present invention, such as Fig. 1 institutesShow, including：S1, according to the three-dimensional data on the road surface to be detected obtained in advance, the deforming depth for obtaining the road surface to be detected is believedBreath；S2, according to the deforming depth information, doubtful disease edge is extracted by edge detection algorithm；S3 is deep according to the deformationSpend information and the doubtful disease edge obtains just positioning disease region, and disease area is positioned by region growing reduction treatmentDomain；S4, the disease attribute set in the disease region is obtained, and according to the disease attribute set, identify the disease regionDamage Types.
Wherein, three-dimensional data is to measure resulting number to road surface to be detected by line scanning three-dimensional measurement sensorAccording to.Specifically, line scanning three-dimensional measurement sensor can measure to obtain measured object surface elevation with respect to situation of change, and the three of acquisitionDimension data can reflect the elevation change information on measured object surface.Line scanning three-dimensional measurement sensor can realize same posture,The profiled outline synchro measure of synchronization, acquisition mode include two ways：Fixed first, three-dimensional measurement sensor is arranged onOn support, in three-dimensional measurement sensor measurement range, testee passes through measured zone with certain speed, is transported in testeeDuring dynamic, the three-D profile data acquisition to testee is realized；Second, three-dimensional measurement sensor is arranged on motion carrierOn, during carrier movement is measured, data acquisition is carried out to testee three-D profile.In the embodiment of the present invention, by lineScanning three-dimensional measurement sensor is arranged on vehicle, is travelled one time on road surface to be detected and is said exemplified by obtaining three-dimensional dataIt is bright, but the protection domain of the embodiment of the present invention is not limited to this.
Wherein, the deforming depth message reflection entirely depth information at road surface diverse location to be detected, such as with coordinateForm characterizes depth value corresponding to each point and each point on road surface to be detected；Deforming depth image then reflects whole road surface to be detectedThe image of depth information at diverse location, the image can intuitively reflect the change in depth situation on whole road surface to be detected.
Wherein, edge is that the most obvious place of grey scale change, edge detection algorithm identify number using this feature on imageBrightness changes obvious point in word image, and the significant changes in image attributes generally reflect the critical event and change of attribute.The edge in the edge reflection disease region in the embodiment of the present invention.
Wherein, disease attribute set includes multiple diseases attribute, and disease attribute set can reflect from shape and depthThe feature in disease region；Make it possible to, according to disease attribute set, classify to the disease in disease region.
In step sl, line scanning three-dimensional measurement sensor can be used to carry out three-dimensional data to road surface to be detected to adoptCollection；After collection, three-dimensional data is handled, further obtains the deforming depth information and deforming depth figure on road surface to be detectedPicture.Fig. 2 is the deforming depth image (a) of the pit disease of the method for identification surface deformation class disease provided in an embodiment of the present inventionSchematic diagram, as shown in Fig. 2 by taking pit disease as an example, on road surface to be detected deeper position, has deeper color, with thisDistinguish the depth value at the diverse location on road surface to be detected.
In step s 2, the deforming depth information obtained according to step S1, deforming depth is believed using edge detection algorithmBreath is solved, and obtained edge is doubtful disease edge；Doubtful disease edge can reflect disease area to a certain extentDomain.
In step s3, according to the step S1 deforming depth information obtained and the doubtful disease side obtained in step S2Edge, just positioning disease region is obtained first；Then the processing reduced by region growing can obtain disease region.Wherein, areaDomain growth reduction is the characteristics of having region clustering and change in depth continuity using surface deformation class disease, to utilize disease areaTopological relation extraction disease region between domain.
In step s 4, the disease region obtained according to step S3, obtains the disease attribute set in the disease region；DiseaseThe features such as the shape and depth in evil attribute set reflection disease region, can be to the disease class in disease region by the attribute setDo not efficiently identified.
The method of identification surface deformation class disease provided in an embodiment of the present invention, by obtaining disease depth information and deformationDepth image, the disease region in road pavement are accurately extracted, and ensure that the integrality in disease region；Also pass through diseaseAttribute is identified exactly to the Damage Types in disease region, is realized and is efficiently and accurately detected and identify surface deformationClass disease.
On the basis of any of the above-described embodiment, the step S1 further comprises：S11, the three-dimensional data is carried outAbnormality value removing processing, data scaling processing and filtering process, obtain control of section profile；S12, to the control of section profileContour feature analyzed, obtain corresponding with control of section profile section nominal contour；S13, by the section controlIt is poor that the wide and described section nominal contour of ratch is made, and obtains the actual elevation of section on the road surface to be detected and the section of standard elevationDepth displacement information；S14, according to the measuring height of section difference information, continuous multiple sections are subjected to splicing, obtained described to be checkedSurvey the deforming depth information on road surface.
Wherein, three-dimensional data is made up of multiple continuous profiled outlines, and each profiled outline mainly contains sectionNominal contour, control of section profile, pavement texture；Wherein, section nominal contour represents normal road surface profile, includes measurement appearanceState (vehicle-mounted three dimension system is due to cross section tilt phenomenon caused by the factors such as roll), but do not include both macro and micro disease profileInformation is (such as：Pit, rut, depression, gather around bag and crack etc.)；Control of section profile is is bonded the profile of road surface data, comprising grandSee disease profile information (such as：Pit, rut, depression and gather around bag etc.), but not comprising microcosmic defect information (such as：Crack etc.)；RoadFace texture is the granuloplastic profile of ground surface material locally tiny fluctuating under normal circumstances.
Wherein, abnormality value removing processing is to remove the value of the apparent error included in three-dimensional data；Data scaling is handledImage space altitude data is converted into object space altitude data, so as to accurately obtain the TP of testee.
Wherein, filtering can use mean filter, gaussian filtering etc., and by taking mean filter as an example, it is typical linear filteringAlgorithm, it refers on image to object pixel to a template, and the template includes adjacent pixels around it (with target picture8 pixel around centered on element, a Filtering Template is formed, that is, removes object pixel in itself), then with all pictures in templateThe average value of element replaces original pixel value.
In step s 11, the three-dimensional data gathered in advance is pre-processed, pretreatment include abnormality value removing processing withData scaling processing；Fig. 3 is the pit disease section wheel of the method for identification surface deformation class disease provided in an embodiment of the present inventionWide schematic diagram, Fig. 4 are that the bag disease profiled outline of gathering around of the method for identification surface deformation class disease provided in an embodiment of the present invention showsIt is intended to, the control of section profile of pit disease is as shown in figure 3, the control of section profile for gathering around bag disease is as shown in Figure 4.
In step s 12, according to the control of section profile obtained in step S11, because control of section profile reflects road surfaceTruth, it is therefore desirable to the contour feature of control of section profile is analyzed, obtains and reflects the disconnected of normal surface conditionsFace nominal contour；Wherein, contour feature can move towards the features such as trend, curvature, peak point including control of section lines of outline.Such as the section nominal contour shown in Fig. 3, because control of section profile is to lower recess, therefore can determine to break to a certain extentFace nominal contour should be on the top of downward peak point；And section nominal contour as shown in Figure 4, due to control of section profileRaise up, therefore can determine that section nominal contour should be in the bottom of upward peak point to a certain extent.
In step s 13, the control of section profile and section nominal contour obtained according to step S12, due to control of sectionProfile contains the profile information and section nominal contour of section disease；Therefore, with section nominal contour SPjSubtract section controlRatch exterior feature CPjIt can obtain measuring height of section difference information Fj={ Fji| i=1,2 ..., n }, (n is single section survey point number).
, will be continuous multiple on road surface to be detected according to the measuring height of section difference information obtained in step S13 in step S14Section is stitched together, and constitutes the deforming depth information F={ F on road surface to be detectedji| j=1,2 ..., m, i=1,2 ..., n }；(wherein, m is splicing section number, and n is single section survey point number)；Deforming depth image then be and deforming depth information pairThe image answered.
By above-mentioned steps, effectively the depth information on road surface to be detected can be extracted, ensure that integrality.
On the basis of any of the above-described embodiment, the step S3 further comprises：S31, by the deforming depth informationMultiple images sub-block is divided into, obtains depth attribute information corresponding to the multiple image subblock difference；Wherein, the depth categoryProperty information includes depth-averaged value, depth value variance and depth value distributed intelligence；S32, with reference to deformation class disease knowledge base, foundationDepth attribute information corresponding to the multiple image subblock and the doubtful disease edge, put to the multiple image subblockReliability is assessed, using the image subblock of high confidence as the just positioning disease region；S33, to the just positioning diseaseEvil region carries out connected component labeling, obtains the region parameter of each connected region；Wherein, the region parameter is grown including regionDegree, region area, regional location and region direction；S34, meet that the connected region of preparatory condition is entered to the region parameterRow edge extends, and obtains the disease region.
It is in step S31, the deforming depth information on road surface to be detected or deforming depth image classifying rationally is mutual into sizeNonoverlapping image subblock (size sm*sn), SU={ SUxy| x=1,2 ..., M, y=1,2 ..., N }, SUxy={ Fji|j∈Xx,i∈Yy, wherein M=m/sm is the number of image subblock in the row direction；N=n/sn be image subblock in a column directionNumber；Xx∈ [(x-1) * sm+1x*sm] and Xx∈ { 1,2 ..., m }, Yy∈ [(y-1) * sn+1y*sn] and Yy∈{1,2,…,n}。
After being divided into multiple images sub-block, by carrying out statistics and analysis to the depth value in each image subblock SU, obtainThe depth value attribute information in image subblock is taken, depth value attribute information can include depth-averaged value, depth value variance and depthAngle value distributed intelligence, wherein, distributed intelligence can be positive and negative distributed intelligence；For example, depth value attribute information can be SUA={ average AVGxy, variance STDxy, positive and negative distributed intelligence PMxy| x=1,2 ..., M, y=1,2 ..., N }, wherein, M is image subblockIn the number of line direction, N is the number of image subblock in a column direction.
In step s 32, deformation class disease knowledge base stores all kinds of diseases (such as the car that deformation class disease includesRut, gather around bag, pit and depression etc.) shape facility, depth characteristic etc., actual parameter is compared and matched with knowledge base energyDamage Types are confirmed to a certain extent.Specifically, according to deformation class disease knowledge base, with reference to image subblock SU, depth attributeInformation SUA and doubtful disease edge, are assessed the confidence level of each image subblock, obtain the higher image subblock of confidence level(confidence level is higher to show that the possibility that the image subblock is disease region is larger) and its positional information (such as coordinate value), are realizedDisease just positions, using the image subblock of the high confidence of acquisition as just positioning disease region；Fig. 2 embodiment of the present invention providesIdentification surface deformation class disease method pit disease first positioning disease region (b) schematic diagram, using pit disease asIt is illustrated just positioning disease region.
In step S33, according to the first positioning disease region obtained in step S32, just positioning disease region is connectedLogical zone marker, recording mark value is FR={ FRji| j=1,2 ..., m, i=1,2 ..., n } (wherein, m is of splicing sectionNumber, n are the measurement point number of single section), and count each connected region UR in connected component labeling image FRu(mark value is u'sConnected region, u=1,2 ..., U；U is the total number of connected region) region parameter, region parameter includes：Zone lengthURLu, region area URAu, regional location URSuWith region direction URDu.Wherein, zone length URLuThe connection for being u for mark valueThe external square long side in region or cornerwise length；Region area URAuThe number of pixels for the connected region for being u for mark value；RegionPosition URSuFor the coordinate of connected region center of gravity；Region direction URDuLeast square fitting method such as can be used to obtain.
In step S34, according to the region parameter obtained in step S33, to each region parameter (such as zone lengthURLuWith region area URAu) meet preparatory condition (such as be respectively greater than default threshold value, wherein, threshold value is known by deformation class diseaseKnow storehouse to obtain) desired disease region carries out edge extension (region direction URD of the bearing of trend according to its geometric shapeuReallyIt is fixed), obtain final disease region.
On the basis of any of the above-described embodiment, the step S4 further comprises：S41, obtain the disease regionEffective length, effective width, length-width ratio and disease position, and the disease is obtained according to the effective length and the effective widthDepth capacity, mean depth and the disease area in evil region；The disease attribute set includes effective length in the disease regionDegree, effective width, length-width ratio, disease position, depth capacity, mean depth and disease area；S42, according to the disease attributeSet, based on deformation class disease knowledge base, obtain the Damage Types that the disease region includes disease.
In step S41, after extracting disease region, disease region DZ is utilizedk(k=1,2 ..., s, wherein s are diseaseTotal number) minimum area boundary rectangle, obtain effective length L, effective width W, length-width ratio LW and the disease position in disease regionLoc (position coordinates on most long line summit in disease region) is put, calculates depth capacity Hmax, the mean depth in disease regionHavgWith disease area Area；Obtain disease attribute set DZAk, (DZAk={ L,W,LW,Loc,Hmax,Havg, Area }, k=1,2 ..., s, s are the total number of disease).
In step S42, using class disease knowledge base is deformed, disease attribute set and the data in knowledge base are comparedPair and matching, and then identify the Damage Types in disease region.
On the basis of any of the above-described embodiment, the step S34 further comprises：S341, to zone length and regionArea is more than the first connected region of predetermined threshold value, and edge extension is carried out along the region direction of first connected region；S342,If the second connected region in elongated area be present, and first connected region and the distance between second connected region are smallIn pre-determined distance, then the line of centres of first connected region and second connected region is obtained；S343, along the centerThe direction of line, first connected region and second connected region are merged, the region for merging to obtain is as newDisease region.
Specifically, to each zone length URLuWith region area URAuMeeting threshold value, (threshold value is by deformation class disease knowledge baseObtain) desired disease region (i.e. the first connected region) progress edge extension (region of the bearing of trend according to its geometric shapeDirection URDuIt is determined that), judge to mark with the presence or absence of other diseases in the elongated area, if in the presence of the second connected region, and bothWith certain topological relation (i.e. closer to the distance, less than pre-determined distance), then the first connected region and the second connected region are obtainedThe line of centres, along the line of centres extend merge the two disease regions (the first connected region and the second connected region)；ByAfter repeatedly merging, terminate if being all not present around small disease region compared with major disease region, disease growth reduction process；Generation compared withThe confidence level in large-scale disease region is higher, is retained；The confidence level in the small range disease region not being merged is relatively low, willIt is rejected；Obtain final disease region DZk(k=1,2 ..., s, wherein s are the total number of disease).
On the basis of any of the above-described embodiment, the step S2 further comprises：S21, according to default rim detectionOperator and the deforming depth information, obtain the gradient component of each pixel；S22, obtain gradient component and be more than predetermined gradientThe pixel of threshold value, as doubtful disease marginal point, the doubtful disease edge is obtained according to the doubtful disease marginal point.
In the step s 21, because deformation class disease region has the larger feature of area, design of embodiment of the present invention 11*11 edge detection operator, road pavement deforming depth information F solve deformation class disease using the method for difference edge detectionDoubtful marginal point is to position disease region；Process is illustrated by taking 11*11 edge detection operator as an example below, rim detectionOperator is as follows：
According to above-mentioned edge detection operator, calculating gradient component is：
In formula, f (j, i) denotation coordination is the deforming depth value at (j, i) place, then the gradient component of pixel (j, i) is bigIt is small to be：
G(j,i)≈|Gj|+|Gi|
In step S22, according to the gradient component obtained in step S21, if G (j, i) ＞ Th, wherein, Th is default ladderSpend threshold value；Then G (j, i) is doubtful disease marginal point.
The proportionate relationship of the Grad and all pixels Grad at class disease edge, Ke Yishe are deformed by a large amount of geo-statisticsDetermine 80% that Th is greatest gradient value, obtain doubtful disease marginal point；All doubtful disease marginal point constitutes doubtful disease sideEdge.
On the basis of any of the above-described embodiment, the step S42 further comprises：S421, according to the deformation class diseaseEvil knowledge base obtains grader；S422, it is special based on disease linear character, depth characteristic, face battle array according to the disease attribute setSign, position feature and deformation extent, the Damage Types of disease are included by the grader acquisition disease region；ItsIn, the linear character includes effective length, effective width, length-width ratio；The depth characteristic includes depth capacity and average depthDegree；The face battle array feature includes disease area；The position feature includes disease position；The deformation extent includes slight becomeShape, moderate deformation and severe deformation.
The schematic diagram of the identification disease classification of the method for Fig. 5 identification surface deformation class diseases provided in an embodiment of the present invention,As shown in figure 5, combining disease knowledge-base design grader first, identification disease area is gone based on disease attributive character using graderThe type in domain；Disease attributive character includes disease linear character, depth characteristic, face battle array feature, position feature and deformation extent；ItsIn, linear character includes effective length, effective width, length-width ratio；Depth characteristic includes disease depth capacity and mean depth；FaceBattle array feature includes disease area；Position feature includes disease position；Deformation extent includes slight deformation, moderate deformation and severe and becomeShape.
Grader is based on disease attributive character, and disease is classified, such as can use following rule：For a certain diseaseEvil region,
Rule 1, threshold value Th is will be greater than according to disease depth informationup" upward class " disease be extracted as gathering around bag disease；
Rule 2, according to effective length (such as rut has obvious length characteristic), effective width (wheel width it is certainRatio), disease depth capacity (being more than threshold value T1max), mean depth (be more than threshold value T1mean), length-width ratio (be more than threshold value Tlw,TlwValue it is related to the sampling interval of data and resolution ratio), and disease present position meets the features such as wheel trajectories, is extracted asRut disease；
Rule 3, (is more than threshold value T2 according to disease depth capacitymax), mean depth (be more than threshold value T2mean), bonded areaFeature (area threshold Tarea) extract depression disease and pit disease；
, can be according to current after determining Damage Types (pit, depression, rut, gathering around bag) in addition, according to classification resultsDisease regional characteristics goes to improve deformation class disease knowledge base.
Fig. 6 is the structural representation of the device of identification surface deformation class disease provided in an embodiment of the present invention, such as Fig. 6 institutesShow, including：Acquisition module 601, for the three-dimensional data according to the road surface to be detected obtained in advance, obtain the road surface to be detectedDeforming depth information；Processing module 602, for according to the deforming depth information, being extracted by edge detection algorithm doubtfulDisease edge；Locating module 603, for obtaining just positioning disease according to the deforming depth information and the doubtful disease edgeRegion, and disease region is positioned by region growing reduction treatment；Identification module 604, for obtaining the disease in the disease regionEvil attribute set, and according to the disease attribute set, identify the Damage Types in the disease region.
Wherein it is possible to the collection of three-dimensional data is carried out to road surface to be detected using line scanning three-dimensional measurement sensor；CollectionAfterwards, acquisition module 601 is handled three-dimensional data, further obtains the deforming depth information and deforming depth on road surface to be detectedImage.Fig. 2 is the deforming depth image of the pit disease of the method for identification surface deformation class disease provided in an embodiment of the present invention(a) schematic diagram, as shown in Fig. 2 by taking pit disease as an example, on road surface to be detected deeper position, has deeper color, withThis distinguishes the depth value at the diverse location on road surface to be detected.
Wherein, the deforming depth information that processing module 602 obtains according to acquisition module 601, utilizes edge detection algorithm pairDeforming depth information is solved, and obtained edge is doubtful disease edge；To a certain extent can at doubtful disease edgeReflect disease region.
Wherein, obtained in locating module 603 obtains according to acquisition module 601 deforming depth information and processing module 602The doubtful disease edge taken, just positioning disease region is obtained first；Then the processing reduced by region growing can obtain diseaseEvil region.Wherein, region growing reduction is that have region clustering and change in depth continuity using surface deformation class diseaseFeature, disease region is extracted using the topological relation between disease region.
Wherein, the disease region that identification module 604 obtains according to locating module 603, the disease in the disease region is obtainedAttribute set；The features such as the shape and depth in disease attribute set reflection disease region, can be to disease by the attribute setThe disease classification in region is efficiently identified.
The device of identification surface deformation class disease provided in an embodiment of the present invention, by obtaining disease depth information and deformationDepth image, the disease region in road pavement are accurately extracted, and ensure that the integrality in disease region；Also pass through diseaseAttribute is identified exactly to the Damage Types in disease region, is realized and is efficiently and accurately detected and identify surface deformationClass disease.
On the basis of any of the above-described embodiment, the acquisition module 601 further comprises：Pretreatment unit, for pairThe three-dimensional data carries out abnormality value removing processing, data scaling processing and filtering process, obtains control of section profile；Analysis is singleMember, for analyzing the contour feature of the control of section profile, obtain section corresponding with the control of section profileNominal contour；Acquiring unit, it is poor for the control of section profile and the section nominal contour to be made, obtain described to be detectedThe actual elevation of section on road surface and the measuring height of section difference information of standard elevation；Concatenation unit, for poor according to the measuring height of sectionInformation, continuous multiple sections are subjected to splicing, obtain the deforming depth information on the road surface to be detected.
On the basis of any of the above-described embodiment, the locating module 603 further comprises：Division unit, for by instituteState deforming depth information and be divided into multiple images sub-block, obtain depth attribute information corresponding to the multiple image subblock difference；Wherein, the depth attribute information includes depth-averaged value, depth value variance and depth value distributed intelligence；Assessment unit, it is used forWith reference to deformation class disease knowledge base, according to depth attribute information corresponding to the multiple image subblock and the doubtful disease sideEdge, the confidence level of the multiple image subblock is assessed, using the image subblock of high confidence as the just positioning diseaseEvil region；Indexing unit, for carrying out connected component labeling to the just positioning disease region, obtain the region of each connected regionParameter；Wherein, the region parameter includes zone length, region area, regional location and region direction；Extension apparatus, it is used forMeet that the connected region of preparatory condition carries out edge extension to the region parameter, obtain the disease region.
On the basis of any of the above-described embodiment, the identification module 604 further comprises：Gather acquiring unit, be used forObtain effective length, effective width, length-width ratio, the depth capacity in disease region, mean depth and the disease in the disease regionArea and disease position；The disease attribute set includes effective length, effective width, length-width ratio, the disease in the disease regionEvil position, depth capacity, mean depth and disease area；Type acquiring unit, for according to the disease attribute set, being based onClass disease knowledge base is deformed, obtains the Damage Types that the disease region includes disease.
On the basis of any of the above-described embodiment, the extension apparatus further comprises：Extend subelement, for regionLength and region area are more than the first connected region of predetermined threshold value, and edge is carried out along the region direction of first connected regionExtension；Line subelement, if for the second connected region, and first connected region and described second in elongated area be presentThe distance between connected region is less than pre-determined distance, then obtains the center of first connected region and second connected regionLine；Merge subelement, for along the direction of the line of centres, by first connected region and second connected regionMerge, the region for merging to obtain is as new disease region.
On the basis of any of the above-described embodiment, the processing module 602 further comprises：Gradient acquiring unit, is used forAccording to default edge detection operator and the deforming depth information, the gradient component of each pixel is obtained；Edge obtains singleMember, the pixel of predetermined gradient threshold value is more than for obtaining gradient component, as doubtful disease marginal point, according to the doubtful diseaseEvil marginal point obtains the doubtful disease edge.
On the basis of any of the above-described embodiment, the type acquiring unit further comprises：Grader obtains subelement,For obtaining grader according to the deformation class disease knowledge base；Type obtains subelement, for according to the disease property setClose, based on disease linear character, depth characteristic, face battle array feature, position feature and deformation extent, institute is obtained by the graderState the Damage Types that disease region includes disease；Wherein, the linear character includes effective length, effective width, length-width ratio；The depth characteristic includes depth capacity and mean depth；The face battle array feature includes disease area；The position feature includesDisease position；The deformation extent includes slight deformation, moderate deformation and severe deformation.
Fig. 7 is the structural representation of the equipment of identification surface deformation class disease provided in an embodiment of the present invention, such as Fig. 3 institutesShow, the equipment includes：At least one processor 701；And at least one memory with the processor 701 communication connection702, wherein：The memory 702 is stored with the programmed instruction that can be performed by the processor 701, and the processor 701 callsThe method that described program instruction is able to carry out the identification surface deformation class disease that the various embodiments described above are provided, such as including：S1,According to the three-dimensional data on the road surface to be detected obtained in advance, the deforming depth information on the acquisition road surface to be detected；S2, according to instituteDeforming depth information is stated, doubtful disease edge is extracted by edge detection algorithm；S3, according to the deforming depth information and describedDoubtful disease edge obtains just positioning disease region, and positions disease region by region growing reduction treatment；S4, described in acquisitionThe disease attribute set in disease region, and according to the disease attribute set, identify the Damage Types in the disease region.
The embodiment of the present invention also provides a kind of non-transient computer readable storage medium storing program for executing, the non-transient computer readable storageMedium storing computer instructs, and the computer instruction makes computer perform the identification surface deformation class disease that corresponding embodiment is providedHarmful method, such as including：S1, according to the three-dimensional data on the road surface to be detected obtained in advance, obtain the road surface to be detectedDeforming depth information；S2, according to the deforming depth information, doubtful disease edge is extracted by edge detection algorithm；S3, according toThe deforming depth information and the doubtful disease edge obtain just positioning disease region, and determining by region growing reduction treatmentPosition disease region；S4, obtains the disease attribute set in the disease region, and according to the disease attribute set, described in identificationThe Damage Types in disease region.
The embodiments such as the equipment of identification surface deformation class disease described above are only schematical, divide wherein being used asUnit from part description can be or may not be it is physically separate, can be as the part that unit is shown orIt may not be physical location, you can with positioned at a place, or can also be distributed on multiple NEs.Can basisIt is actual to need to select some or all of module therein to realize the purpose of this embodiment scheme.Ordinary skill peopleMember is not in the case where paying performing creative labour, you can to understand and implement.
Through the above description of the embodiments, those skilled in the art can be understood that each embodiment canRealized by the mode of software plus required general hardware platform, naturally it is also possible to pass through hardware.Based on such understanding, onThe part that technical scheme substantially in other words contributes to prior art is stated to embody in the form of software product, shouldComputer software product can store in a computer-readable storage medium, such as ROM/RAM, magnetic disc, CD, including some fingersMake to cause a computer equipment (can be personal computer, server, or network equipment etc.) to perform each implementationSome Part Methods of example or embodiment.
The method, apparatus and equipment of identification surface deformation class disease provided in an embodiment of the present invention, by calculating three-dimensional roadThe elevation difference of face nominal contour and controlling profile, obtain cross section deformation elevation information；By will be spelled along the section of direction of trafficPick up and, form surface deformation depth image；Pass through the figure by surface deformation depth image classifying rationally into size non-overlapping copiesAs sub-block, statistical analysis is carried out to the depth value in each segmentation sub-block, obtains the average of depth value in sub-block, variance, positive and negativeDistributed intelligence；Realize quick, the accurate extraction of surface deformation depth information.
In surface deformation disease position fixing process, disease edge is extracted by using difference edge detection, with reference to deformation classDisease knowledge base and disease sub-block depth information, realize the first positioning to disease；There is region clustering using class disease is deformedProperty and change in depth continuity feature, based on the disease target sub-block after first positioning, carry out Morphological scale-space, utilize its topologyRelation carries out region growing reduction, obtains the high disease region of confidence level, and then realizes deformation class disease extracted region and ensureThe integrality in disease region.
Disease knowledge-base design grader is deformed by combining road, it is special using disease linear character, face battle array feature, positionSign, deformation extent classified the surface deformation disease region extracted, it is determined that deformation Damage Types, while perfect threeTie up surface deformation class disease knowledge base.
Finally it should be noted that：The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；AlthoughThe present invention is described in detail with reference to the foregoing embodiments, it will be understood by those within the art that：It still may be usedTo be modified to the technical scheme described in foregoing embodiments, or equivalent substitution is carried out to which part technical characteristic；And these modification or replace, do not make appropriate technical solution essence depart from various embodiments of the present invention technical scheme spirit andScope.