CN110930720B - Vehicle identification method and device - Google Patents
Vehicle identification method and device Download PDFInfo
- Publication number
- CN110930720B CN110930720B CN201911189552.0A CN201911189552A CN110930720B CN 110930720 B CN110930720 B CN 110930720B CN 201911189552 A CN201911189552 A CN 201911189552A CN 110930720 B CN110930720 B CN 110930720B
- Authority
- CN
- China
- Prior art keywords
- vehicle image
- vehicle
- point
- identified
- concave
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/04—Detecting movement of traffic to be counted or controlled using optical or ultrasonic detectors
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/065—Traffic control systems for road vehicles by counting the vehicles in a section of the road or in a parking area, i.e. comparing incoming count with outgoing count
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a vehicle identification method, which comprises the following steps: acquiring a vehicle image to be identified of a road section to be identified; extracting pits and bumps of the vehicle image to be recognized, and constructing a pit model and a bump model of the vehicle image to be recognized according to the pits and the bumps; and matching the vehicle image to be recognized with a vehicle image database based on the concave point model and the convex point model to obtain a vehicle recognition result of the vehicle image to be recognized and obtain the traffic flow condition of the road section to be recognized. The invention also discloses a vehicle identification device, which can effectively solve the problem of high difficulty in extracting the characteristics of the road traffic vehicles in the prior art, can stably extract the image characteristics of the vehicles and has higher identification degree.
Description
Technical Field
The invention relates to the technical field of vehicle identification, in particular to a vehicle identification method and device.
Background
With the increasing of the automobile holding capacity and the increasing of the road traffic pressure, the safety management problem related to the automobile is increasingly highlighted, and in order to realize the optimized management and scheduling of the running automobile, the number condition of the automobile can be obtained through the effective detection and identification of the automobile characteristics, so that the visual information reference is provided for a driver and a vehicle management scheduling center. The vehicle identification has important application value in the fields of vehicle safety management, road traffic control and the like, and the research on the vehicle feature extraction method has good application prospect in the aspect of detecting illegal crimes related to vehicles.
At present, a feature extraction system of a road traffic vehicle performs edge and information enhancement processing on an acquired vehicle image through an edge contour detection module and an enhancement processing module, and processes vehicle corner distribution information in an invariant region through a feature extraction module in a simulation mode so as to extract vehicle pixel feature points. However, when the distance from the camera is too far or the field information amount is large in the processing process, the difficulty in extracting the features of the road traffic vehicle is large.
Disclosure of Invention
The embodiment of the invention provides a vehicle identification method and device, which can effectively solve the problem of high difficulty in extracting the characteristics of road traffic vehicles in the prior art, can stably extract the image characteristics of the vehicles and has high identification degree.
An embodiment of the present invention provides a vehicle identification method, including:
acquiring a vehicle image to be identified of a road section to be identified;
extracting pits and bumps of the vehicle image to be recognized, and constructing a pit model and a bump model of the vehicle image to be recognized according to the pits and the bumps;
and matching the vehicle image to be recognized with a vehicle image database based on the concave point model and the convex point model to obtain a vehicle recognition result of the vehicle image to be recognized and obtain the traffic flow condition of the road section to be recognized.
As an improvement of the above solution, the image of the vehicle to be recognized is obtained by the following steps:
acquiring a vehicle image of a road section to be identified in real time;
and carrying out normalization processing on the vehicle image to obtain the vehicle image to be identified.
As an improvement of the above scheme, the normalizing the vehicle image to obtain the vehicle image to be recognized specifically includes:
carrying out gridding processing on the vehicle image to obtain a grid vehicle image;
calculating the mass center of each top point of the roof on the grid vehicle image grid, and taking a sphere with the mass center as the sphere center as a candidate area of the roof center;
calculating the shape index of any point in the candidate region, and determining the center point of the car roof according to the shape index;
and constructing a coordinate model of the grid vehicle image according to the roof center point to obtain the vehicle image to be identified.
As an improvement of the above scheme, the calculating a shape index of any point in the candidate region, and determining a roof center point according to the shape index specifically includes:
the shape index is obtained according to equation (1):
wherein, si (P) is a shape index of any point P in the candidate region, Kmax is a maximum main surface ratio of the point P, Kmin is a minimum main surface ratio of the point P;
and acquiring a corresponding point when the shape index reaches a preset roof center condition as the roof center point.
As an improvement of the above scheme, the constructing a coordinate model of the vehicle image according to the roof center point specifically includes:
and establishing the coordinate model by taking a central axis where each discrete point of the grid vehicle image is located as a coordinate y axis, taking a straight line which is perpendicular to the coordinate y axis and passes through the center point of the roof as a coordinate z axis, taking an intersection point of the coordinate y axis and the coordinate z axis as a coordinate origin, and taking a direction perpendicular to a yoz plane as a coordinate x axis.
As an improvement of the above scheme, the extracting the pits and the bumps of the to-be-recognized vehicle image specifically includes:
fitting each vertex of the vehicle image mesh to be identified into a local curved surface;
calculating the surface rate extreme value coefficient of each point on the local surface;
and screening points corresponding to the extreme coefficient of the surface rate when the extreme coefficient of the surface rate reaches a preset concave point condition as concave points, and screening points corresponding to the extreme coefficient of the surface rate when the extreme coefficient of the surface rate reaches a preset convex point condition as convex points.
As an improvement of the above scheme, the calculating a curvature ratio extremum coefficient of each point on the local curved surface specifically includes:
obtaining the surface ratio extreme value coefficient according to a formula (2):
wherein e is the extreme coefficient of the surface ratio, k is the main surface ratio of any point on the local surface, and t1、t2For the coordinate origin after the fitting of the local surface, a principal direction corresponding to the principal surface curvature, c0、c1、c2、c3And fitting coefficients in a polynomial to the local surface.
As an improvement of the above scheme, the pairing the to-be-recognized vehicle image with a vehicle image database based on the concave point model and the convex point model to obtain a vehicle recognition result of the to-be-recognized vehicle image and obtain a traffic flow condition of the to-be-recognized road segment includes:
carrying out three-dimensional uniform rasterization processing on the concave point model and the convex point model of the vehicle image to be recognized to obtain the spatial distribution density of concave points and the spatial distribution density of convex points of all grids;
constructing a three-dimensional distribution density array of the concave point model and a three-dimensional distribution density array of the convex point model according to the concave point spatial distribution density and the convex point spatial distribution density;
respectively obtaining the number of elements falling in each preset density interval in the three-dimensional distribution density array of each model to construct a concave point spatial distribution density histogram and a convex point spatial distribution density histogram;
matching the vehicle image to be recognized with the vehicle in the vehicle image database based on the concave point spatial distribution density histogram and the convex point spatial distribution density histogram to obtain a vehicle recognition result of the vehicle image to be recognized;
and acquiring the traffic flow condition of the road section to be identified according to the vehicle identification result of the vehicle image to be identified.
As an improvement of the above scheme, the matching the to-be-recognized vehicle image with the vehicle in the vehicle image database based on the concave point spatial distribution density histogram and the convex point spatial distribution density histogram to obtain the vehicle recognition result of the to-be-recognized vehicle image specifically includes:
calculating a first matching degree between the concave point space distribution density histogram of the vehicle image to be identified and the concave point space distribution density histogram of the vehicle in the vehicle image database;
calculating a second matching degree between the convex point space distribution density histogram of the vehicle image to be recognized and the convex point space distribution density histogram of the vehicle in the vehicle image database;
obtaining the matching degree of the vehicle image to be recognized and the vehicle in the vehicle image database according to the first matching degree and the second matching degree;
and screening out a corresponding vehicle when the matching degree reaches a preset matching condition from the vehicle image database as a vehicle recognition result of the vehicle image to be recognized.
Another embodiment of the present invention correspondingly provides a vehicle identification device, including:
the image acquisition module is used for acquiring an image of a vehicle to be identified on a road section to be identified;
the concave-convex point extraction module is used for extracting concave points and convex points of the vehicle image to be identified and constructing a concave point model and a convex point model of the vehicle image to be identified according to the concave points and the convex points;
and the matching identification module is used for matching the vehicle image to be identified with a vehicle image database based on the concave point model and the convex point model to obtain a vehicle identification result of the vehicle image to be identified and obtain the traffic flow condition of the road section to be identified.
Compared with the prior art, the vehicle identification method and the vehicle identification device disclosed by the embodiment of the invention have the advantages that the concave point model and the convex point model of the vehicle image to be identified are constructed according to the concave point and the convex point by acquiring the vehicle image to be identified of the road section to be identified, the vehicle image to be identified is matched with the vehicle image database based on the concave point model and the convex point model, the vehicle identification result of the vehicle image to be identified is obtained, and the traffic flow condition of the road section to be identified is obtained. The invention adopts a concave-convex characteristic extraction mode, obtains the vehicle quantity condition only by distinguishing whether the target is a motor vehicle or not, solves the problems of complex field condition, too far vehicle distance from a camera and more vehicle types, can stably extract the vehicle image characteristic, thereby effectively solving the problem of high difficulty in extracting the characteristic of the road traffic vehicle in the prior art, has higher identification degree, can quickly identify the vehicle characteristic and more accurately and efficiently obtains the traffic flow condition.
Drawings
Fig. 1 is a schematic flow chart of a vehicle identification method according to an embodiment of the present invention;
fig. 2 is a schematic diagram of an embodiment of a pair of images of a vehicle to be recognized in a vehicle recognition method according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a vehicle identification device according to a second embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example one
Fig. 1 is a schematic flow chart of a vehicle identification method according to an embodiment of the present invention.
The vehicle identification method provided by the embodiment can be executed by a vehicle identification terminal. In this embodiment, the vehicle identification terminal is preferably a vehicle identification device (even a cloud server, etc.), the vehicle identification device may be implemented in a software and/or hardware manner, and the vehicle identification terminal may be formed by two or more physical entities or may be formed by one physical entity.
Specifically, referring to fig. 1, the vehicle identification method includes steps S101 to S103.
S101, obtaining an image of the vehicle to be identified of the road section to be identified.
Preferably, the image of the vehicle to be recognized is obtained by the following steps:
acquiring a vehicle image of a road section to be identified in real time;
and carrying out normalization processing on the vehicle image to obtain the vehicle image to be identified.
Specifically, the vehicle image of the specific traffic section is obtained in real time, wherein the number of vehicles in the vehicle image is not particularly limited. Furthermore, different vehicle images are adjusted to be uniform preset standard front postures by carrying out normalization processing on the vehicle images, so that normalization of all vehicles is realized.
In the above embodiment, the normalization process of the vehicle image is preferably:
carrying out gridding processing on the vehicle image to obtain a grid vehicle image;
calculating the mass center of each top point of the roof on the grid vehicle image grid, and taking a sphere with the mass center as the sphere center as a candidate area of the roof center;
calculating the shape index of any point in the candidate region, and determining the center point of the car roof according to the shape index;
and constructing a coordinate model of the grid vehicle image according to the roof center point to obtain the vehicle image to be identified.
It should be noted that the center of the roof is the most obvious and most prominent part of the vehicle, so the center of the roof is the preferred reference point in the vehicle image normalization. After the gridding processing of the vehicle image, calculating the mass center of each vertex of the roof on the grid of the grid vehicle image, then taking a sphere taking the mass center as the sphere center as a candidate area of the roof center, and then determining the roof center point by calculating the shape index of each point in the candidate area. Furthermore, as each point of the vehicle is basically distributed on the surface of a cuboid and is symmetrical on the left side and the right side, in order to realize the normalization of all vehicles, the central axis of the cuboid where each discrete point of the vehicle image is located is taken as the Y axis under a new coordinate system, the straight line which is perpendicular to the Y axis and passes through the center point of the roof is taken as the Z axis of the new coordinate system, the intersection point of two coordinate axes is taken as the origin of the new coordinate system, and the direction which is perpendicular to the YOZ plane is taken as the X axis to establish a right-hand coordinate system. Through the establishment of the coordinate model, the acquired different vehicle images can be adjusted to be in a unified standard front posture. Of course, the size and the area of the vehicle roof of the vehicle image to be identified are basically consistent through the normalization of the size and the rectangular cutting taking the center of the vehicle roof as the center of the cuboid body.
In the above embodiment, preferably, the roof center point is determined by:
the shape index is obtained according to equation (1):
wherein, si (P) is a shape index of any point P in the candidate region, Kmax is a maximum main surface ratio of the point P, Kmin is a minimum main surface ratio of the point P;
and acquiring a corresponding point when the shape index reaches a preset roof center condition as the roof center point.
It should be noted that the closer any point of the roof is to the center of the roof, the closer the maximum and minimum principal curvatures of that point of the roof are, and therefore the point is the center of the roof when the maximum and minimum principal curvatures are equal. That is, the roof center conditions are: the shape index of any point on the roof is equal to a preset shape index threshold. Wherein, the shape index threshold is preferably set to,the shape index corresponding to the point when the maximum main surface ratio is equal to the minimum main surface ratio. More specifically, the roof center condition may be expressed by the equation:
s102, extracting the pits and the bumps of the vehicle image to be recognized, and constructing a pit model and a bump model of the vehicle image to be recognized according to the pits and the bumps.
Specifically, the concave points and the convex points are used as extreme points of the change of the curvature rate of the inner part of the local area on the curved surface along the main direction, and the important characteristics of the corresponding curved surface are fully represented. For an arbitrary three-dimensional object, the judgment of the pits and the bumps is relative, and the pits and the bumps can be mutually converted when the normal direction of the surface of the object is regulated to change. In order to extract the concave points and the convex points on the vehicle image to be recognized, local surface fitting is firstly carried out at each vertex of the vehicle image to be recognized. And further, judging whether each vertex is a concave point or a convex point according to the geometrical characteristic information on the local curved surface. Furthermore, for the vehicle image to be recognized after normalization processing, all concave points and convex points of the image are extracted by calculating the surface rate extreme value coefficient of each vertex, and then a concave point model and a convex point model of the vehicle image to be recognized are respectively constructed by all the concave points and the convex points.
In the foregoing embodiment, preferably, the extracting the pits and the bumps of the to-be-recognized vehicle image specifically includes:
fitting each vertex of the vehicle image mesh to be identified into a local curved surface;
calculating the surface rate extreme value coefficient of each point on the local surface;
and screening points corresponding to the extreme coefficient of the surface rate when the extreme coefficient of the surface rate reaches a preset concave point condition as concave points, and screening points corresponding to the extreme coefficient of the surface rate when the extreme coefficient of the surface rate reaches a preset convex point condition as convex points.
For example, a local curved surface z ═ f (x, y) is established at each vertex of the vehicle image mesh to be identified, and a fitting method of a neighboring normal cube can be adopted to fit a cubic polynomial shown by the following formula through each vertex and each vertex in the neighborhood thereof according to a least square method:
the vertex set used in the fitting process is taken from each vertex in the k (k is 1, 2, …) neighborhood of the corresponding vertex on the mesh, and the larger the selected k value, the smoother the fitted surface. It should be noted that some insignificant pits and bumps are lost during the fitting process.
In the above embodiment, preferably, the curvature extremum coefficient is obtained according to formula (2):
wherein e is the extreme coefficient of the surface ratio, k is the main surface ratio of any point on the local surface, and t1、t2For the coordinate origin after the fitting of the local surface, a principal direction corresponding to the principal surface curvature, c0、c1、c2、c3And fitting coefficients in a polynomial to the local surface.
It should be noted that the pits and the bumps, which are extreme points of the change of the curvature in the main direction in the local region on the curved surface, may be extracted by various methods. Illustratively, whether a certain point on the local curved surface is a concave point or a convex point is judged by calculating the coefficient of the curvature extremum. The definition of the curved surface rate extremum coefficient is as follows: for any point p, k on the curved surface z ═ f (x, y)maxAnd kminRespectively representing the maximum principal surface ratio and the minimum principal surface ratio, t, of the surface at the pointmaxAnd tminRespectively representing the two principal directions corresponding to the maximum and minimum principal curvatures obtained at that point, emaxAnd eminThe first derivatives of the maximum and minimum principal curvature along the two corresponding principal directions, respectively, are called curvature extremal coefficients, i.e.,
the differential geometry theory shows that the local shape of the local curved surface is completely determined by the size and the direction of the two main curvature, so that the embodiment is very effective in characterizing the local curved surface by means of curvature extreme coefficient. Thus, the curvature extremum coefficient can be obtained by equation (2).
Illustratively, after obtaining the curvature coefficient extremum coefficient value of any point of the local curved surface, the pit condition corresponds to: the minimum value of the curvature extremum coefficient value of the point p is zero, the first derivative of the curvature extremum coefficient value of the point p is greater than zero, and the minimum principal curvature of the point p is less than the absolute value of the maximum principal curvature of the point, so the point p is a concave point. More specifically, the pit condition may be expressed by the following equation: e.g. of the typemin＝0，
And k ismin<|kmax|。
Similarly, after obtaining the extreme value coefficient value of the curvature of any point of the local curved surface, the bump condition corresponds to: the maximum value of the curvature extreme value coefficient value of the point p is zero, the first derivative of the curvature extreme value coefficient value of the point p is less than zero, and the maximum principal curvature of the point p is greater than the absolute value of the minimum principal curvature of the point, so the point p is a convex point. More specifically, the bump condition may be expressed by the following equation: e.g. of the typemin＝0，
And k ismin<|kmax|。
S103, based on the concave point model and the convex point model, the vehicle image to be recognized is matched with a vehicle image database to obtain a vehicle recognition result of the vehicle image to be recognized, and the traffic flow condition of the road section to be recognized is obtained.
In the above embodiment, preferably, referring to fig. 2, it is a schematic diagram of an embodiment of a to-be-recognized vehicle image pairing in the vehicle recognition method provided in the first embodiment of the present invention, and the schematic diagram includes steps S1031 to S1035.
And S1031, carrying out three-dimensional uniform rasterization on the concave point model and the convex point model of the vehicle image to be recognized, and obtaining the spatial distribution density of the concave points and the spatial distribution density of the convex points of all grids.
It can be understood that the number and the position distribution information of concave and convex points can fully represent the remarkable characteristics of the concave and convex points due to the fact that the concave and convex points correspond to different vehicle images to be identified. After the concave points and the convex points of the vehicle images are extracted and normalized, all the vehicle images to be recognized are basically consistent in posture, area and size. In the matching process, a concave point model and a convex point model extracted from a vehicle image to be recognized and a vehicle image in a vehicle image database are subjected to three-dimensional uniform rasterization processing. Further, the number of pits falling inside each grid is counted in the pit model, and the number of bumps falling inside each grid is counted in the bump model, so that the spatial distribution density of pits and the spatial distribution density of bumps of all grids are obtained.
S1032, building a three-dimensional distribution density array of the concave point model and a three-dimensional distribution density array of the convex point model according to the concave point spatial distribution density and the convex point spatial distribution density.
And constructing a three-dimensional distribution density array of the concave point model and a three-dimensional distribution density array of the convex point model according to the spatial distribution density of the concave points and the spatial distribution density of the convex points. More specifically, the three-dimensional distribution density array may be expressed as
Can be represented by formula
And (4) calculating. Wherein M is1、M2、M3The number of the grids divided along the X-axis, the Y-axis and the Z-axis, nijkThe numbers of grid concave (convex) points with serial numbers of i, j and k along the direction of the X, Y, Z axis are respectively, and N is the total number of concave (convex) points in the concave (convex) point model.
And S1033, respectively obtaining the number of elements falling in each preset density interval in the three-dimensional distribution density array of each model to construct a concave point spatial distribution density histogram and a convex point spatial distribution density histogram.
In this embodiment, before step S1033, the method further includes dividing all possible distribution density values into L small regions, each small region corresponding to a stage of the distribution density, for example, the L-th distribution density value is a region [ f [ ]l,fl+1)(l∈[0,L-1]，f00). Thus, the number of elements falling in the density interval corresponding to each density level is counted in the three-dimensional distribution density array, and a concave point space distribution density histogram corresponding to the concave point model and a convex point space distribution density histogram corresponding to the convex point model are obtained. Illustratively, the histogram may be a 1-D discrete function of the form shown in the following equation:
(L ═ 0,1, …, L-1), where M is the total number of elements in the three-dimensional distribution array, i.e., M ═ M1×M2×M3，NlThe distribution density value in the F array falls in a class I interval [ Fl,fl+1) The number of elements in the table.
S1034, matching the vehicle image to be recognized with the vehicle in the vehicle image database based on the concave point space distribution density histogram and the convex point space distribution density histogram to obtain a vehicle recognition result of the vehicle image to be recognized.
In a preferred embodiment, step S1034 is specifically:
calculating a first matching degree between the concave point space distribution density histogram of the vehicle image to be identified and the concave point space distribution density histogram of the vehicle in the vehicle image database;
calculating a second matching degree between the convex point space distribution density histogram of the vehicle image to be recognized and the convex point space distribution density histogram of the vehicle in the vehicle image database;
obtaining the matching degree of the vehicle image to be recognized and the vehicle in the vehicle image database according to the first matching degree and the second matching degree;
and screening out a corresponding vehicle when the matching degree reaches a preset matching condition from the vehicle image database as a vehicle recognition result of the vehicle image to be recognized.
In this embodiment, after obtaining the spatial distribution density histogram of the concave points corresponding to the concave point model, the histogram intersection method may be adopted to obtain the spatial distribution density histogram H of the concave points of the vehicle image to be identifiedp(l) And a vehicle concave model space distribution density histogram H in a vehicle image databaseG(l) A first degree of matching M (P, G). Similarly, obtaining a convex point space distribution density histogram H of the vehicle image to be identifiedp(l) And the spatial distribution density histogram H of the vehicle convex model in the vehicle image databaseG(l) A second degree of matching M (P, G). More specifically, the first degree of matching or the second degree of matching may be represented by the formula:
further, according to the first matching degree and the second matching degree, the matching degree of the vehicle image to be recognized and the vehicle in the vehicle image database is obtained. Furthermore, the matching process adopts a k-nearest neighbor method to select N with the maximum matching degree with the spatial distribution density histogram of the vehicle image to be identified in the vehicle image databasegAnd (5) the vehicle model enters the next step of identification. And then, screening a vehicle image database rapidly in the identification process, and screening a vehicle meeting the matching condition as a vehicle identification result of the vehicle image to be identified.
Wherein the matching condition is as follows: the matching degree of the vehicle image to be recognized and the vehicles in the vehicle image database is greater than a preset matching threshold value. Illustratively, the matching condition is that the matching degree is greater than a matching threshold of 90%.
And S1035, acquiring the traffic flow condition of the road section to be identified according to the vehicle identification result of the vehicle image to be identified.
It can be understood that the present embodiment obtains the number of vehicles only by distinguishing whether the target is a motor vehicle, so as to obtain the traffic flow condition of the road section to be identified.
According to the vehicle identification method provided by the embodiment of the invention, the vehicle image to be identified of the road section to be identified is obtained, the concave point and the convex point of the vehicle image to be identified are extracted, the concave point model and the convex point model of the vehicle image to be identified are constructed according to the concave point and the convex point, the vehicle image to be identified is matched with the vehicle image database based on the concave point model and the convex point model, the vehicle identification result of the vehicle image to be identified is obtained, and the traffic flow condition of the road section to be identified is obtained. The invention adopts a concave-convex characteristic extraction mode, obtains the vehicle quantity condition only by distinguishing whether the target is a motor vehicle or not, solves the problems of complex field condition, too far vehicle distance from a camera and more vehicle types, can stably extract the vehicle image characteristic, thereby effectively solving the problem of high difficulty in extracting the characteristic of the road traffic vehicle in the prior art, has higher identification degree, can quickly identify the vehicle characteristic and more accurately and efficiently obtains the traffic flow condition.
Example two
Referring to fig. 3, a schematic structural diagram of a vehicle identification device according to a second embodiment of the present invention is shown, including:
the image acquisition module 201 is used for acquiring a vehicle image to be identified of a road section to be identified;
the concave-convex point extraction module 202 is used for extracting concave points and convex points of the vehicle image to be identified, and constructing a concave point model and a convex point model of the vehicle image to be identified according to the concave points and the convex points;
and the matching identification module 203 is configured to pair the vehicle image to be identified with a vehicle image database based on the concave point model and the convex point model to obtain a vehicle identification result of the vehicle image to be identified, and obtain a traffic flow condition of the road section to be identified.
Preferably, the image acquiring module 201 includes:
the vehicle image acquisition unit is used for acquiring a vehicle image of a road section to be identified in real time;
and the normalization processing unit is used for performing normalization processing on the vehicle image to obtain the vehicle image to be identified.
Preferably, the normalization processing unit includes:
the gridding processing unit is used for carrying out gridding processing on the vehicle image to obtain a gridding vehicle image;
the center of mass calculating unit is used for calculating the center of mass of each top point of the roof on the grid vehicle image grid, and a sphere taking the center of mass as the sphere center is taken as a candidate area of the roof center;
the vehicle roof center point determining unit is used for calculating the shape index of any point in the candidate area and determining the vehicle roof center point according to the shape index;
and the to-be-identified vehicle image determining unit is used for constructing a coordinate model of the grid vehicle image according to the roof center point to obtain the to-be-identified vehicle image.
Preferably, the roof center point determining unit includes:
a shape index calculation unit for obtaining the shape index according to formula (1):
wherein, si (P) is a shape index of any point P in the candidate region, Kmax is a maximum main surface ratio of the point P, Kmin is a minimum main surface ratio of the point P;
and the shape index screening unit is used for acquiring a corresponding point when the shape index reaches a preset roof center condition as the roof center point.
Preferably, the vehicle image to be recognized determining unit includes:
and the coordinate model building unit is used for building the coordinate model by taking the central axis of each discrete point of the grid vehicle image as a coordinate y axis, taking a straight line which is perpendicular to the coordinate y axis and passes through the center point of the roof as a coordinate z axis, taking the intersection point of the coordinate y axis and the coordinate z axis as a coordinate origin, and taking the direction perpendicular to the yoz plane as a coordinate x axis.
Preferably, the concave-convex point extracting module 202 comprises:
the local curved surface fitting unit is used for fitting each vertex of the vehicle image mesh to be identified into a local curved surface;
the curved surface rate extreme value coefficient calculating unit is used for calculating the curved surface rate extreme value coefficient of each point on the local curved surface;
and the concave-convex point identification unit is used for screening corresponding points as concave points when the curved surface rate extreme value coefficient reaches a preset concave point condition, and screening corresponding points as convex points when the curved surface rate extreme value coefficient reaches a preset convex point condition.
Preferably, the curvature extremum coefficient calculating unit includes:
the specific calculation unit of the surface rate extreme coefficient is used for obtaining the surface rate extreme coefficient according to a formula (2):
wherein e is the extreme coefficient of the surface ratio, k is the main surface ratio of any point on the local surface, and t1、t2For the coordinate origin after the fitting of the local surface, a principal direction corresponding to the principal surface curvature, c0、c1、c2、c3And fitting coefficients in a polynomial to the local surface.
Preferably, the matching identification module 203 includes:
the concave-convex point spatial distribution density acquisition unit is used for carrying out three-dimensional uniform rasterization processing on the concave point model and the convex point model of the vehicle image to be identified so as to acquire the concave point spatial distribution density and the convex point spatial distribution density of all grids;
the concave-convex point three-dimensional distribution density array construction unit is used for constructing a three-dimensional distribution density array of the concave point model and a three-dimensional distribution density array of the convex point model according to the concave point spatial distribution density and the convex point spatial distribution density;
the concave-convex point spatial distribution density histogram construction unit is used for respectively obtaining the number of elements falling in each preset density interval in the three-dimensional distribution density array of each model so as to construct a concave point spatial distribution density histogram and a convex point spatial distribution density histogram;
the matching unit is used for matching the vehicle image to be recognized with the vehicle in the vehicle image database based on the concave point spatial distribution density histogram and the convex point spatial distribution density histogram to obtain a vehicle recognition result of the vehicle image to be recognized;
and the traffic flow condition acquisition unit is used for acquiring the traffic flow condition of the road section to be identified according to the vehicle identification result of the vehicle image to be identified.
Preferably, the matching unit includes:
the first matching degree calculation unit is used for calculating a first matching degree between the concave point space distribution density histogram of the vehicle image to be identified and the concave point space distribution density histogram of the vehicle in the vehicle image database;
the second matching degree calculation unit is used for calculating a second matching degree between the bump space distribution density histogram of the vehicle image to be recognized and the bump space distribution density histogram of the vehicle in the vehicle image database;
the matching degree obtaining unit is used for obtaining the matching degree of the vehicle image to be recognized and the vehicle in the vehicle image database according to the first matching degree and the second matching degree;
and the vehicle screening unit is used for screening out a vehicle corresponding to the matching degree reaching a preset matching condition from the vehicle image database as a vehicle identification result of the vehicle image to be identified.
According to the vehicle identification device provided by the embodiment of the invention, the vehicle image to be identified of the road section to be identified is obtained, the concave point and the convex point of the vehicle image to be identified are extracted, the concave point model and the convex point model of the vehicle image to be identified are constructed according to the concave point and the convex point, the vehicle image to be identified is matched with the vehicle image database based on the concave point model and the convex point model, the vehicle identification result of the vehicle image to be identified is obtained, and the traffic flow condition of the road section to be identified is obtained. The invention adopts a concave-convex characteristic extraction mode, obtains the vehicle quantity condition only by distinguishing whether the target is a motor vehicle or not, solves the problems of complex field condition, too far vehicle distance from a camera and more vehicle types, can stably extract the vehicle image characteristic, thereby effectively solving the problem of high difficulty in extracting the characteristic of the road traffic vehicle in the prior art, has higher identification degree, can quickly identify the vehicle characteristic and more accurately and efficiently obtains the traffic flow condition.
It should be noted that the above-described device embodiments are merely illustrative, where the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. In addition, in the drawings of the embodiment of the apparatus provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, and may be specifically implemented as one or more communication buses or signal lines. One of ordinary skill in the art can understand and implement it without inventive effort.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention.
Claims (8)
1. A vehicle identification method, characterized by comprising:
acquiring a vehicle image to be identified of a road section to be identified; the method comprises the steps of acquiring a vehicle image of a road section to be identified in real time; carrying out normalization processing on the vehicle image to obtain the vehicle image to be identified; carrying out gridding processing on the vehicle image to obtain a grid vehicle image; calculating the mass center of each top point of the roof on the grid vehicle image grid, and taking a sphere with the mass center as the sphere center as a candidate area of the roof center; calculating the shape index of any point in the candidate region, and determining the center point of the car roof according to the shape index; according to the roof center point, a coordinate model of the grid vehicle image is built, and the vehicle image to be identified is obtained;
extracting pits and bumps of the vehicle image to be recognized, and constructing a pit model and a bump model of the vehicle image to be recognized according to the pits and the bumps;
and matching the vehicle image to be recognized with a vehicle image database based on the concave point model and the convex point model to obtain a vehicle recognition result of the vehicle image to be recognized and obtain the traffic flow condition of the road section to be recognized.
2. The vehicle identification method according to claim 1, wherein the calculating of the shape index of any point in the candidate region and the determining of the roof center point according to the shape index specifically comprises:
the shape index is obtained according to equation (1):
wherein, si (P) is a shape index of any point P in the candidate region, Kmax is a maximum main surface ratio of the point P, Kmin is a minimum main surface ratio of the point P;
and acquiring a corresponding point when the shape index reaches a preset roof center condition as the roof center point.
3. The vehicle identification method according to claim 1, wherein the constructing a coordinate model of the vehicle image according to the roof center point specifically comprises:
and establishing the coordinate model by taking a central axis where each discrete point of the grid vehicle image is located as a coordinate y axis, taking a straight line which is perpendicular to the coordinate y axis and passes through the center point of the roof as a coordinate z axis, taking an intersection point of the coordinate y axis and the coordinate z axis as a coordinate origin, and taking a direction perpendicular to a yoz plane as a coordinate x axis.
4. The vehicle identification method according to claim 1, wherein the extracting of the pits and the bumps of the image of the vehicle to be identified specifically comprises:
fitting each vertex of the vehicle image mesh to be identified into a local curved surface;
calculating the surface rate extreme value coefficient of each point on the local surface;
and screening points corresponding to the extreme coefficient of the surface rate when the extreme coefficient of the surface rate reaches a preset concave point condition as concave points, and screening points corresponding to the extreme coefficient of the surface rate when the extreme coefficient of the surface rate reaches a preset convex point condition as convex points.
5. The method according to claim 4, wherein the calculating the extremum coefficient of curvature for each point on the local surface specifically includes:
obtaining the surface ratio extreme value coefficient according to a formula (2):
wherein e is the extreme coefficient of the surface ratio, k is the main surface ratio of any point on the local surface, and t1、t2For the coordinate origin after the fitting of the local surface, a principal direction corresponding to the principal surface curvature, c0、c1、c2、c3And fitting coefficients in a polynomial to the local surface.
6. The vehicle identification method according to claim 1, wherein the pairing the to-be-identified vehicle image with a vehicle image database based on the concave point model and the convex point model to obtain a vehicle identification result of the to-be-identified vehicle image and obtain a traffic flow condition of the to-be-identified road section comprises:
carrying out three-dimensional uniform rasterization processing on the concave point model and the convex point model of the vehicle image to be recognized to obtain the spatial distribution density of concave points and the spatial distribution density of convex points of all grids;
constructing a three-dimensional distribution density array of the concave point model and a three-dimensional distribution density array of the convex point model according to the concave point spatial distribution density and the convex point spatial distribution density;
respectively obtaining the number of elements falling in each preset density interval in the three-dimensional distribution density array of each model to construct a concave point spatial distribution density histogram and a convex point spatial distribution density histogram;
matching the vehicle image to be recognized with the vehicle in the vehicle image database based on the concave point spatial distribution density histogram and the convex point spatial distribution density histogram to obtain a vehicle recognition result of the vehicle image to be recognized;
and acquiring the traffic flow condition of the road section to be identified according to the vehicle identification result of the vehicle image to be identified.
7. The vehicle identification method according to claim 6, wherein the matching the vehicle image to be identified with the vehicle in the vehicle image database based on the histogram of spatial distribution density of concave points and the histogram of spatial distribution density of convex points to obtain the vehicle identification result of the vehicle image to be identified specifically comprises:
calculating a first matching degree between the concave point space distribution density histogram of the vehicle image to be identified and the concave point space distribution density histogram of the vehicle in the vehicle image database;
calculating a second matching degree between the convex point space distribution density histogram of the vehicle image to be recognized and the convex point space distribution density histogram of the vehicle in the vehicle image database;
obtaining the matching degree of the vehicle image to be recognized and the vehicle in the vehicle image database according to the first matching degree and the second matching degree;
and screening out a corresponding vehicle when the matching degree reaches a preset matching condition from the vehicle image database as a vehicle recognition result of the vehicle image to be recognized.
8. A vehicle identification device characterized by comprising:
the image acquisition module is used for acquiring an image of a vehicle to be identified on a road section to be identified; the method comprises the steps of acquiring a vehicle image of a road section to be identified in real time; carrying out normalization processing on the vehicle image to obtain the vehicle image to be identified; carrying out gridding processing on the vehicle image to obtain a grid vehicle image; calculating the mass center of each top point of the roof on the grid vehicle image grid, and taking a sphere with the mass center as the sphere center as a candidate area of the roof center; calculating the shape index of any point in the candidate region, and determining the center point of the car roof according to the shape index; according to the roof center point, a coordinate model of the grid vehicle image is built, and the vehicle image to be identified is obtained;
the concave-convex point extraction module is used for extracting concave points and convex points of the vehicle image to be identified and constructing a concave point model and a convex point model of the vehicle image to be identified according to the concave points and the convex points;
and the matching identification module is used for matching the vehicle image to be identified with a vehicle image database based on the concave point model and the convex point model to obtain a vehicle identification result of the vehicle image to be identified and obtain the traffic flow condition of the road section to be identified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911189552.0A CN110930720B (en) | 2019-11-28 | 2019-11-28 | Vehicle identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911189552.0A CN110930720B (en) | 2019-11-28 | 2019-11-28 | Vehicle identification method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110930720A CN110930720A (en) | 2020-03-27 |
| CN110930720B true CN110930720B (en) | 2020-11-13 |
Family
ID=69846746
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911189552.0A Active CN110930720B (en) | 2019-11-28 | 2019-11-28 | Vehicle identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110930720B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102799857A (en) * | 2012-06-19 | 2012-11-28 | 东南大学 | Video multi-vehicle outline detection method |
| CN106781513A (en) * | 2016-11-28 | 2017-05-31 | 东南大学 | The recognition methods of vehicle behavior in a kind of urban transportation scene of feature based fusion |
| CN108171203A (en) * | 2018-01-17 | 2018-06-15 | 百度在线网络技术（北京）有限公司 | For identifying the method and apparatus of vehicle |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102043957B (en) * | 2011-01-11 | 2016-07-20 | 北京邮电大学 | Method of Vehicle Segmentation based on concave spots of image |
| WO2012128656A1 (en) * | 2011-03-22 | 2012-09-27 | Gan Mikhail Abramovich | Device for displaying visual information on the windshield of a vehicle |
| CN106257446B (en) * | 2015-06-19 | 2019-05-31 | 苏州市大创信息运用有限公司 | Geography information Freehandhand-drawing system based on longitude and latitude |
| KR101914041B1 (en) * | 2016-08-03 | 2018-11-02 | 주식회사 코렌 | Optical lens assembly and electronic apparatus having the same |
| CN109102495A (en) * | 2018-07-05 | 2018-12-28 | 广州杰赛科技股份有限公司 | Target image determines method and system, computer equipment, computer storage medium |
| CN109784216B (en) * | 2018-12-28 | 2023-06-20 | 华南理工大学 | Vehicle-mounted thermal imaging pedestrian detection Rois extraction method based on probability map |
-
2019
- 2019-11-28 CN CN201911189552.0A patent/CN110930720B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102799857A (en) * | 2012-06-19 | 2012-11-28 | 东南大学 | Video multi-vehicle outline detection method |
| CN106781513A (en) * | 2016-11-28 | 2017-05-31 | 东南大学 | The recognition methods of vehicle behavior in a kind of urban transportation scene of feature based fusion |
| CN108171203A (en) * | 2018-01-17 | 2018-06-15 | 百度在线网络技术（北京）有限公司 | For identifying the method and apparatus of vehicle |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110930720A (en) | 2020-03-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107292276B (en) | Vehicle-mounted point cloud clustering method and system | |
| CN108681692B (en) | Method for identifying newly added buildings in remote sensing image based on deep learning | |
| CN110348297B (en) | Detection method, system, terminal and storage medium for identifying stereo garage | |
| CN113095205B (en) | Point cloud target detection method based on improved Hough voting | |
| CN113205063A (en) | Visual identification and positioning method for defects of power transmission conductor | |
| CN108550166B (en) | Spatial target image matching method | |
| CN111401188B (en) | Traffic police gesture recognition method based on human body key point characteristics | |
| CN113470090A (en) | Multi-solid-state laser radar external reference calibration method based on SIFT-SHOT characteristics | |
| CN106204557A (en) | A kind of extracting method of the non-complete data symmetrical feature estimated with M based on extension Gaussian sphere | |
| CN114612665B (en) | Pose estimation and dynamic vehicle detection method based on normal vector histogram features | |
| CN106156752A (en) | A kind of model recognizing method based on inverse projection three-view diagram | |
| CN115563732B (en) | Spraying track simulation optimization method and device based on virtual reality | |
| CN111145129A (en) | Point cloud denoising method based on hyper-voxels | |
| CN112084453A (en) | Three-dimensional virtual display system, method, computer equipment, terminal and storage medium | |
| CN111783721A (en) | Lane line extraction method of laser point cloud and electronic equipment | |
| CN117132630A (en) | Point cloud registration method based on second-order spatial compatibility measurement | |
| CN112241964A (en) | Light strip center extraction method for line structured light non-contact measurement | |
| CN117830297B (en) | Weld joint identification method, welding device and electronic equipment | |
| CN111310720A (en) | Pedestrian re-identification method and system based on graph metric learning | |
| CN111985507A (en) | Rock three-dimensional point cloud joint trace extraction method | |
| CN117011299B (en) | Reference point cloud quality assessment method and system integrating graph resampling and gradient characteristics | |
| CN110930720B (en) | Vehicle identification method and device | |
| CN112115737B (en) | Vehicle orientation determining method and device and vehicle-mounted terminal | |
| WO2023131203A1 (en) | Semantic map updating method, path planning method, and related apparatuses | |
| CN111310772A (en) | Point line feature selection method and system for binocular vision SLAM |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |