CN114693863A - Vehicle weight recognition method and device based on laser radar camera - Google Patents
Vehicle weight recognition method and device based on laser radar camera Download PDFInfo
- Publication number
- CN114693863A CN114693863A CN202011598701.1A CN202011598701A CN114693863A CN 114693863 A CN114693863 A CN 114693863A CN 202011598701 A CN202011598701 A CN 202011598701A CN 114693863 A CN114693863 A CN 114693863A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- similarity
- vehicle
- cloud data
- dimensional
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 85
- 239000013598 vector Substances 0.000 claims description 77
- 238000001514 detection method Methods 0.000 claims description 58
- 238000004590 computer program Methods 0.000 claims description 53
- 238000004422 calculation algorithm Methods 0.000 claims description 30
- 238000000605 extraction Methods 0.000 claims description 20
- 238000004364 calculation method Methods 0.000 claims description 16
- 230000003068 static effect Effects 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 14
- 238000012163 sequencing technique Methods 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 4
- 238000013499 data model Methods 0.000 description 70
- 230000008859 change Effects 0.000 description 32
- 238000010586 diagram Methods 0.000 description 11
- 230000008569 process Effects 0.000 description 11
- 238000009616 inductively coupled plasma Methods 0.000 description 6
- 230000009466 transformation Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 206010034719 Personality change Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
- G06T7/344—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods involving models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The application relates to a vehicle re-identification method, a vehicle re-identification device, computer equipment and a storage medium.
Description
Technical Field
The application relates to the technical field of intelligent transportation, in particular to a three-dimensional point cloud data model reconstruction method, a target re-identification method, a device, computer equipment and a storage medium.
Background
With the development of intelligent transportation technology, the requirement on the reliability of traffic information is higher and higher, and how to obtain road information as rich as possible is one of the keys for improving the reliability. The laser radar is one of main sensors of a roadside sensing system, and a vehicle can be positioned and tracked through a sensing algorithm.
However, the cost of the laser radar is limited, the radar with a too high beam cannot be used, the installation position of the roadside laser radar is relatively high, the roadside laser radar is relatively far away from the vehicle, generally, the number of laser lines scanned on the vehicle is relatively small, and available information is relatively limited. The single-frame vehicle point cloud data is also local data, and geometric information such as the length, the width and the like of the vehicle cannot be accurately acquired by using the single-frame point cloud data, so that the vehicle cannot be described in detail.
Disclosure of Invention
In view of the above, it is necessary to provide a three-dimensional point cloud data model reconstruction method, an object re-identification method, an apparatus, a computer device, and a storage medium that can solve the above-mentioned problems.
A three-dimensional point cloud data model reconstruction method comprises the following steps:
carrying out target detection on a plurality of continuous frame point cloud data to obtain a target detection result of each frame of point cloud data; the target detection result of each frame of point cloud data comprises the same target object;
calculating an axis vector according to static point clouds in a plurality of continuous frame point cloud data, wherein the axis vector has a fixed included angle with the ground;
and registering the target detection result of each frame of point cloud data based on the axis vector, and outputting a three-dimensional point cloud data model containing the same target object according to the registration result.
A vehicle weight recognition method, comprising:
acquiring point cloud data and a plane image of a target vehicle, wherein the point cloud data comprises a plurality of continuous frames of the target vehicle;
acquiring appearance information and two-dimensional characteristics of the target vehicle based on the plane image;
acquiring a contour curve of a target vehicle based on the point cloud data;
selecting a plurality of candidate re-identification information from the re-identification information of the query library by using the appearance information; each candidate re-identification information corresponds to one candidate vehicle, a plurality of re-identification information are stored in the query library, and each re-identification information comprises a vehicle identity, pre-stored appearance information, pre-stored two-dimensional characteristics and a pre-stored contour curve;
calculating the similarity between the contour curve and each pre-stored contour curve to obtain a first similarity;
calculating the similarity between the two-dimensional features and each pre-stored two-dimensional feature to obtain a second similarity;
and outputting a re-recognition result based on the first similarity and the second similarity of each candidate vehicle.
A method of object re-identification, comprising:
acquiring point cloud data of a target, wherein the point cloud data comprises a plurality of continuous frames of the target;
processing the plurality of continuous frames by using a three-dimensional point cloud data model reconstruction method to construct a three-dimensional point cloud data model of the target;
calculating a contour curve of the target using the three-dimensional model;
obtaining a re-recognition result matched from the query library by utilizing the similarity between the profile curve and a pre-stored profile curve; the pre-stored profile curve is a pre-stored profile curve in a plurality of pieces of re-identification information pre-stored in the query library.
An apparatus for reconstructing a three-dimensional point cloud data model, the apparatus comprising:
the acquisition module is used for carrying out target detection on a plurality of continuous frame point cloud data to obtain a target detection result of each frame of point cloud data; the target detection result of each frame of point cloud data comprises the same target object;
the system comprises an axis vector calculation module, a data acquisition module and a data processing module, wherein the axis vector calculation module is used for calculating an axis vector according to a static target point in a plurality of continuous frame point cloud data, and the axis vector and the ground have a fixed included angle;
and the registration module is used for registering the target detection result of each frame of point cloud data based on the axis vector and outputting a three-dimensional point cloud data model containing the same target object according to the registration result.
An object re-identification apparatus, the apparatus comprising:
the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring point cloud data of a target, and the point cloud data comprises a plurality of continuous frames of the target;
the matching module is used for constructing a three-dimensional point cloud data model of the target by utilizing a three-dimensional point cloud data model reconstruction method; and calculating a contour curve of the target using the three-dimensional model;
the similarity calculation module is used for obtaining a re-identification result matched from the query library by utilizing the similarity between the profile curve and a pre-stored profile curve; the pre-stored profile curve is a pre-stored profile curve in a plurality of pieces of re-identification information pre-stored in the query library.
A computer device comprising a memory storing a computer program and a processor implementing the steps of the method in embodiments of the application when executing the computer program.
A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method in the embodiments of the application.
According to the three-dimensional point cloud data model reconstruction method, the target re-identification device, the computer equipment and the storage medium, the acquired new frame of point cloud registration process is limited by the acquired axis vector, and the method limits the three-dimensional registration of the three-dimensional point cloud to two-dimensional during execution, so that the difficulty of registration operation and the data volume of the execution process are greatly reduced, and therefore the three-dimensional point cloud data model reconstruction method of the embodiment has low requirements on computing resources of the equipment and is high in computing speed. The method for reconstructing the three-dimensional point cloud data model has the advantages that the obtained continuous multi-frame point cloud is processed by the method for reconstructing the three-dimensional point cloud data model, the target contour line with richer content can be obtained compared with a single-frame point cloud, and when the contour obtained by the point cloud is used for re-recognition, the method is accurate and reliable.
Drawings
FIG. 1 is a diagram of an application environment in one embodiment;
FIG. 2 is a schematic flow chart of a three-dimensional point cloud data model reconstruction method according to an embodiment;
FIG. 3 is a schematic flow chart illustrating the refinement step of step S230 in one embodiment;
FIG. 4 is a schematic flow chart diagram illustrating a method for re-identifying an object in one embodiment;
FIG. 5 is a schematic flow chart illustrating vehicle re-identification according to one embodiment;
FIG. 6 is a schematic flow chart illustrating the step of refining step S420 in one embodiment;
FIG. 7 is a diagram of a smallest outer enclosure box in one embodiment;
FIG. 8 is a graph of profile curve effects in one embodiment;
FIG. 9 is a block diagram of an embodiment of a three-dimensional point cloud data model reconstruction apparatus;
FIG. 10 is a block diagram of an object re-identification apparatus in one embodiment;
FIG. 11 is a block diagram of an object re-identification apparatus in one embodiment;
FIG. 12 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of and not restrictive on the broad application.
The three-dimensional point cloud data model reconstruction method provided by the application can be applied to the application environment shown in fig. 1. Wherein lidar 102 is in communication with processor 104 via a network. Optionally, the laser radar may be a roadside laser radar or a vehicle-end laser radar. Optionally, on the premise that the processor has sufficient computing power, the processor 104 may be a cloud server, a vehicle-end processor, a roadside computing unit, a roadside edge server, or a random combination of the above processor forms. The processor form is not particularly limited in the present application.
In an embodiment, as shown in fig. 2, a three-dimensional point cloud data model reconstruction method is provided, where a processor in fig. 1 is an edge server on the road side, and a lidar is a lidar on the road side, and the method is applied to the edge server 104 in fig. 1 to perform the method as an example, and includes the following steps:
step 210, performing target detection on a plurality of continuous frame point cloud data to obtain a target detection result of each frame of point cloud data.
And the target detection result of each frame of point cloud data comprises the same target object.
The point cloud data in this step is obtained by scanning the roadside lidar 102 in a preset area. After the point cloud data is obtained, the edge server 104 operates a preset point cloud target detection algorithm to perform target identification on the point cloud data, so as to obtain a target detection result of each frame of point cloud data.
Optionally, after the point cloud data is obtained, the edge server 104 may determine continuous frame cloud data included in the point cloud data to check whether there is abnormal data. As a mode, the point cloud frames with too few points or suddenly and sharply reduced points can be removed, so that the validity of data in the point cloud data set is ensured.
Step 220, calculating an axis vector according to the static target points in the plurality of continuous frame point cloud data.
Wherein, the axis vector and the ground have a fixed included angle.
The edge server 104 may also determine the type of the point cloud data when performing target identification on the point cloud data, and the type of the point cloud data is divided into static point cloud and dynamic point cloud. The static point cloud corresponds to a static object in the point cloud data, which may be a stationary road, building, etc., and the dynamic point cloud corresponds to a dynamic object, which may be a moving vehicle, pedestrian, non-motorized vehicle, etc. Optionally, the static point cloud is a ground point, and the axis vector is perpendicular or parallel to the ground. Alternatively, for more refined data processing, an axis vector may be calculated for each frame of point cloud data, and multiple axis vectors may be calculated for multiple frames of point clouds, and their average value may be taken to obtain an axis vector. Optionally, the mean value may be further normalized to obtain an axis vector.
And 230, registering the target detection result of each frame of point cloud data based on the axis vector, and outputting a three-dimensional point cloud data model containing the same target object according to the registration result.
Specifically, when implementing step S230, the edge server 104 may cyclically execute the steps of calculating a pose change between current frame point cloud data and previous frame point cloud data by using the axis vector as a constraint, registering a detection result of the current frame point cloud data to the global three-dimensional point cloud data model by using the pose change until a target detection result of the last frame point cloud data is registered to the global three-dimensional point cloud data model, and outputting a three-dimensional reconstruction model including the same target object. It should be noted that the global three-dimensional point cloud data model is a three-dimensional point cloud data model obtained before the current frame point cloud data is registered, if the axis vector is perpendicular to the ground, the change value of the pose change value in the direction of the axis vector is 0, and if the axis vector is parallel to the ground, the change value of the pose change value in the direction perpendicular to the axis vector is 0.
In the method for reconstructing the three-dimensional point cloud data model, the acquired axis vector is used for limiting the acquired new frame point cloud registration process, and the method limits the three-dimensional registration of the three-dimensional point cloud to be two-dimensional during execution, so that the difficulty of registration operation and the data volume of the execution process are greatly reduced, and therefore, the method for reconstructing the three-dimensional point cloud data model has low requirement on the computing resources of equipment and high computing speed. According to the method, the point cloud data acquired by the road side laser radar or the point cloud data acquired by the vehicle-mounted laser radar on the flat road are applied to three-dimensional reconstruction, and the effect is remarkable.
In one embodiment, as shown in fig. 3, the step of registering the detection result of the current frame point cloud data to the global three-dimensional point cloud data model by using the pose change value may be implemented by:
step 311, the pose change between the current frame point cloud data and the previous frame point cloud data is used as the initial pose.
Optionally, after a new frame of point cloud data is obtained, the attitude change between the current frame of point cloud and the previous frame of point cloud is calculated by using an icp algorithm and is used as the initial pose of the current frame registered to the global three-dimensional point cloud model. In the step of executing, the previously calculated axis vector can be added into the process of optimizing the pose by the icp algorithm, so that the pose transformation is only performed around the axis vector, and no rotation component and translation component exist in the direction of the axis vector, which can reduce the operation of the step and reduce the requirement of the coarse registration process on computing resources.
And step 312, taking the axis vector as a limit, and adjusting the initial pose by using a fine registration algorithm to register the target detection result of the current frame point cloud data to the global three-dimensional point cloud data model.
Optionally, the initial pose obtained by calculation is used as input, and the current frame point cloud is registered to the global three-dimensional point cloud model by using the nick algorithm and taking the axis vector as a limiting condition again.
The icp algorithm and the nic algorithm are common algorithms in the prior art, and details of implementation of the algorithms are not described in detail herein. In this embodiment, in order to fully utilize the obtained axis vector to realize the process of point cloud registration to the global three-dimensional point cloud model, other point cloud registration algorithms in the prior art may also be applied adaptively, and this application is not listed one by one.
Optionally, after each execution of step 312, in order to bring the latest global three-dimensional point cloud model closer to the next frame of point cloud data (which facilitates registration), the method may further include: and performing inverse transformation on the global three-dimensional point cloud data model by using the registration parameters output in the current step, so that the global three-dimensional point cloud data model is moved to the pose of the current frame point cloud data.
The embodiment applies the axis vectors to a mature algorithm of coarse registration and fine registration very appropriately, which reduces the difficulty of implementation of the scheme.
Based on the same inventive concept, as shown in fig. 4, in one embodiment, a target re-identification method is further provided, which includes:
step 310, point cloud data of a target is obtained, wherein the point cloud data comprises a plurality of continuous frames of the target.
And 320, constructing a three-dimensional point cloud data model of the target by using the three-dimensional point cloud data model reconstruction method in the embodiment.
Step 330, calculating a contour curve of the target by using the three-dimensional model.
Step 340, obtaining a re-identification result matched from the query library by utilizing the similarity between the profile curve and a pre-stored profile curve; the pre-stored profile curve is a pre-stored profile curve in a plurality of pieces of re-identification information pre-stored in the query library.
In the embodiment, the obtained continuous multi-frame point cloud is processed by using the three-dimensional point cloud data model reconstruction method, so that a target contour line with richer content can be obtained compared with a single-frame point cloud, and when the contour obtained by using the point cloud is used for re-identification, the method is accurate and reliable.
When the target is re-identified, the point cloud and the plane image may be combined together to re-identify the target.
In an embodiment of the present application, as shown in fig. 5, a method for re-identifying a vehicle (a type of target) is further provided, where the processor in fig. 1 is an edge server on the road side, the lidar is a lidar on the road side, and the device for acquiring a planar image in the method is a camera on the road side, and the method is applied to the edge server 104 in fig. 1 to perform the method for example, and the method includes:
step S410, point cloud data and a plane image of a target vehicle are obtained, wherein the point cloud data comprises a plurality of continuous frames of the target vehicle.
The point cloud data in this step may be point cloud data obtained by scanning the roadside laser radar 102 in a preset area. The planar image is captured by a conventional capturing device (e.g., a roadside camera), and appearance information (e.g., a body color, a model, a license plate color) of the target vehicle is included in the planar image.
Step S420, acquiring appearance information and two-dimensional characteristics of the target vehicle based on the plane image; and acquiring a contour curve of the target vehicle based on the point cloud data.
After the point cloud data and the plane image are obtained, the edge server 104 operates a preset algorithm to extract the characteristics of the point cloud data and the image, and obtains appearance information and two-dimensional characteristics of the target vehicle based on the plane image; and acquiring a contour curve of the target vehicle based on the point cloud data. Alternatively, a pre-trained feature extraction network (e.g., a CNN convolutional neural network) may be used to extract features in the planar image, so as to obtain two-dimensional features of the image.
Meanwhile, processing the point cloud data by using a point cloud three-dimensional reconstruction algorithm to obtain a three-dimensional point cloud model of the target vehicle; and calculating a contour curve of the target vehicle by using the three-dimensional point cloud model. Further, a three-dimensional model of the target vehicle may be constructed by using a 3D model construction algorithm (e.g., an ICP reconstruction algorithm), the outliers may be removed after down-sampling the constructed three-dimensional model, and then a minimum outer bounding box of the outliers may be calculated, and based on the minimum outer bounding box, a contour curve in the side view direction may be calculated, so as to obtain a contour curve of the target vehicle. Feature extraction may be performed on the target profile curve image to obtain a second multi-dimensional feature vector of the profile curve image. Isolated points can be removed by using a statistical Gaussian distribution method, the number of the neighborhood points is 20, and the allowed standard deviation of the average distance between the point clouds is 2 m.
Step S430, selecting a plurality of candidate re-identification information from the re-identification information of the query library by using the appearance information;
each candidate re-identification information corresponds to one candidate vehicle, a plurality of pieces of re-identification information are stored in the query library, and each piece of re-identification information comprises a vehicle identity, pre-stored appearance information, pre-stored two-dimensional characteristics and a pre-stored profile curve.
Candidate re-identification information matched with the appearance information can be firstly inquired from a query library based on the appearance information; because each piece of heavy identification information comprises the vehicle identity, the pre-stored appearance information, the pre-stored two-dimensional characteristic and the pre-stored profile curve, the two-dimensional characteristic and the pre-stored profile curve corresponding to the candidate vehicle are obtained after the candidate heavy identification information is obtained.
Step S440, calculating the similarity between the profile curve and each pre-stored profile curve to obtain a first similarity, and calculating the similarity between the two-dimensional characteristic and each pre-stored two-dimensional characteristic to obtain a second similarity.
Alternatively, the two-dimensional feature obtained in step S430 may be a vectorized feature. And then similarity calculation can be carried out by utilizing the vectorized data, and the operation process is simple. The first similarity may be a similarity based on areas of the profile curve and the pre-stored profile curve, or may be obtained by performing cosine similarity calculation on the profile curve and each of the pre-stored profile curves.
Step S450, outputting a re-recognition result based on the first similarity and the second similarity of each of the candidate vehicles.
The first similarity and the second similarity of each candidate vehicle can be weighted and calculated respectively to obtain a weighted similarity value; and sequencing the candidate vehicles according to the weighted similarity values, and outputting a re-identification result according to a sequencing result. The weighting value may be set according to actual needs, for example, a weighting value of 50% may be set for each of the first similarity value and the second similarity value, or a weighting value of 40% and a weighting value of 60% may be set for each of the first similarity value and the second similarity value. Through the weighted similarity value, the target vehicle can be identified more accurately. In addition, a preset threshold value can be further set, and the target vehicle can be identified again by comparing the preset threshold value with the weighted similarity value. When the threshold is used for re-identification, if the final weighting result is that the final weighting result exceeds the preset threshold, the query target vehicle is not in the query library, re-identification information of the target vehicle needs to be constructed based on the information of the target vehicle, and the re-identification information is added into the query library; if the similarity exceeds the preset threshold, the vehicle with the highest similarity can be set as the target recognition result.
According to the vehicle re-identification method, firstly, rough matching is carried out from a query library by utilizing the vehicle appearance to obtain a candidate vehicle, then, the multi-frame three-dimensional point cloud is utilized to obtain the side profile characteristics of the vehicle, fine matching is carried out by combining the image characteristics of the vehicle, and a target vehicle is obtained from the candidate vehicle.
If the vehicle re-identification method utilizes the three-dimensional point cloud model to obtain the contour of the target vehicle, the three-dimensional point cloud model can be constructed by utilizing the three-dimensional point cloud model construction method provided by the embodiments in the application, and how to construct the three-dimensional point cloud model can be referred to the related description in the above, and details are not repeated here.
In one embodiment, as shown in fig. 6, calculating the contour curve of the target vehicle according to the minimum outer bounding box may include:
step S421, obtaining the vertex coordinates of the minimum bounding box.
When this step is implemented, the minimum outer bounding rectangle of the vehicle can be first found by adopting an "AABB" mode, and the following vectors are respectively calculated:
Pmin=[xmin ymin zmin]
Pmax=[xmax ymax zmax]
c=(Pmin+Pmax)/2
s=Pmax-Pmin
wherein, Pmin,PmaxIs a pair of points of diagonal angles in the minimum outer surrounding frame, c is a coordinate of a central point, and s is PminPoint of direction PmaxVector of (2). From these four vectors, the coordinates of the 8 points of the smallest bounding box can then be calculated. Fig. 7 is a schematic diagram of the resulting minimum bounding box.
And step S422, calculating a vehicle head plane equation and a vehicle tail plane equation based on the plane formed by the vertex coordinates.
With continued reference to FIG. 7, we set the equations for calculating the plane 1 in which points 1, 2, 6, 5 lie and the plane 2 in which points 3, 4, 8, 7 lie.
And S423, acquiring a contour curve of the target vehicle according to the nose plane equation and the tail plane equation.
Specifically, the height values of the points within the range may be counted every preset distance according to the plane, and the average value of the first N points with the largest height value may be used as the height within the corresponding interval to obtain the profile curve. In connection with the example shown in fig. 7, starting from the plane 1, the height values of the points in the range are counted once every xcm (x is a preset value, which may be 10cm, for example), and the profile curve is obtained by taking the average value of the first n (n is a preset value, n is 5, for example) points with the largest height value as the height in this interval. Fig. 8 shows a contour curve effect diagram obtained by the method of the embodiment.
It should be understood that although the various steps in the flow charts of fig. 2-6 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 2-6 may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, which are not necessarily performed in sequence, but may be performed in turn or alternately with other steps or at least some of the other steps.
In one embodiment, as shown in fig. 9, there is provided a three-dimensional point cloud data model reconstruction apparatus, including:
an obtaining module 910, configured to perform target detection on multiple continuous frame point cloud data to obtain a target detection result of each frame of point cloud data; the target detection result of each frame of point cloud data comprises the same target object;
an axis vector calculating module 920, for calculating an axis vector according to a static point cloud in a plurality of continuous frame point cloud data, wherein the axis vector has a fixed included angle with the ground;
a registration module 930, configured to register the target detection result of each frame of point cloud data based on the axis vector, and output a three-dimensional point cloud data model including the same target object according to the registration result.
In one embodiment, the registration module 930 is configured to perform the steps of calculating a pose change between the current frame point cloud data and the previous frame point cloud data using the axis vector as a constraint, registering the detection result of the current frame point cloud data onto the global three-dimensional point cloud data model using the pose change until the target detection result of the last frame point cloud data is registered onto the global three-dimensional point cloud data model, outputting a three-dimensional reconstruction model including the same target object, wherein, the global three-dimensional point cloud data model is a three-dimensional point cloud data model obtained before the current frame point cloud data is registered, if the axis vector is vertical to the ground, the change value of the pose change value in the direction of the axis vector is 0, and if the axis vector is parallel to the ground, the change value of the pose change value in the direction perpendicular to the axis vector is 0.
In one embodiment, the registration module 930 takes the pose change between the current frame point cloud data and the previous frame point cloud data as the initial pose; and taking the axis vector as a limit, and adjusting the initial pose by using a fine registration algorithm so as to register the target detection result of the current frame point cloud data to the global three-dimensional point cloud data model.
In one embodiment, the registration module 930 performs inverse transformation on the global three-dimensional point cloud data model by using the registration parameters output in the current step, so that the global three-dimensional point cloud data model is moved to the pose of the current frame point cloud data.
In one embodiment, the registration module 930 calculates the pose change between the current frame point cloud data and the previous frame point cloud data by using an ICP algorithm to obtain the initial pose.
In one embodiment, the registration module 930 uses the axis vector as a constraint and utilizes a nic algorithm to adjust the initial pose so as to register the target detection result of the current frame point cloud data to the global three-dimensional point cloud data model.
In one embodiment, as shown in fig. 10, the present application further provides an object re-identification apparatus, including:
an obtaining module 11, configured to obtain point cloud data of a target, where the point cloud data includes multiple continuous frames of the target;
the matching module 12 is configured to construct a three-dimensional point cloud data model of the target by using the three-dimensional point cloud data model reconstruction method in the foregoing embodiment; and calculating a contour curve of the target using the three-dimensional model;
the similarity calculation module 13 is configured to obtain a re-recognition result matched from the query library by using the similarity between the profile curve and a pre-stored profile curve; the pre-stored profile curve is a pre-stored profile curve in a plurality of pieces of re-identification information pre-stored in the query library.
In one embodiment, as shown in fig. 11, the present application also provides a vehicle weight recognition apparatus including:
an obtaining module 21, configured to obtain point cloud data and a planar image of a target vehicle, where the point cloud data includes multiple continuous frames of the target vehicle;
a feature extraction module 22, configured to obtain appearance information and two-dimensional features of the target vehicle based on the planar image;
a matching module 23, configured to select multiple candidate re-identification information from the re-identification information in the query library by using the appearance information; each candidate re-identification information corresponds to one candidate vehicle, a plurality of re-identification information are stored in the query library, and each re-identification information comprises a vehicle identity, pre-stored appearance information, pre-stored two-dimensional characteristics and a pre-stored contour curve;
the similarity calculation module 24 is configured to calculate similarities between the contour curves and the pre-stored contour curves to obtain first similarities; calculating the similarity between the two-dimensional features and each pre-stored two-dimensional feature to obtain a second similarity; and outputting a re-recognition result based on the first similarity and the second similarity of each of the candidate vehicles.
In one embodiment, the feature extraction module 22 is configured to perform feature extraction on the planar image by using a trained feature extraction network to obtain a two-dimensional feature of the target vehicle.
In one embodiment, the feature extraction module 22 is configured to process the point cloud data by using a point cloud three-dimensional reconstruction algorithm to obtain a three-dimensional point cloud model of the target vehicle; and calculating a contour curve of the target vehicle by using the three-dimensional point cloud model.
In one embodiment, feature extraction module 22 is configured to calculate a minimum outer bounding box of the target vehicle using the three-dimensional point cloud model; and calculating the contour curve of the target vehicle according to the minimum outer surrounding frame.
In one embodiment, the feature extraction module 22 is configured to perform downsampling on the three-dimensional point cloud model and eliminate outliers.
In one embodiment, the feature extraction module 22 is configured to remove isolated points using a statistical gaussian distribution method, with a neighborhood number of 20, and an allowable average distance standard deviation across the point cloud of 2 m.
In one embodiment, feature extraction module 22 is configured to obtain vertex coordinates of the minimum bounding box; calculating a vehicle head plane equation and a vehicle tail plane equation based on the plane formed by the vertex coordinates; and acquiring a contour curve of the target vehicle according to the vehicle head plane equation and the vehicle tail plane equation.
In one embodiment, the feature extraction module 22 is configured to count height values of points within a range every preset distance according to a plane, and use an average value of the top N points with the largest height value as a height in a corresponding interval to obtain a profile curve.
In one embodiment, the similarity calculation module 24 is configured to output the re-recognition result based on the first similarity and the second similarity of each candidate vehicle, and includes: respectively carrying out weighted calculation on the first similarity and the second similarity of each candidate vehicle to obtain a weighted similarity value; and sequencing the candidate vehicles according to the weighted similarity values, and outputting a re-identification result according to a sequencing result.
In one embodiment, the similarity calculation module 24 is configured to compare the weighted similarity value with a preset threshold; and if the weighted similarity value exceeds the preset threshold value, identifying the candidate vehicle with the highest similarity in the weighted similarity values as a target vehicle.
In one embodiment, the similarity calculation module 24 is further configured to generate re-identification information of the target vehicle and add the re-identification information to the query library if the weighted similarity value does not exceed the preset threshold.
In one embodiment, the feature extraction module 22 is configured to perform target detection on a plurality of continuous frame point clouds to obtain a target detection result of each frame point cloud; the target detection result of each frame of point cloud comprises the same target object; calculating an axis vector according to a static target point in a plurality of continuous frame point clouds, wherein the axis vector and the ground have a fixed included angle; and registering the target detection result of each frame of point cloud based on the axis vector, and outputting a three-dimensional point cloud model containing the same target object according to the registration result.
Regarding the three-dimensional point cloud data model reconstruction device and the target re-identification device, the specific limitations of the vehicle re-identification device can be referred to the limitations of the three-dimensional point cloud data model reconstruction method, the target re-identification method and the vehicle re-identification method, and are not described herein again. The three-dimensional point cloud data model reconstruction device, the target re-identification device and each module in the vehicle re-identification device can be completely or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 12. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used for storing point cloud data or re-identification information. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a three-dimensional point cloud data model reconstruction method, a target re-identification method, and a vehicle re-identification method.
Those skilled in the art will appreciate that the architecture shown in fig. 12 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
carrying out target detection on a plurality of continuous frame point cloud data to obtain a target detection result of each frame of point cloud data; the target detection result of each frame of point cloud data comprises the same target object;
calculating an axis vector according to static point clouds in a plurality of continuous frame point cloud data, wherein the axis vector has a fixed included angle with the ground;
and registering the target detection result of each frame of point cloud data based on the axis vector, and outputting a three-dimensional point cloud data model containing the same target object according to the registration result.
In one embodiment, the processor, when executing the computer program, performs the following loop steps: calculating the pose change between the current frame point cloud data and the previous frame point cloud data by taking the axis vector as a limit, registering the detection result of the current frame point cloud data to a global three-dimensional point cloud data model by utilizing the pose change until the target detection result of the last frame point cloud data is registered to the global three-dimensional point cloud data model, and outputting a three-dimensional reconstruction model containing the same target object, wherein the global three-dimensional point cloud data model is a three-dimensional point cloud data model obtained before the current frame point cloud data is registered, if the axis vector is vertical to the ground, the pose change value of the pose change value in the direction of the axis vector is 0, and if the axis vector is parallel to the ground, the pose change value of the pose change value in the vertical direction of the axis vector is 0
In one embodiment, the processor, when executing the computer program, performs the steps of: taking the pose change between the current frame point cloud data and the previous frame point cloud data as an initial pose; and taking the axis vector as a limit, and adjusting the initial pose by using a fine registration algorithm so as to register the target detection result of the current frame point cloud data to the global three-dimensional point cloud data model.
In one embodiment, the processor, when executing the computer program, performs the steps of: and performing inverse transformation on the global three-dimensional point cloud data model by using the registration parameters output in the current step, so that the global three-dimensional point cloud data model is moved to the pose of the current frame point cloud data.
In one embodiment, the processor, when executing the computer program, performs the steps of: and calculating the pose change between the current frame point cloud data and the previous frame point cloud data by utilizing an ICP (inductively coupled plasma) algorithm to obtain the initial pose.
In one embodiment, the processor, when executing the computer program, performs the steps of: and taking the axis vector as a limit, and adjusting the initial pose by using an nick algorithm so as to register the target detection result of the current frame point cloud data to the global three-dimensional point cloud data model.
In one embodiment, another computer device is provided, comprising a memory having a computer program stored therein and a processor that when executed performs the steps of:
acquiring point cloud data of a target, wherein the point cloud data comprises a plurality of continuous frames of the target;
processing the plurality of continuous frames by using a three-dimensional point cloud data model reconstruction method to construct a three-dimensional point cloud data model of the target;
calculating a contour curve of the target using the three-dimensional model;
obtaining a re-recognition result matched from the query library by utilizing the similarity between the profile curve and a pre-stored profile curve; the pre-stored profile curve is a pre-stored profile curve in a plurality of pieces of re-identification information pre-stored in the query library.
In one embodiment, there is provided another computer apparatus comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program: in one embodiment, the processor when executing the computer program further performs the steps of:
acquiring point cloud data and a plane image of a target vehicle, wherein the point cloud data comprises a plurality of continuous frames of the target vehicle;
acquiring appearance information and two-dimensional characteristics of the target vehicle based on the plane image;
acquiring a contour curve of a target vehicle based on the point cloud data;
selecting a plurality of candidate re-identification information from the re-identification information of the query library by using the appearance information; each candidate re-identification information corresponds to one candidate vehicle, a plurality of re-identification information are stored in the query library, and each re-identification information comprises a vehicle identity, pre-stored appearance information, pre-stored two-dimensional characteristics and a pre-stored contour curve;
calculating the similarity between the contour curve and each pre-stored contour curve to obtain a first similarity;
calculating the similarity between the two-dimensional features and each pre-stored two-dimensional feature to obtain a second similarity;
and outputting a re-recognition result based on the first similarity and the second similarity of each candidate vehicle.
In one embodiment, the processor, when executing the computer program, performs the steps of: and performing feature extraction on the plane image by using the trained feature extraction network to obtain the two-dimensional features of the target vehicle.
In one embodiment, the processor, when executing the computer program, performs the steps of: processing the point cloud data by using a point cloud three-dimensional reconstruction algorithm to obtain a three-dimensional point cloud model of the target vehicle; and calculating a contour curve of the target vehicle by using the three-dimensional point cloud model.
In one embodiment, the processor, when executing the computer program, performs the steps of: calculating a minimum outer enclosure frame of the target vehicle by using the three-dimensional point cloud model; and calculating the contour curve of the target vehicle according to the minimum outer surrounding frame.
In one embodiment, the processor, when executing the computer program, performs the steps of: and performing down-sampling processing on the three-dimensional point cloud model, and removing outliers.
In one embodiment, the processor, when executing the computer program, performs the steps of: and removing isolated points by using a statistical Gaussian distribution method, wherein the number of the neighborhood points is 20, and the allowable cross-point cloud average distance standard deviation is 2 m.
In one embodiment, the processor, when executing the computer program, performs the steps of: acquiring the vertex coordinates of the minimum outer surrounding frame; calculating a vehicle head plane equation and a vehicle tail plane equation based on the plane formed by the vertex coordinates; and acquiring a contour curve of the target vehicle according to the vehicle head plane equation and the vehicle tail plane equation.
In one embodiment, the processor, when executing the computer program, performs the steps of: and according to the plane, counting the height values of the points within the range every preset distance, and taking the average value of the first N points with the maximum height value as the height in the corresponding interval to obtain the profile curve.
In one embodiment, the processor when executing the computer program implements the steps of: respectively carrying out weighted calculation on the first similarity and the second similarity of each candidate vehicle to obtain a weighted similarity value; and sequencing the candidate vehicles according to the weighted similarity values, and outputting a re-identification result according to a sequencing result.
In one embodiment, the processor, when executing the computer program, performs the steps of: comparing the weighted similarity value with a preset threshold value; and if the weighted similarity value exceeds the preset threshold value, identifying the candidate vehicle with the highest similarity in the weighted similarity values as a target vehicle.
In one embodiment, the processor, when executing the computer program, performs the steps of: and if the weighted similarity value does not exceed the preset threshold value, generating re-identification information of the target vehicle, and adding the re-identification information to the query library.
In one embodiment, the processor, when executing the computer program, performs the steps of: carrying out target detection on a plurality of continuous frame point clouds to obtain a target detection result of each frame point cloud; the target detection result of each frame of point cloud comprises the same target object; calculating an axis vector according to a static target point in a plurality of continuous frame point clouds, wherein the axis vector and the ground have a fixed included angle; and registering the target detection result of each frame of point cloud based on the axis vector, and outputting a three-dimensional point cloud model containing the same target object according to the registration result.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
carrying out target detection on a plurality of continuous frame point cloud data to obtain a target detection result of each frame of point cloud data; the target detection result of each frame of point cloud data comprises the same target object;
calculating an axis vector according to static point clouds in a plurality of continuous frame point cloud data, wherein the axis vector has a fixed included angle with the ground;
and registering the target detection result of each frame of point cloud data based on the axis vector, and outputting a three-dimensional point cloud data model containing the same target object according to the registration result.
In one embodiment, the computer program when executed by the processor implements the following loop steps: taking the axis vector as a limit, calculating the pose change between the current frame point cloud data and the previous frame point cloud data, registering the detection result of the current frame point cloud data to a global three-dimensional point cloud data model by using the pose change until the target detection result of the last frame point cloud data is registered to the global three-dimensional point cloud data model, and outputting a three-dimensional reconstruction model containing the same target object, wherein the global three-dimensional point cloud data model is a three-dimensional point cloud data model obtained before the current frame point cloud data is registered, if the axis vector is vertical to the ground, the pose change value of the pose change value in the direction of the axis vector is 0, and if the axis vector is parallel to the ground, the pose change value of the pose change value in the vertical direction of the axis vector is 0
In one embodiment, the computer program when executed by the processor implements the steps of: taking the pose change between the current frame point cloud data and the previous frame point cloud data as an initial pose; and taking the axis vector as a limit, and adjusting the initial pose by using a fine registration algorithm so as to register the target detection result of the current frame point cloud data to the global three-dimensional point cloud data model.
In one embodiment, the computer program when executed by the processor implements the steps of: and performing inverse transformation on the global three-dimensional point cloud data model by using the registration parameters output in the current step, so that the global three-dimensional point cloud data model is moved to the pose of the current frame point cloud data.
In one embodiment, the computer program when executed by the processor implements the steps of: and calculating the pose change between the current frame point cloud data and the previous frame point cloud data by utilizing an ICP (inductively coupled plasma) algorithm to obtain the initial pose.
In one embodiment, the computer program when executed by the processor implements the steps of: and taking the axis vector as a limit, and adjusting the initial pose by using an nick algorithm so as to register the target detection result of the current frame point cloud data to the global three-dimensional point cloud data model.
In one embodiment, another computer-readable storage medium is provided, having a computer program stored thereon, the computer program, when executed by a processor, implementing the steps of:
acquiring point cloud data of a target, wherein the point cloud data comprises a plurality of continuous frames of the target;
processing the plurality of continuous frames by using a three-dimensional point cloud data model reconstruction method to construct a three-dimensional point cloud data model of the target;
calculating a contour curve of the target using the three-dimensional model;
obtaining a re-recognition result matched from the query library by utilizing the similarity between the profile curve and a pre-stored profile curve; the pre-stored profile curve is a pre-stored profile curve in a plurality of pieces of re-identification information pre-stored in the query library.
In one embodiment, another computer-readable storage medium is provided, having a computer program stored thereon, the computer program, when executed by a processor, implementing the steps of:
acquiring point cloud data and a plane image of a target vehicle, wherein the point cloud data comprises a plurality of continuous frames of the target vehicle;
acquiring appearance information and two-dimensional characteristics of the target vehicle based on the plane image;
acquiring a contour curve of a target vehicle based on the point cloud data;
selecting a plurality of candidate re-identification information from the re-identification information of the query library by using the appearance information; each candidate re-identification information corresponds to one candidate vehicle, a plurality of re-identification information are stored in the query library, and each re-identification information comprises a vehicle identity, pre-stored appearance information, pre-stored two-dimensional characteristics and a pre-stored contour curve;
calculating the similarity between the contour curve and each pre-stored contour curve to obtain a first similarity;
calculating the similarity between the two-dimensional characteristics and each pre-stored two-dimensional characteristic to obtain a second similarity;
and outputting a re-recognition result based on the first similarity and the second similarity of each candidate vehicle.
In one embodiment, the computer program when executed by the processor implements the steps of: and performing feature extraction on the plane image by using the trained feature extraction network to obtain the two-dimensional features of the target vehicle.
In one embodiment, the computer program when executed by the processor implements the steps of: processing the point cloud data by using a point cloud three-dimensional reconstruction algorithm to obtain a three-dimensional point cloud model of the target vehicle; and calculating a contour curve of the target vehicle by using the three-dimensional point cloud model.
In one embodiment, the computer program when executed by the processor implements the steps of: calculating a minimum outer enclosure frame of the target vehicle by using the three-dimensional point cloud model; and calculating the contour curve of the target vehicle according to the minimum outer surrounding frame.
In one embodiment, the computer program when executed by the processor implements the steps of: and performing down-sampling processing on the three-dimensional point cloud model, and removing outliers.
In one embodiment, the computer program when executed by the processor implements the steps of: and removing isolated points by using a statistical Gaussian distribution method, wherein the number of the neighborhood points is 20, and the allowed standard deviation of the average distance between the point clouds is 2 m.
In one embodiment, the computer program when executed by the processor implements the steps of: acquiring the vertex coordinates of the minimum outer surrounding frame; calculating a vehicle head plane equation and a vehicle tail plane equation based on the plane formed by the vertex coordinates; and acquiring a contour curve of the target vehicle according to the vehicle head plane equation and the vehicle tail plane equation.
In one embodiment, the computer program when executed by the processor implements the steps of: and according to the plane, counting the height values of the points within the range every preset distance, and taking the average value of the first N points with the maximum height value as the height in the corresponding interval to obtain the profile curve.
In one embodiment, the computer program when executed by the processor implements the steps of: respectively carrying out weighted calculation on the first similarity and the second similarity of each candidate vehicle to obtain a weighted similarity value; and sequencing the candidate vehicles according to the weighted similarity values, and outputting a re-identification result according to a sequencing result.
In one embodiment, the computer program when executed by the processor implements the steps of: comparing the weighted similarity value with a preset threshold value; and if the weighted similarity value exceeds the preset threshold value, identifying the candidate vehicle with the highest similarity in the weighted similarity values as a target vehicle.
In one embodiment, the computer program when executed by the processor implements the steps of: and if the weighted similarity value does not exceed the preset threshold value, generating re-identification information of the target vehicle, and adding the re-identification information to the query library.
In one embodiment, the computer program when executed by the processor implements the steps of: carrying out target detection on a plurality of continuous frame point clouds to obtain a target detection result of each frame point cloud; the target detection result of each frame of point cloud comprises the same target object; calculating an axis vector according to a static target point in a plurality of continuous frame point clouds, wherein the axis vector and the ground have a fixed included angle; and registering the target detection result of each frame of point cloud based on the axis vector, and outputting a three-dimensional point cloud model containing the same target object according to the registration result.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (15)
1. A vehicle weight recognition method, characterized by comprising:
acquiring point cloud data and a plane image of a target vehicle, wherein the point cloud data comprises a plurality of continuous frames of the target vehicle;
acquiring appearance information and two-dimensional characteristics of the target vehicle based on the plane image;
acquiring a contour curve of a target vehicle based on the point cloud data;
selecting a plurality of candidate re-identification information from the re-identification information of the query library by using the appearance information; each candidate heavy identification information corresponds to one candidate vehicle, a plurality of heavy identification information is stored in the query library, and each heavy identification information comprises a vehicle identity, prestored appearance information, prestored two-dimensional characteristics and prestored profile curves;
calculating the similarity between the contour curve and each pre-stored contour curve to obtain a first similarity;
calculating the similarity between the two-dimensional features and each pre-stored two-dimensional feature to obtain a second similarity;
and outputting a re-recognition result based on the first similarity and the second similarity of each candidate vehicle.
2. The method of claim 1, wherein obtaining appearance information and two-dimensional features of the vehicle based on the planar image comprises:
and performing feature extraction on the plane image by using the trained feature extraction network to obtain the two-dimensional features of the target vehicle.
3. The method of claim 1, wherein obtaining contour information of a vehicle based on the point cloud data comprises:
processing the point cloud data by using a point cloud three-dimensional reconstruction algorithm to obtain a three-dimensional point cloud model of the target vehicle;
and calculating a contour curve of the target vehicle by using the three-dimensional point cloud model.
4. The method of claim 3, wherein calculating a contour curve of the target vehicle using the three-dimensional point cloud model comprises:
calculating a minimum outer enclosure frame of the target vehicle by using the three-dimensional point cloud model;
and calculating the contour curve of the target vehicle according to the minimum outer surrounding frame.
5. The method of claim 4, wherein prior to calculating the minimum outer bounding box of the target vehicle using the three-dimensional point cloud model, the method further comprises:
and performing down-sampling processing on the three-dimensional point cloud model, and removing outliers.
6. The method of claim 5, wherein downsampling the three-dimensional point cloud model and rejecting outliers comprises:
and removing isolated points by using a statistical Gaussian distribution method, wherein the number of the neighborhood points is 20, and the allowed standard deviation of the average distance between the point clouds is 2 m.
7. The method of any of claims 4-6, wherein calculating the contour curve of the target vehicle from the minimum outer bounding box comprises:
acquiring the vertex coordinates of the minimum outer surrounding frame;
calculating a vehicle head plane equation and a vehicle tail plane equation based on the plane formed by the vertex coordinates;
and acquiring a contour curve of the target vehicle according to the vehicle head plane equation and the vehicle tail plane equation.
8. The method of claim 7, wherein obtaining the contour curve of the target vehicle from the nose plane equation and the tail plane equation comprises:
and according to the plane, counting the height values of the points within the range every preset distance, and taking the average value of the first N points with the maximum height value as the height in the corresponding interval to obtain the profile curve.
9. The method according to any one of claims 1 to 6, wherein outputting a re-recognition result based on the first similarity and the second similarity of each of the candidate vehicles includes:
respectively carrying out weighted calculation on the first similarity and the second similarity of each candidate vehicle to obtain a weighted similarity value;
and sequencing the candidate vehicles according to the weighted similarity values, and outputting a re-identification result according to a sequencing result.
10. The method of claim 9, wherein ranking the plurality of candidate vehicles according to the weighted similarity values and outputting a re-recognition result according to the ranking result comprises:
comparing the weighted similarity value with a preset threshold value;
and if the weighted similarity value exceeds the preset threshold value, identifying the candidate vehicle with the highest similarity in the weighted similarity values as a target vehicle.
11. The method of claim 7, further comprising:
and if the weighted similarity value does not exceed the preset threshold value, generating re-identification information of the target vehicle, and adding the re-identification information to the query library.
12. The method of claim 3, wherein processing the point cloud data using a point cloud three-dimensional reconstruction algorithm to obtain a three-dimensional point cloud model of the target vehicle comprises:
carrying out target detection on a plurality of continuous frame point clouds to obtain a target detection result of each frame point cloud; the target detection result of each frame of point cloud comprises the same target object;
calculating an axis vector according to a static target point in a plurality of continuous frame point clouds, wherein the axis vector and the ground have a fixed included angle;
and registering the target detection result of each frame of point cloud based on the axis vector, and outputting a three-dimensional point cloud model containing the same target object according to the registration result.
13. A vehicle weight recognition apparatus, characterized by comprising:
the system comprises an acquisition module, a display module and a display module, wherein the acquisition module is used for acquiring point cloud data and a plane image of a target vehicle, and the point cloud data comprises a plurality of continuous frames of the target vehicle;
the feature extraction module is used for acquiring appearance information and two-dimensional features of the target vehicle based on the plane image;
the matching module is used for selecting a plurality of candidate re-identification information from the re-identification information of the query library by utilizing the appearance information; each candidate heavy identification information corresponds to one candidate vehicle, a plurality of heavy identification information is stored in the query library, and each heavy identification information comprises a vehicle identity, prestored appearance information, prestored two-dimensional characteristics and prestored profile curves;
the similarity calculation module is used for calculating the similarity between the contour curve and each pre-stored contour curve to obtain a first similarity; calculating the similarity between the two-dimensional characteristics and each pre-stored two-dimensional characteristic to obtain a second similarity; and outputting a re-recognition result based on the first similarity and the second similarity of each of the candidate vehicles.
14. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 12.
15. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 12.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011598701.1A CN114693863A (en) | 2020-12-29 | 2020-12-29 | Vehicle weight recognition method and device based on laser radar camera |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011598701.1A CN114693863A (en) | 2020-12-29 | 2020-12-29 | Vehicle weight recognition method and device based on laser radar camera |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114693863A true CN114693863A (en) | 2022-07-01 |
Family
ID=82131723
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011598701.1A Pending CN114693863A (en) | 2020-12-29 | 2020-12-29 | Vehicle weight recognition method and device based on laser radar camera |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114693863A (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103854320A (en) * | 2012-12-05 | 2014-06-11 | 上海海事大学 | Automatic vehicle type identification device and method based on laser radar |
CN104376334A (en) * | 2014-11-12 | 2015-02-25 | 上海交通大学 | Pedestrian comparison method based on multi-scale feature fusion |
CN105718882A (en) * | 2016-01-19 | 2016-06-29 | 上海交通大学 | Resolution adaptive feature extracting and fusing for pedestrian re-identification method |
US20170213093A1 (en) * | 2016-01-27 | 2017-07-27 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for detecting vehicle contour based on point cloud data |
CN110458086A (en) * | 2019-08-07 | 2019-11-15 | 北京百度网讯科技有限公司 | Vehicle recognition methods and device again |
CN111783654A (en) * | 2020-06-30 | 2020-10-16 | 苏州科达科技股份有限公司 | Vehicle weight identification method and device and electronic equipment |
-
2020
- 2020-12-29 CN CN202011598701.1A patent/CN114693863A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103854320A (en) * | 2012-12-05 | 2014-06-11 | 上海海事大学 | Automatic vehicle type identification device and method based on laser radar |
CN104376334A (en) * | 2014-11-12 | 2015-02-25 | 上海交通大学 | Pedestrian comparison method based on multi-scale feature fusion |
CN105718882A (en) * | 2016-01-19 | 2016-06-29 | 上海交通大学 | Resolution adaptive feature extracting and fusing for pedestrian re-identification method |
US20170213093A1 (en) * | 2016-01-27 | 2017-07-27 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for detecting vehicle contour based on point cloud data |
CN110458086A (en) * | 2019-08-07 | 2019-11-15 | 北京百度网讯科技有限公司 | Vehicle recognition methods and device again |
CN111783654A (en) * | 2020-06-30 | 2020-10-16 | 苏州科达科技股份有限公司 | Vehicle weight identification method and device and electronic equipment |
Non-Patent Citations (2)
Title |
---|
MATTHEW BERGER: "A Survey of Surface Reconstruction from Point Clouds", 《COMPUTER GRAPHICS FORUM》, 31 January 2017 (2017-01-31) * |
彭亲利: "基于建筑物重复特性的三维点云重建研究", 《中国优秀硕士学位论文全文库》, 15 February 2019 (2019-02-15) * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11488308B2 (en) | Three-dimensional object detection method and system based on weighted channel features of a point cloud | |
US10262190B2 (en) | Method, system, and computer program product for recognizing face | |
Nedevschi et al. | Stereo-based pedestrian detection for collision-avoidance applications | |
CN111353512B (en) | Obstacle classification method, obstacle classification device, storage medium and computer equipment | |
CN110879994A (en) | Three-dimensional visual inspection detection method, system and device based on shape attention mechanism | |
CN111709923B (en) | Three-dimensional object detection method, three-dimensional object detection device, computer equipment and storage medium | |
CN111723721A (en) | Three-dimensional target detection method, system and device based on RGB-D | |
EP3703008A1 (en) | Object detection and 3d box fitting | |
Zhang et al. | Fast moving pedestrian detection based on motion segmentation and new motion features | |
US20230021591A1 (en) | Model generation method, model generation apparatus, non-transitory storage medium, mobile object posture estimation method, and mobile object posture estimation apparatus | |
WO2021018690A1 (en) | Method for determining an environmental model of a scene | |
CN116246119A (en) | 3D target detection method, electronic device and storage medium | |
KR102592668B1 (en) | Facial recognition method and device | |
El Jaafari et al. | A novel approach for on-road vehicle detection and tracking | |
CN111353385A (en) | Pedestrian re-identification method and device based on mask alignment and attention mechanism | |
CN114693862A (en) | Three-dimensional point cloud data model reconstruction method, target re-identification method and device | |
CN114092850A (en) | Re-recognition method and device, computer equipment and storage medium | |
CN111339973A (en) | Object identification method, device, equipment and storage medium | |
Cai et al. | Deep representation and stereo vision based vehicle detection | |
CN114693863A (en) | Vehicle weight recognition method and device based on laser radar camera | |
KR20160148806A (en) | Object Detecter Generation Method Using Direction Information, Object Detection Method and Apparatus using the same | |
CN111986299A (en) | Point cloud data processing method, device, equipment and storage medium | |
EP3076370B1 (en) | Method and system for selecting optimum values for parameter set for disparity calculation | |
CN109146969B (en) | Pedestrian positioning method, device and processing equipment and storage medium thereof | |
CN111079551A (en) | Finger vein identification method and device based on singular value decomposition and storage medium |
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 | ||
TA01 | Transfer of patent application right | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20231212 Address after: 430200, 7th floor, Building 3, Phase II, Modern Service Industry Demonstration Base, Huazhong University of Science and Technology Science Park, Guandong Street, Donghu New Technology Development Zone, Wuhan City, Hubei Province Applicant after: Wuhan Wanji Photoelectric Technology Co.,Ltd. Address before: Wanji space, building 12, Zhongguancun Software Park, yard 8, Dongbei Wangxi Road, Haidian District, Beijing 100193 Applicant before: BEIJING WANJI TECHNOLOGY Co.,Ltd. |