CN116168090A - Equipment parameter calibration method and device - Google Patents
Equipment parameter calibration method and device Download PDFInfo
- Publication number
- CN116168090A CN116168090A CN202310449374.0A CN202310449374A CN116168090A CN 116168090 A CN116168090 A CN 116168090A CN 202310449374 A CN202310449374 A CN 202310449374A CN 116168090 A CN116168090 A CN 116168090A
- Authority
- CN
- China
- Prior art keywords
- parameters
- external
- point cloud
- parameter
- optimized
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 76
- 238000000611 regression analysis Methods 0.000 claims abstract description 121
- 230000007613 environmental effect Effects 0.000 claims description 70
- 230000008859 change Effects 0.000 abstract description 10
- 238000005457 optimization Methods 0.000 description 17
- 238000010586 diagram Methods 0.000 description 11
- 238000004422 calculation algorithm Methods 0.000 description 9
- 238000004891 communication Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 5
- 230000008447 perception Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- 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
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Traffic Control Systems (AREA)
Abstract
The disclosure provides a device parameter calibration method and device, wherein the method comprises the following steps: acquiring an environment image and point cloud data corresponding to a target vehicle; based on a preset initial external parameter and a preset internal parameter of the image acquisition equipment, matching point cloud data with an environment image and determining a first external parameter; based on the first regression analysis network and according to the preset initial external parameters and the preset internal parameters, matching point cloud data with the environment image and determining a second external parameter; and determining a first optimized external parameter according to the first external parameter and the second external parameter, and calibrating the first optimized external parameter as a target external parameter of the image acquisition equipment. By adopting the method, the problem that the calibration parameters of the image acquisition equipment are invalid due to the position change of the image acquisition equipment can be avoided, and the calibrated target external parameters are more accurate.
Description
Technical Field
The disclosure relates to the technical field of parameter calibration, in particular to a device parameter calibration method and device.
Background
With the development of automatic driving technology, the application fields of automatic driving vehicles are becoming wider and wider. In order to enable an autonomous vehicle to travel, it is generally necessary to install sensors for collecting environmental data on the autonomous vehicle, and precise calibration of parameters of the sensors is one of the prerequisites for the unmanned system to function properly.
Although the unmanned vehicle can be initialized and calibrated before leaving the factory, the position of a sensor of the unmanned vehicle may change in the long-term driving process, for example, the sensor may change due to loosening of a mounting structure, and the sensor may fail in parameters of the sensor calibrated before leaving the factory, so that the environmental data collected by the sensor is inaccurate, and the normal driving of the automatic driving vehicle is affected.
Therefore, how to design a method capable of accurately calibrating sensor parameters of an unmanned vehicle is important for normal running of the autonomous vehicle.
Disclosure of Invention
The disclosure provides a device parameter calibration method and device, which are used for at least solving the technical problems existing in the prior art.
According to a first aspect of the present disclosure, there is provided a device parameter calibration method, the method comprising:
acquiring an environment image and point cloud data corresponding to a target vehicle, wherein the environment image and the point cloud data are images and space points acquired for the target environment respectively;
based on a preset initial external parameter and a preset internal parameter of the image acquisition equipment, matching the point cloud data with the environment image, and determining a first external parameter;
Based on a first regression analysis network, matching the point cloud data with the environment image according to the preset initial external parameters and the preset internal parameters, and determining a second external parameter;
determining a first optimized external parameter according to the first external parameter and the second external parameter, and calibrating the first optimized external parameter as a target external parameter of the image acquisition equipment, wherein the target external parameter represents the attitude information of the image acquisition equipment relative to a radar.
In an embodiment, the method further comprises:
based on a second regression analysis network, matching the point cloud data with the environment image according to the first optimized external parameters and the preset internal parameters, and determining a second optimized external parameters;
and matching the point cloud data with the environment image according to the second optimized external parameters and the preset internal parameters, determining a third optimized external parameters, and calibrating the third optimized external parameters as target external parameters of the image acquisition equipment.
In an embodiment, the matching the point cloud data with the environmental image according to the second optimized external parameter and the preset internal parameter, determining a third optimized external parameter, and calibrating the third optimized external parameter as a target external parameter of the image acquisition device includes:
Determining the number of matching points, in the point cloud data, of the space points and the environment image based on the second optimized external parameters and the preset internal parameters;
determining the number of the current maximum space points matched with the environment image in the space points of the point cloud data according to the number of the matching points;
if the number of the matching points is larger than the current maximum number of the space points, determining the second optimized external parameters as third optimized external parameters, and updating the current maximum number of the space points to the number of the matching points;
if the number of the matching points is not greater than the current maximum number of the spatial points, determining the first optimized external parameters as third optimized external parameters;
determining whether the iteration number of the second regression analysis network reaches a preset iteration number;
if yes, the third optimized external parameter is calibrated as a target external parameter of the image acquisition equipment, and the iteration is terminated;
and if not, adjusting parameters of the second regression analysis network according to the number of the matching points, and after updating the preset initial external parameters by adopting the third optimized external parameters, returning to execute the preset initial external parameters and the preset internal parameters based on the image acquisition equipment, and matching the point cloud data with the environment image.
In an embodiment, the determining, based on the second optimized external parameter and the preset internal parameter, the number of matching points where the spatial points in the point cloud data match the environmental image includes:
projecting the spatial points in the point cloud data to a pixel plane according to the second optimized external parameters and the preset internal parameters;
and determining the number of spatial points matched with the environment image in the spatial points projected onto the pixel plane as the number of matched points.
In an embodiment, before the matching the point cloud data with the environmental image and determining the first external parameter based on the preset initial external parameter and the preset internal parameter of the image acquisition device, the method further includes:
extracting point cloud characteristics of the point cloud data and image characteristics of the environment image;
the matching the point cloud data and the environmental image based on the preset initial external parameters and the preset internal parameters of the image acquisition equipment, and determining a first external parameter comprises the following steps:
based on a preset initial external reference and a preset internal reference of the image acquisition equipment, projecting the space points in the point cloud data to a pixel plane;
and determining the number of first space points which are projected to the space points on the pixel plane and matched with the environment image according to the image characteristics and the point cloud characteristics, and determining the matching parameters corresponding to the number of the first space points as first external parameters.
In an embodiment, the matching the point cloud data and the environmental image based on the first regression analysis network and according to the preset initial external parameters and the preset internal parameters, and determining the second external parameters includes:
inputting the preset initial external parameters, the preset internal parameters, the point cloud features and the image features into a first regression analysis network to obtain the number of second space points matched with the environment image in the space points of the point cloud data, and determining the current parameters of the first regression analysis network as the second external parameters.
In an embodiment, the determining a first optimized external parameter according to the first external parameter and the second external parameter includes:
comparing the first space point number with the second space point number;
if the first number of spatial points is greater than the second number of spatial points, determining the first outlier as a first optimized outlier;
and if the first number of space points is not greater than the second number of space points, determining the second external parameters as first optimized external parameters.
In an embodiment, the method further comprises:
and if the number of the first space points is larger than the number of the second space points, adjusting parameters of the first regression analysis network, adopting the first optimized external parameters to update the preset initial external parameters, and returning to execute the preset initial external parameters and the preset internal parameters based on the image acquisition equipment to match the point cloud data with the environment image.
According to a second aspect of the present disclosure, there is provided a device parameter calibration method applied to a vehicle-mounted computing platform, the method comprising:
acquiring an environment image and point cloud data corresponding to a target vehicle, wherein the environment image and the point cloud data are images and space points acquired for the target environment respectively;
based on a preset initial external parameter and a preset internal parameter of the image acquisition equipment, matching the point cloud data with the environment image, and determining a first external parameter;
based on a first regression analysis network, matching the point cloud data with the environment image according to the preset initial external parameters and the preset internal parameters, and determining a second external parameter;
determining a first optimized external parameter according to the first external parameter and the second external parameter;
the first optimized external parameters are sent to a cloud computing platform, so that the cloud computing platform matches the point cloud data with the environment image based on a second regression analysis network according to the first optimized external parameters and the preset internal parameters, after a second optimized external parameters are determined, matches the point cloud data with the environment image according to the second optimized external parameters and the preset internal parameters, determines a third optimized external parameters, and sends the third optimized external parameters to the vehicle-mounted computing platform;
And receiving the third optimized external parameters sent by the vehicle-mounted computing platform, and calibrating the third optimized external parameters as target external parameters of the image acquisition equipment, wherein the target external parameters represent the attitude information of the image acquisition equipment relative to a radar.
According to a third aspect of the present disclosure, there is provided an apparatus parameter calibration device, the device comprising:
the first data acquisition module is used for acquiring an environment image and point cloud data corresponding to a target vehicle, wherein the environment image and the point cloud data are images and space points acquired for the target environment respectively;
the first matching module is used for matching the point cloud data with the environment image based on a preset initial external parameter and a preset internal parameter of the image acquisition equipment and determining a first external parameter;
the second matching module is used for matching the point cloud data with the environment image based on the first regression analysis network and according to the preset initial external parameters and the preset internal parameters, and determining a second external parameter;
the first parameter calibration module is used for determining a first optimized external parameter according to the first external parameter and the second external parameter, and calibrating the first optimized external parameter as a target external parameter of the image acquisition equipment, wherein the target external parameter represents the attitude information of the image acquisition equipment relative to a radar.
According to a fourth aspect of the present disclosure, there is provided an electronic device comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the methods described in the present disclosure.
According to a fifth aspect of the present disclosure, there is provided a non-transitory computer readable storage medium storing computer instructions for causing a computer to perform the method described in the present disclosure.
By adopting the method and the device provided by the embodiment of the disclosure, the environment image and the point cloud data corresponding to the target vehicle are acquired; based on a preset initial external parameter and a preset internal parameter of the image acquisition equipment, matching point cloud data with an environment image and determining a first external parameter; based on the first regression analysis network and according to the preset initial external parameters and the preset internal parameters, matching point cloud data with the environment image and determining a second external parameter; and determining a first optimized external parameter according to the first external parameter and the second external parameter, and calibrating the first optimized external parameter as a target external parameter of the image acquisition equipment. The external parameters of the image acquisition equipment are not affected by the position of the image acquisition equipment when calibrated, so that the problem that the calibration parameters of the image acquisition equipment are invalid due to the position change of the image acquisition equipment can be avoided. And the regression analysis network is adopted to optimize the external parameters of the image acquisition equipment, so that the marked target external parameters are more accurate.
It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the disclosure, nor is it intended to be used to limit the scope of the disclosure. Other features of the present disclosure will become apparent from the following specification.
Drawings
The above, as well as additional purposes, features, and advantages of exemplary embodiments of the present disclosure will become readily apparent from the following detailed description when read in conjunction with the accompanying drawings. Several embodiments of the present disclosure are illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings, in which:
in the drawings, the same or corresponding reference numerals indicate the same or corresponding parts.
FIG. 1 is a schematic flow chart of an implementation of a device parameter calibration method according to an embodiment of the disclosure;
FIG. 2 illustrates a point cloud and image matching flow chart provided by an embodiment of the present disclosure;
fig. 3 illustrates a schematic diagram of matching a point cloud with an image according to an embodiment of the disclosure;
FIG. 4 illustrates an optimization exogenous determination flow chart provided by an embodiment of the present disclosure;
FIG. 5 shows a schematic diagram of an external parameter calibration flow provided by an embodiment of the present disclosure;
FIG. 6 is a schematic flow chart of a method for calibrating parameters of a device according to an embodiment of the disclosure;
FIG. 7 is a schematic flow chart of an implementation of the method for calibrating device parameters applied to a vehicle-mounted computing platform according to an embodiment of the disclosure;
FIG. 8 is a schematic structural diagram of an apparatus parameter calibration device according to an embodiment of the present disclosure;
FIG. 9 is a schematic structural diagram of an apparatus parameter calibration device applied to a vehicle-mounted computing platform according to an embodiment of the disclosure;
fig. 10 shows a schematic diagram of a composition structure of an electronic device according to an embodiment of the present disclosure.
Detailed Description
In order to make the objects, features and advantages of the present disclosure more comprehensible, the technical solutions in the embodiments of the present disclosure will be clearly described in conjunction with the accompanying drawings in the embodiments of the present disclosure, and it is apparent that the described embodiments are only some embodiments of the present disclosure, but not all embodiments. Based on the embodiments in this disclosure, all other embodiments that a person skilled in the art would obtain without making any inventive effort are within the scope of protection of this disclosure.
Because the existing sensor parameter calibration method can cause calibration parameter failure due to the position change of the sensor, the normal running of the automatic driving vehicle is affected. Therefore, in order to avoid such a problem, the embodiment of the disclosure provides a device parameter calibration method and device. The method provided by the embodiment of the disclosure can be applied to a vehicle-mounted computing platform configured in a vehicle and also can be applied to a cloud computing platform.
The technical solutions of the embodiments of the present disclosure will be described below with reference to the drawings in the embodiments of the present disclosure.
Fig. 1 shows a schematic flow chart of an implementation of a device parameter calibration method according to an embodiment of the disclosure, as shown in fig. 1, where the method includes:
s101, acquiring an environment image and point cloud data corresponding to a target vehicle.
The environment image and the point cloud data are respectively an image and a space point which are acquired aiming at a target environment, and the target environment is the environment where the target vehicle is located.
In the present disclosure, an environmental image for a target environment may be acquired by an image acquisition device configured on a target vehicle and/or an environmental image for a target environment may be acquired by an image acquisition device configured in an edge-aware terminal. After the edge perception terminal collects the environment image, if the electronic device executing the device parameter calibration method provided by the disclosure is a vehicle-mounted computing platform, the edge perception terminal can send the collected environment image to the vehicle-mounted computing platform through the internet of things, and if the electronic device executing the device parameter calibration method provided by the disclosure is a cloud computing platform, the edge perception terminal can send the collected environment image to the cloud computing platform through the internet of things.
In the present disclosure, the point cloud data refers to radar point clouds acquired by a radar device for a target environment.
S102, matching the point cloud data with the environment image based on a preset initial external parameter and a preset internal parameter of the image acquisition equipment, and determining a first external parameter.
In the present disclosure, the preset initial external parameters refer to initial parameters set for the target external parameters according to the actual application scenario. The preset internal parameters refer to internal parameters of an image pickup apparatus disposed on the target vehicle, for example, if the image pickup apparatus is an in-vehicle camera, the preset internal parameters are internal parameters of the in-vehicle camera of the target vehicle.
In an embodiment, before the matching the point cloud data with the environmental image and determining the first external parameter based on the preset initial external parameter and the preset internal parameter of the image capturing device, the device parameter calibration method further includes step A1:
and A1, extracting point cloud characteristics of the point cloud data and image characteristics of the environment image.
In the disclosure, any feature extraction algorithm may be used to extract the point cloud features of the point cloud data and the image features of the environmental image, for example, a convolutional neural network may be used to extract the convolutional features of the environmental image as the image features, and the convolutional features of the point cloud data may be extracted as the point cloud features. Alternatively, semantic segmentation operators can be used to extract point cloud features of the point cloud data and image features of the environmental image.
In the disclosure, the point cloud features reflect semantic information of point cloud regions corresponding to different targets in the point cloud data, and the image features reflect semantic information of image regions corresponding to different targets in the environmental image.
On the basis of step A1, fig. 2 shows a flowchart for matching point cloud with an image, where, as shown in fig. 2, the matching the point cloud data with the environmental image based on a preset initial external parameter and a preset internal parameter of an image acquisition device, and determining a first external parameter may include:
s201, based on preset initial external parameters and preset internal parameters of the image acquisition equipment, the spatial points in the point cloud data are projected to a pixel plane.
In this step, a pixel plane may be constructed on the environmental image using a preset initial external parameter and a preset internal parameter of the image capturing device. For example, fig. 3 shows a schematic diagram of matching a point cloud with an image, where, as shown in fig. 3, a pixel plane may be constructed on an environmental image 301 by using a preset initial external parameter and a preset internal parameter of an image acquisition device, and a spatial point in point cloud data 302 is projected onto the pixel plane, so as to obtain an image 303 where the spatial point coincides with the environmental image.
S202, according to the image features and the point cloud features, determining the number of first space points which are projected to the space points on the pixel plane and matched with the environment image, and determining matching parameters corresponding to the number of the first space points as first external parameters.
Because the point cloud features reflect the semantic information of the point cloud regions corresponding to different targets, and the image features reflect the semantic information of the image regions corresponding to different targets, in this step, for each spatial point, whether the semantic information of the spatial point is matched with the semantic information of the pixel point in the environmental image corresponding to the spatial point can be determined according to the image features and the point cloud features, and if so, the spatial point is determined to be the first spatial point which is projected to the spatial point on the pixel plane and is matched with the environmental image. Specifically, according to the image characteristics and the Point cloud characteristics, a PnP (periodic-n-Point) algorithm is adopted to calculate spatial parameters of the image acquisition device and the radar. Specifically, the number of first space points can be determined according to whether the semantic information of each space point is matched with the semantic category of the pixel point in the environmental image corresponding to the space point, and the algorithm parameter for solving the number of the first space points is determined as the first external parameter.
Still referring to fig. 3 as an example, after the space points in the point cloud data 302 are projected onto the pixel plane to obtain an image 303 where the space points coincide with the environmental image, according to the point cloud features of the point cloud data 302 and the image features of the environmental image 301, it may be determined whether the semantic category represented by each space point in the image 303 where the space points coincide with the environmental image is consistent with the semantic category represented by the pixel point corresponding to the space point in the environmental image 301, and if the semantic category is consistent, the space point is represented as the first space point matched with the environmental image. For example, if the semantic category of the spatial point a304 on the person object in the image 303 where the spatial point coincides with the environmental image is a person, it may be determined that the spatial point a304 is the same as the semantic category corresponding to the pixel point corresponding to the spatial point a304 in the environmental image 301, and both represent the person, and it may be determined that the spatial point a304 is the first spatial point matched with the environmental image 301; if the semantic category of the spatial point B305 on the person object in the image 303 where the spatial point coincides with the environmental image is a vehicle, it may be determined that the semantic category of the spatial point B305 is different from the semantic category corresponding to the pixel point corresponding to the spatial point B305 in the environmental image 301, and the semantic category corresponding to the pixel point corresponding to the spatial point B305 in the environmental image 301 is a person, and it may be determined that the spatial point B305 is not matched with the environmental image 301, that is, it may be determined that the spatial point B305 is an erroneous projection point.
In another embodiment, the matching the point cloud data with the environmental image based on the preset initial external parameter and the preset internal parameter of the image capturing device, and determining the first external parameter may include: and (3) inputting preset initial external parameters and preset internal parameters of the image acquisition equipment, point cloud data and an environment image into the point cloud and image matching model by adopting a pre-trained point cloud and image matching model, and determining the model parameters of the point cloud and image matching model corresponding to the maximum number of the first space points matched with the environment image in the space points in the output point cloud data as the first external parameters by adjusting the model parameters of the point cloud and image matching model.
And S103, matching the point cloud data with the environment image based on a first regression analysis network and according to the preset initial external parameters and the preset internal parameters, and determining a second external parameter.
In the present disclosure, based on a first regression analysis network and according to the preset initial external parameters and the preset internal parameters, matching the point cloud data with the environmental image, and determining a second external parameter, specifically may include step B1:
And B1, inputting the preset initial external parameters, the preset internal parameters, the point cloud characteristics and the image characteristics into a first regression analysis network to obtain the number of second space points matched with the environment image in the space points of the point cloud data, and determining the current parameters of the first regression analysis network as the second external parameters.
In the disclosure, the first regression analysis network may adopt a lightweight regression analysis network, then input preset initial external parameters, preset internal parameters, point cloud features and image features into the lightweight regression analysis network, output a second spatial point matched with an environmental image among spatial points of point cloud data, and then determine the number of the second spatial points, and take the current parameters of the lightweight regression analysis network as the second external parameters.
S104, determining a first optimized external parameter according to the first external parameter and the second external parameter, and calibrating the first optimized external parameter as a target external parameter of the image acquisition equipment.
The target external parameters represent attitude information of the image acquisition equipment relative to a radar.
In the disclosure, determining a first optimized external parameter according to the first external parameter and the second external parameter may specifically include the following steps C1-C3:
And C1, comparing the first space point number with the second space point number.
And C2, if the first number of space points is larger than the second number of space points, determining the first external parameters as first optimized external parameters.
And C3, if the first number of space points is not larger than the second number of space points, determining the second external parameters as first optimized external parameters.
If the number of the first space points is larger than that of the second space points, the matching effect of the first external parameters on the environment image and the point cloud data is better, so that the first external parameters can be determined to be the first optimized external parameters, and the target parameters are calibrated by the first optimized external parameters. If the number of the first space points is not larger than the number of the second space points, the matching effect of the second external parameters on the environment image and the point cloud data is better, so that the second external parameters can be determined to be the first optimized external parameters, and the target parameters are calibrated by the first optimized external parameters.
Also, in an embodiment, the method further includes step D1:
and D1, if the number of the first space points is larger than the number of the second space points, adjusting parameters of the first regression analysis network, updating the preset initial external parameters by adopting the first optimized external parameters, and returning to execute the preset initial external parameters and the preset internal parameters based on the image acquisition equipment to match the point cloud data with the environment image.
Specifically, if the number of the first space points is greater than the number of the second space points, it is also indicated that the matching effect of the second external parameter corresponding to the current first regression analysis network on the environmental image and the point cloud data is not as good as that of the first external parameter, so in order to further iterate the first regression analysis network to make the matching positive and accurate, the second space point number determined based on the first regression analysis network currently can be determined as a negative sample and fed back to the first regression analysis network, so that parameters of the first regression analysis network can be adjusted according to the direction of the negative sample. If the number of the first space points is not greater than the number of the second space points, the matching effect of the second external parameters corresponding to the current first regression analysis network on the environmental image and the point cloud data is also stronger than that of the first external parameters, so that in order to further iterate the first regression analysis network to enable the matching to be accurate, the number of the second space points determined based on the first regression analysis network can be determined to be positive samples and fed back to the first regression analysis network, and parameters of the first regression analysis network can be adjusted according to the direction of the positive samples. And after the first optimized external parameter is calibrated as the target external parameter of the image acquisition device, continuing to execute the step S102 until the number of the second space points determined based on the first regression analysis network is greater than the number of the first space points, and when the number of the second space points determined after the plurality of iterations of the first regression analysis network is within a stable interval, the iteration of the first regression analysis network can be determined to be completed, and the second parameter determined by the last iteration of the first regression analysis network is calibrated as the target external parameter of the image acquisition device.
By adopting the method provided by the embodiment of the disclosure, the environment image and the point cloud data corresponding to the target vehicle are acquired; based on a preset initial external parameter and a preset internal parameter of the image acquisition equipment, matching point cloud data with an environment image and determining a first external parameter; based on the first regression analysis network and according to the preset initial external parameters and the preset internal parameters, matching point cloud data with the environment image and determining a second external parameter; and determining a first optimized external parameter according to the first external parameter and the second external parameter, and calibrating the first optimized external parameter as a target external parameter of the image acquisition equipment. The external parameters of the image acquisition equipment are not affected by the position of the image acquisition equipment when calibrated, so that the problem that the calibration parameters of the image acquisition equipment are invalid due to the position change of the image acquisition equipment can be avoided. And the regression analysis network is adopted to optimize the external parameters of the image acquisition equipment, so that the marked target external parameters are more accurate.
In an implementation manner, fig. 4 shows an optimization external parameter determination flowchart provided by an embodiment of the disclosure, and as shown in fig. 4, the device parameter calibration method may further include:
s401, matching the point cloud data with the environment image based on a second regression analysis network and according to the first optimized external parameters and the preset internal parameters, and determining a second optimized external parameters.
In the disclosure, the second regression analysis network may employ a complex regression analysis network, then input preset initial external parameters, preset internal parameters, point cloud data and an environmental image into the complex regression analysis network, output the spatial points matched with the environmental image among the spatial points of the point cloud data, then determine the number of matched spaces, and take the current parameters of the complex regression analysis network as the second optimized external parameters.
S402, matching the point cloud data with the environment image according to the second optimized external parameters and the preset internal parameters, determining a third optimized external parameters, and calibrating the third optimized external parameters as target external parameters of the image acquisition equipment.
In this disclosure, fig. 5 shows a schematic diagram of an external parameter calibration flow provided by an embodiment of the disclosure, as shown in fig. 5, the matching the point cloud data and the environmental image according to the second optimized external parameter and the preset internal parameter, and determining a third optimized external parameter, where the third optimized external parameter is calibrated as a target external parameter of the image capturing device may include:
s501, determining the number of matching points, in which the space points in the point cloud data are matched with the environment image, based on the second optimized external parameters and the preset internal parameters.
Specifically, in an embodiment, the determining, based on the second optimized external parameter and the preset internal parameter, the number of matching points where the spatial points in the point cloud data match the environmental image may include the following steps E1-E2:
and E1, projecting the space points in the point cloud data to a pixel plane according to the second optimized external parameters and the preset internal parameters.
And E2, determining the number of space points which are projected to the space points on the pixel plane and matched with the environment image as the number of matching points.
Because the point cloud features reflect the semantic information of the point cloud regions corresponding to different targets, and the image features reflect the semantic information of the image regions corresponding to different targets, in this embodiment, the image features of the environmental image and the point cloud features of the point cloud data may be extracted in advance. And projecting the space points in the point cloud data onto the pixel plane where the environment image is located according to the second optimized external parameters and the preset internal parameters. Then, for each space point, according to the image characteristics and the point cloud characteristics, whether the semantic category corresponding to the space point is matched with the semantic category of the pixel point in the environment image corresponding to the space point or not can be determined, and if so, the space point is determined to be a matching point matched with the environment image. Specifically, spatial parameters of the image acquisition device and the radar can be solved by adopting a PnP algorithm according to the image characteristics and the point cloud characteristics, specifically, position parameters when the semantic category of each spatial point is matched with the semantic category of the pixel point in the environment image corresponding to the spatial point can be solved, and the number of the matched points is determined according to the position parameters.
In another embodiment, the determining, based on the second optimized external parameter and the preset internal parameter, the number of matching points where the spatial points in the point cloud data match the environmental image may include: and (3) inputting preset initial external parameters, preset internal parameters, point cloud data and an environment image into the point cloud and image matching model by adopting a pre-trained point cloud and image matching model, enabling the number of spatial points matched with the environment image in the spatial points in the output point cloud data to be maximum by adjusting model parameters of the point cloud and image matching model, and determining the maximum number of the spatial points output as the number of matched points matched with the environment image in the point cloud data.
S502, determining the number of the current maximum space points matched with the environment image in the space points of the point cloud data according to the number of the matching points.
Specifically, if the number of the current matching points is determined based on the second optimized external parameters output after the first iteration of the second regression analysis network, the number of the current matching points can be determined as the number of the current maximum spatial points; if the number of the current matching points is determined based on the second optimized external parameters output by the second regression analysis network after multiple iterations, the number of the matching points is determined according to the second optimized external parameters output by the second regression analysis network after the previous iteration before the number of the current matching points, and the maximum number in the number of the current matching points and the number of the matching points obtained before can be determined as the current maximum number of the space points.
S503, if the number of the matching points is larger than the current maximum number of the space points, determining the second optimized external parameters as third optimized external parameters, and updating the current maximum number of the space points as the number of the matching points.
S504, if the number of the matching points is not greater than the current maximum number of the spatial points, determining the first optimized external parameters as third optimized external parameters.
S505, determining whether the iteration number of the second regression analysis network reaches a preset iteration number.
The preset iteration number may be set according to an actual application scenario, for example, may be set to 1000 times or 2000 times, etc.
S506, if so, calibrating the third optimized external parameter as the target external parameter of the image acquisition equipment and terminating the iteration.
S507, if not, adjusting parameters of the second regression analysis network according to the number of the matching points, and after updating the preset initial external parameters by adopting the third optimized external parameters, returning to execute the preset initial external parameters and the preset internal parameters based on the image acquisition equipment, and matching the point cloud data with the environment image.
In this step, if the number of the current matching points is greater than the number of the current maximum spatial points, the number of the current matching points may be marked as a positive sample and fed back to the second regression analysis network, so that the second regression analysis network may adjust network parameters according to the direction of the positive sample, so that the matching between the point cloud and the image may be better performed by the second optimized external parameters output by the second regression analysis network after the parameters are adjusted, the second optimized external parameters are determined as third optimized external parameters, and after the preset initial external parameters are updated by using the third optimized external parameters, the step of executing the preset initial external parameters and the preset internal parameters based on the image acquisition device is returned to perform matching between the point cloud data and the environmental image, so as to optimize the second regression analysis network.
If the number of the current matching points is not greater than the number of the current maximum space points, which indicates that a large lifting space exists for the matching effect of the second optimization external parameters output by the second regression analysis network on the point cloud and the image, the number of the current matching points can be marked as a negative sample and fed back to the second regression analysis network, so that the second regression analysis network can adjust network parameters according to the direction of the negative sample, the matching of the point cloud and the image can be better achieved by the second optimization external parameters output by the second regression analysis network after the parameters are adjusted, the first optimization external parameters are determined to be third optimization external parameters, and after the third optimization external parameters are used for updating the preset initial external parameters, the step of executing the matching of the point cloud data and the environment image based on the preset initial external parameters and the preset internal parameters is returned to optimize the second regression analysis network.
And calibrating the third optimized external parameters as target external parameters of the image acquisition equipment until the iteration times of the second regression analysis network reach the preset iteration times, or when the number of the plurality of matching points determined based on the plurality of second optimized external parameters output by the second regression analysis network after the plurality of iterations is in a stable interval.
By adopting the method provided by the embodiment of the disclosure, the external parameters of the image acquisition equipment are not affected by the position of the image acquisition equipment when calibrated, so that the problem of invalid calibration parameters of the image acquisition equipment caused by the position change of the image acquisition equipment can be avoided. And on the basis of optimizing the external parameters of the image acquisition equipment by using the first regression analysis network, the external parameters of the image acquisition equipment are further optimized by using the second regression analysis network, so that the accuracy of the marked target external parameters is further improved.
Fig. 6 illustrates a specific embodiment of an apparatus parameter calibration method provided by an embodiment of the present disclosure, and fig. 6 illustrates a flowchart of an apparatus parameter calibration method provided by an embodiment of the present disclosure. As shown in fig. 6, the electronic device performing the device parameter calibration method provided in the present disclosure may be a vehicle-mounted computing platform or a cloud computing platform.
As shown in fig. 6, an environment image and point cloud data may be acquired through an on-vehicle sensor 601 configured on a target vehicle, where the environment image is an image acquired by an image acquisition device configured in the target vehicle for a target environment, the point cloud data is acquired by a radar device, and/or the environment image and the point cloud data for the target environment are acquired through a radar device configured by an image acquisition device in an edge sensing terminal 602, where after the edge sensing terminal 602 acquires the environment image, if an electronic device that performs a device parameter calibration method provided by the present disclosure is an on-vehicle computing platform 603, the edge sensing terminal may send the acquired environment image and point cloud data to an on-vehicle gateway 605 through an internet of things 604, and then send the environment image and the point cloud data to the on-vehicle computing platform through the on-vehicle gateway, and if an electronic device that performs a device parameter calibration method provided by the present disclosure is a cloud computing platform 606, the edge sensing terminal may send the acquired environment image and the point cloud data to the on-vehicle computing platform through the internet of things.
As also shown in fig. 6, the image data refers to an environmental image, and the Lei Dadian cloud is pointing cloud data. The present embodiment can divide parameter calibration into three phases: the method comprises a semantic perception stage, an external parameter calculation optimization stage and an external parameter refinement adjustment optimization stage. The semantic perception stage is to extract image semantic features of an environment image in the data cloud by using an image semantic segmentation operator, extract point cloud semantic features of point cloud data in the data cloud by using a point cloud semantic segmentation operator, and extract the extracted image semantic features, namely image features of the environment image, and the extracted point cloud semantic features, namely point cloud features of the point cloud data.
As also shown in fig. 6, the extrinsic computation optimization phase: the second external parameters are determined by using the preset internal parameters, the image semantic features, the point cloud semantic features, the preset initial external parameters and the lightweight regression analysis network, specifically, the optimized external parameters output by the lightweight regression analysis network are determined to be the second external parameters, and the specific process of determining the second external parameters can refer to the foregoing embodiments and will not be repeated here. And projecting the space points in the point cloud data to the pixel plane by using the preset initial external parameters and the preset internal parameters, determining the first space point quantity matched with the environment image in the space points projected to the pixel plane according to semantic information comprising the image semantic features and the point cloud semantic features, and determining the matching parameters corresponding to the first space point quantity as the first external parameters. Specifically, pnP calculation can be performed according to a matching relationship between the spatial points projected onto the pixel plane and the environmental image, so as to determine the number of first spatial points, and then a matching parameter corresponding to the number of first spatial points is determined as a first external parameter. The method for determining the first external parameter may refer to the foregoing embodiments, and will not be described herein. And, according to the first external parameters determined based on the PnP algorithm and output by the lightweight regression analysis network in fig. 6, the external parameters with the largest number of matching points corresponding to projection matching in the first external parameters and the second external parameters output by using the two matching paths are determined as the first optimized external parameters. In addition, positive and negative samples can be judged according to the first optimized external parameters, and training network parameters can be adjusted, specifically, when the number of matching points for projection matching by using the first external parameters is larger than that of matching points for projection matching by using the second external parameters, the matching effect of the environment image and the point cloud data by the second external parameters corresponding to the current lightweight regression analysis network is not as good as that of the first external parameters, so that the second external parameters determined based on the lightweight regression analysis network can be judged to be positive and accurate for further iterating the lightweight regression analysis network, and the negative samples can be fed back to the lightweight regression analysis network, so that the parameters of the lightweight regression analysis network can be adjusted according to the direction of the negative samples; when the number of the matching points for projection matching by using the first external parameters is not greater than the number of the matching points for projection matching by using the second external parameters, the second external parameters corresponding to the current lightweight regression analysis network can be marked as positive samples and fed back to the lightweight regression analysis network, so that the lightweight regression analysis network can adjust network parameters according to the direction of the positive samples. And the first optimized external parameters can be used as the optimal external parameters to update the preset initial external parameters, and the initial step of executing the external parameter calculation optimization stage is returned to update until the number of the matching points for projection matching by using the first external parameters is larger than the number of the matching points for projection matching by using the second external parameters, and the lightweight regression analysis network can determine that the lightweight regression analysis network is completed in an iteration mode when the number of the matching points determined by the plurality of iterations of the lightweight regression analysis network is within a stable interval, and the second parameter determined by the last iteration of the lightweight regression analysis network is calibrated as the target external parameter of the image acquisition equipment.
Still as shown in fig. 6, the extrinsic refinement adjustment optimization phase includes: based on the complex regression analysis network, matching the point cloud data with the environment image according to the first optimized external parameters and the preset internal parameters, and determining a second optimized external parameters. Then carrying out projection calculation matching point number processing according to the second optimized external parameters, the preset internal parameters, the point cloud data and the environment image, judging whether the number of the matching points is increased compared with the number of the current maximum space points, if the number of the matching points is not increased compared with the number of the current maximum space points, marking the matching points as negative samples, and feeding back the negative samples to the complex regression analysis network so as to enable the complex regression analysis network to adjust network parameters according to the direction of the negative samples, so that the second optimized external parameters output by the complex regression analysis network after the parameters are adjusted can have better effect in matching the point cloud and the image, and adopting the first optimized external parameters as the optimal external parameters to update the preset initial parameters; if the number of the matched points is increased compared with the number of the current maximum space points, the number of the current maximum space points is updated to be the matched points, meanwhile, the matched points are marked as positive samples and fed back to the complex regression analysis network, so that the complex regression analysis network adjusts network parameters according to the direction of the positive samples, a second optimized external parameter output by the complex regression analysis network after the parameters are adjusted can achieve a better effect in matching point cloud and images, and the second optimized external parameter is used as an optimal external parameter to update preset initial parameters; and returning to an initial step of executing the external parameter calculation optimization stage to optimize the complex regression analysis network until the iteration times of the complex regression analysis network reach the preset iteration times, or calibrating the second optimized external parameters as preset initial parameters of the image acquisition equipment when the number of the plurality of matching points determined by the second optimized external parameters output by the complex regression analysis network after a plurality of iterations is in a stable interval.
By adopting the equipment parameter calibration method disclosed by the invention, the regression analysis network can be continuously updated aiming at the positive and negative samples of the regression analysis network in the process of updating the target external parameters, the long-term effectiveness of the whole external parameter optimization algorithm is ensured, and the optimization external parameters of different stages can be adopted aiming at different use scenes, for example, the first optimization external parameters or the second optimization external parameters are selected, so that the method has stronger flexibility.
The embodiment of the disclosure also provides a device parameter calibration method applied to the vehicle-mounted computing platform. The technical solutions of the embodiments of the present disclosure will be described below with reference to the drawings in the embodiments of the present disclosure.
Fig. 7 is a schematic flow chart of an implementation of a device parameter calibration method applied to a vehicle-mounted computing platform according to an embodiment of the disclosure, where, as shown in fig. 7, the method includes:
s701, acquiring an environment image and point cloud data corresponding to a target vehicle.
The environment image and the point cloud data are an image and a space point which are acquired for a target environment respectively.
S702, matching the point cloud data with the environment image based on a preset initial external parameter and a preset internal parameter of the image acquisition device, and determining a first external parameter.
In an embodiment, the on-board computing platform may extract point cloud features of the point cloud data and image features of the environmental image. After extracting the point cloud features and the image features, the vehicle-mounted computing platform matches the point cloud data with the environmental image based on a preset initial external parameter and a preset internal parameter of the image acquisition device, and the step of determining the first external parameter may include: based on a preset initial external reference and a preset internal reference of the image acquisition equipment, projecting the space points in the point cloud data to a pixel plane; and determining the number of first space points which are projected to the space points on the pixel plane and matched with the environment image according to the image characteristics and the point cloud characteristics, and determining the matching parameters corresponding to the number of the first space points as first external parameters.
S703, matching the point cloud data with the environment image based on a first regression analysis network and according to the preset initial external parameters and the preset internal parameters, and determining a second external parameter.
In an embodiment, the step of the vehicle-mounted computing platform matching the point cloud data with the environmental image and determining the second external parameter based on the first regression analysis network and according to the preset initial external parameter and the preset internal parameter may include: inputting the preset initial external parameters, the preset internal parameters, the point cloud features and the image features into a first regression analysis network to obtain the number of second space points matched with the environment image in the space points of the point cloud data, and determining the current parameters of the first regression analysis network as the second external parameters.
S704, determining a first optimized external parameter according to the first external parameter and the second external parameter.
In an embodiment, the step of determining, by the on-board computing platform, the first optimized external parameter according to the first external parameter and the second external parameter may include: comparing the first space point number with the second space point number; if the first number of spatial points is greater than the second number of spatial points, determining the first outlier as a first optimized outlier; and if the first number of space points is not greater than the second number of space points, determining the second external parameters as first optimized external parameters.
In an embodiment, if the vehicle-mounted computing platform determines that the number of the first space points is greater than the number of the second space points, parameters of the first regression analysis network may be adjusted, the first optimized external parameters are adopted to update the preset initial external parameters, and the step of matching the point cloud data with the environmental image based on the preset initial external parameters and the preset internal parameters of the image acquisition device is performed.
And S705, sending the first optimized external parameters to a cloud computing platform, so that the cloud computing platform matches the point cloud data with the environment image based on a second regression analysis network and according to the first optimized external parameters and the preset internal parameters, and after determining a second optimized external parameters, matches the point cloud data with the environment image according to the second optimized external parameters and the preset internal parameters, and determines a third optimized external parameters, and sends the third optimized external parameters to the vehicle-mounted computing platform.
In an embodiment, the step of determining, by the cloud computing platform, the third optimized external parameters includes: determining the number of matching points, in the point cloud data, of the space points and the environment image based on the second optimized external parameters and the preset internal parameters; determining the number of the current maximum space points matched with the environment image in the space points of the point cloud data according to the number of the matching points; if the number of the matching points is larger than the number of the current maximum space points, updating the number of the current maximum space points to the number of the matching points, and determining the second optimized external parameters as third optimized external parameters; and if the number of the matching points is not greater than the current maximum number of the spatial points, determining the first optimized external parameters as third optimized external parameters.
In an embodiment, the step of determining, by the cloud computing platform, the number of matching points that match the spatial points in the point cloud data with the environmental image includes: projecting the space points in the point cloud data to a pixel plane according to the second optimized external parameters and the preset internal parameters; the number of spatial points which are matched with the environment image among the spatial points projected onto the pixel plane is determined as the number of matching points.
S706, receiving the third optimized external parameters sent by the vehicle-mounted computing platform, and calibrating the third optimized external parameters as target external parameters of the image acquisition equipment.
The target external parameters represent attitude information of the image acquisition equipment relative to a radar.
Specifically, the step of determining the target external parameter by the vehicle-mounted computing platform may include: determining whether the iteration number of the second regression analysis network reaches a preset iteration number; if yes, calibrating the third optimized external parameter as a target external parameter of the image acquisition equipment; and if not, adjusting parameters of the second regression analysis network according to the number of the matching points, and after updating the preset initial external parameters by adopting the third optimized external parameters, returning to execute the preset initial external parameters and the preset internal parameters based on the image acquisition equipment, and matching the point cloud data with the environment image.
By adopting the method provided by the embodiment of the disclosure, the external parameters of the image acquisition equipment are not affected by the position of the image acquisition equipment when calibrated, so that the problem of invalid calibration parameters of the image acquisition equipment caused by the position change of the image acquisition equipment can be avoided. And the regression analysis network is adopted to optimize the external parameters of the image acquisition equipment, so that the marked target external parameters are more accurate. And a part of algorithms for optimizing the external parameters can be transferred to the cloud computing equipment, so that the speed of optimizing the external parameters can be increased by utilizing abundant computing resources of the cloud computing equipment, the external parameter optimizing time delay is reduced, and the efficiency of optimizing the external parameters is improved.
Based on the same inventive concept, according to the device parameter calibration method provided by the above embodiment of the present disclosure, correspondingly, another embodiment of the present disclosure further provides a device parameter calibration apparatus, where the device may be applied to a cloud computing platform or a vehicle computing platform, and the structural schematic diagram of the device parameter calibration apparatus is shown in fig. 8, and specifically includes:
a first data obtaining module 801, configured to obtain an environmental image and point cloud data corresponding to a target vehicle, where the environmental image and the point cloud data are an image and a spatial point collected for a target environment respectively;
a first matching module 802, configured to match the point cloud data with the environmental image based on a preset initial external parameter and a preset internal parameter of the image capturing device, and determine a first external parameter;
a second matching module 803, configured to match the point cloud data with the environmental image based on a first regression analysis network and according to the preset initial external parameter and the preset internal parameter, and determine a second external parameter;
the first parameter calibration module 804 is configured to determine a first optimized external parameter according to the first external parameter and the second external parameter, and calibrate the first optimized external parameter as a target external parameter of the image acquisition device, where the target external parameter characterizes pose information of the image acquisition device relative to a radar.
By adopting the device provided by the embodiment of the disclosure, the external parameters of the image acquisition equipment are not affected by the position of the image acquisition equipment when calibrated, so that the problem that the calibration parameters of the image acquisition equipment are invalid due to the position change of the image acquisition equipment can be avoided. And on the basis of optimizing the external parameters of the image acquisition equipment by using the first regression analysis network, the external parameters of the image acquisition equipment are further optimized by using the second regression analysis network, so that the accuracy of the marked target external parameters is further improved.
In an embodiment, the device further comprises:
a second parameter calibration module (not shown in the figure) for matching the point cloud data with the environmental image based on a second regression analysis network and according to the first optimized external parameter and the preset internal parameter, and determining a second optimized external parameter; and matching the point cloud data with the environment image according to the second optimized external parameters and the preset internal parameters, determining a third optimized external parameters, and calibrating the third optimized external parameters as target external parameters of the image acquisition equipment.
In an embodiment, the second parameter calibration module is specifically configured to determine, based on the second optimized external parameter and the preset internal parameter, a number of matching points that match the spatial point in the point cloud data with the environmental image; determining the number of the current maximum space points matched with the environment image in the space points of the point cloud data according to the number of the matching points; if the number of the matching points is larger than the current maximum number of the space points, determining the second optimized external parameters as third optimized external parameters, and updating the current maximum number of the space points to the number of the matching points; if the number of the matching points is not greater than the current maximum number of the spatial points, determining the first optimized external parameters as third optimized external parameters; determining whether the iteration number of the second regression analysis network reaches a preset iteration number; if yes, the third optimized external parameter is calibrated as a target external parameter of the image acquisition equipment, and the iteration is terminated; and if not, adjusting parameters of the second regression analysis network according to the number of the matching points, and after updating the preset initial external parameters by adopting the third optimized external parameters, returning to execute the preset initial external parameters and the preset internal parameters based on the image acquisition equipment, and matching the point cloud data with the environment image.
In an embodiment, the second parameter calibration module is specifically configured to project spatial points in the point cloud data to a pixel plane according to the second optimized external parameter and the preset internal parameter; and determining the number of spatial points matched with the environment image in the spatial points projected onto the pixel plane as the number of matched points.
In an embodiment, the device further comprises:
a feature extraction module (not shown in the figure) for extracting point cloud features of the point cloud data and image features of the environmental image;
the first matching module 802 is specifically configured to project spatial points in the point cloud data to a pixel plane based on a preset initial external parameter and a preset internal parameter of an image acquisition device; and determining the number of first space points which are projected to the space points on the pixel plane and matched with the environment image according to the image characteristics and the point cloud characteristics, and determining the matching parameters corresponding to the number of the first space points as first external parameters.
In an embodiment, the second matching module 803 is specifically configured to input the preset initial external parameter, the preset internal parameter, the point cloud feature, and the image feature into a first regression analysis network, obtain a second number of spatial points matched with the environmental image in spatial points of the point cloud data, and determine a current parameter of the first regression analysis network as the second external parameter.
In an embodiment, the first parameter calibration module 804 is specifically configured to compare the first number of spatial points with the second number of spatial points; if the first number of spatial points is greater than the second number of spatial points, determining the first outlier as a first optimized outlier; and if the first number of space points is not greater than the second number of space points, determining the second external parameters as first optimized external parameters.
In an embodiment, the first parameter calibration module 804 is further configured to adjust parameters of the first regression analysis network if the number of the first spatial points is greater than the number of the second spatial points, update the preset initial external parameters with the first optimized external parameters, and return to execute the step of matching the point cloud data with the environmental image based on the preset initial external parameters and the preset internal parameters of the image capturing device.
Based on the same inventive concept, according to the device parameter calibration method applied to the vehicle-mounted computing platform provided in the above embodiment of the present disclosure, correspondingly, another embodiment of the present disclosure further provides a device parameter calibration apparatus applied to the vehicle-mounted computing platform, where a structural schematic diagram of the device parameter calibration apparatus is shown in fig. 9, and the device parameter calibration apparatus specifically includes:
The second data acquisition module 901 is configured to acquire an environmental image and point cloud data corresponding to a target vehicle, where the environmental image and the point cloud data are an image and a spatial point acquired for the target environment respectively;
a third matching module 902, configured to match the point cloud data with the environmental image based on a preset initial external parameter and a preset internal parameter of the image capturing device, and determine a first external parameter;
a fourth matching module 903, configured to match the point cloud data with the environmental image based on a first regression analysis network and according to the preset initial external parameter and the preset internal parameter, and determine a second external parameter;
a third parameter calibration module 904, configured to determine a first optimized external parameter according to the first external parameter and the second external parameter; the first optimized external parameters are sent to a cloud computing platform, so that the cloud computing platform matches the point cloud data with the environment image based on a second regression analysis network according to the first optimized external parameters and the preset internal parameters, after a second optimized external parameters are determined, matches the point cloud data with the environment image according to the second optimized external parameters and the preset internal parameters, determines a third optimized external parameters, and sends the third optimized external parameters to the vehicle-mounted computing platform; receiving the third optimized external parameters sent by the vehicle-mounted computing platform; and calibrating the third optimized external parameter as a target external parameter of the image acquisition equipment, wherein the target external parameter represents attitude information of the image acquisition equipment relative to a radar.
By adopting the device provided by the embodiment of the disclosure, the external parameters of the image acquisition equipment are not affected by the position of the image acquisition equipment when calibrated, so that the problem that the calibration parameters of the image acquisition equipment are invalid due to the position change of the image acquisition equipment can be avoided. And the regression analysis network is adopted to optimize the external parameters of the image acquisition equipment, so that the marked target external parameters are more accurate. And the partial algorithm for optimizing the external parameters can be transferred to the cloud computing equipment, so that the speed of optimizing the external parameters can be increased by utilizing abundant computing resources of the cloud computing equipment, and the efficiency of optimizing the external parameters is improved.
According to embodiments of the present disclosure, the present disclosure also provides an electronic device and a readable storage medium.
Fig. 10 shows a schematic block diagram of an example electronic device 1000 that may be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the disclosure described and/or claimed herein.
As shown in fig. 10, the electronic device 1000 includes a computing unit 1001 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 1002 or a computer program loaded from a storage unit 1008 into a Random Access Memory (RAM) 1003. In the RAM 1003, various programs and data required for the operation of the electronic apparatus 1000 can also be stored. The computing unit 1001, the ROM 1002, and the RAM 1003 are connected to each other by a bus 1004. An input/output (I/O) interface 1005 is also connected to bus 1004.
Various components in the electronic device 1000 are connected to the I/O interface 1005, including: an input unit 1006 such as a keyboard, a mouse, and the like; an output unit 1007 such as various types of displays, speakers, and the like; a storage unit 1008 such as a magnetic disk, an optical disk, or the like; and communication unit 1009 such as a network card, modem, wireless communication transceiver, etc. Communication unit 1009 allows electronic device 1000 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunications networks.
The computing unit 1001 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 1001 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1001 performs the respective methods and processes described above, such as the device parameter calibration method. For example, in some embodiments, the device parameter calibration method may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as the storage unit 1008. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 1000 via the ROM 1002 and/or the communication unit 1009. When the computer program is loaded into RAM 1003 and executed by computing unit 1001, one or more steps of the device parameter calibration method described above may be performed. Alternatively, in other embodiments, the computing unit 1001 may be configured to perform the device parameter calibration method in any other suitable way (e.g., by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuitry, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems-on-a-chip (SOCs), complex Programmable Logic Devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), and the internet.
The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server may be a cloud server, a server of a distributed system, or a server incorporating a blockchain.
It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present disclosure may be performed in parallel or sequentially or in a different order, provided that the desired results of the technical solutions of the present disclosure are achieved, and are not limited herein.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present disclosure, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.
The foregoing is merely specific embodiments of the disclosure, but the protection scope of the disclosure is not limited thereto, and any person skilled in the art can easily think about changes or substitutions within the technical scope of the disclosure, and it is intended to cover the scope of the disclosure. Therefore, the protection scope of the present disclosure shall be subject to the protection scope of the claims.
Claims (10)
1. A method for calibrating parameters of a device, the method comprising:
acquiring an environment image and point cloud data corresponding to a target vehicle, wherein the environment image and the point cloud data are images and space points acquired for the target environment respectively;
based on a preset initial external parameter and a preset internal parameter of the image acquisition equipment, matching the point cloud data with the environment image, and determining a first external parameter;
based on a first regression analysis network, matching the point cloud data with the environment image according to the preset initial external parameters and the preset internal parameters, and determining a second external parameter;
determining a first optimized external parameter according to the first external parameter and the second external parameter, and calibrating the first optimized external parameter as a target external parameter of the image acquisition equipment, wherein the target external parameter represents the attitude information of the image acquisition equipment relative to a radar.
2. The method according to claim 1, wherein the method further comprises:
based on a second regression analysis network, matching the point cloud data with the environment image according to the first optimized external parameters and the preset internal parameters, and determining a second optimized external parameters;
And matching the point cloud data with the environment image according to the second optimized external parameters and the preset internal parameters, determining a third optimized external parameters, and calibrating the third optimized external parameters as target external parameters of the image acquisition equipment.
3. The method according to claim 2, wherein the matching the point cloud data and the environmental image according to the second optimized external parameters and the preset internal parameters, and determining a third optimized external parameters, and calibrating the third optimized external parameters as target external parameters of the image acquisition device, includes:
determining the number of matching points, in the point cloud data, of the space points and the environment image based on the second optimized external parameters and the preset internal parameters;
determining the number of the current maximum space points matched with the environment image in the space points of the point cloud data according to the number of the matching points;
if the number of the matching points is larger than the current maximum number of the space points, determining the second optimized external parameters as third optimized external parameters, and updating the current maximum number of the space points to the number of the matching points;
if the number of the matching points is not greater than the current maximum number of the spatial points, determining the first optimized external parameters as third optimized external parameters;
Determining whether the iteration number of the second regression analysis network reaches a preset iteration number;
if yes, the third optimized external parameter is calibrated as a target external parameter of the image acquisition equipment, and the iteration is terminated;
and if not, adjusting parameters of the second regression analysis network according to the number of the matching points, and after updating the preset initial external parameters by adopting the third optimized external parameters, returning to execute the preset initial external parameters and the preset internal parameters based on the image acquisition equipment, and matching the point cloud data with the environment image.
4. The method of claim 3, wherein the determining the number of matching points for which the spatial points in the point cloud data match the environmental image based on the second optimized external parameter and the preset internal parameter comprises:
projecting the spatial points in the point cloud data to a pixel plane according to the second optimized external parameters and the preset internal parameters;
and determining the number of spatial points matched with the environment image in the spatial points projected onto the pixel plane as the number of matched points.
5. The method of claim 1, wherein prior to matching the point cloud data and the environmental image and determining a first external reference based on a preset initial external reference and a preset internal reference of the image acquisition device, the method further comprises:
Extracting point cloud characteristics of the point cloud data and image characteristics of the environment image;
the matching the point cloud data and the environmental image based on the preset initial external parameters and the preset internal parameters of the image acquisition equipment, and determining a first external parameter comprises the following steps:
based on a preset initial external reference and a preset internal reference of the image acquisition equipment, projecting the space points in the point cloud data to a pixel plane;
and determining the number of first space points which are projected to the space points on the pixel plane and matched with the environment image according to the image characteristics and the point cloud characteristics, and determining the matching parameters corresponding to the number of the first space points as first external parameters.
6. The method of claim 5, wherein the matching the point cloud data and the environmental image based on the first regression analysis network and according to the preset initial external parameters and the preset internal parameters, and determining a second external parameter, comprises:
inputting the preset initial external parameters, the preset internal parameters, the point cloud features and the image features into a first regression analysis network to obtain the number of second space points matched with the environment image in the space points of the point cloud data, and determining the current parameters of the first regression analysis network as the second external parameters.
7. The method of claim 6, wherein the determining a first optimized external parameter from the first external parameter and the second external parameter comprises:
comparing the first space point number with the second space point number;
if the first number of spatial points is greater than the second number of spatial points, determining the first outlier as a first optimized outlier;
and if the first number of space points is not greater than the second number of space points, determining the second external parameters as first optimized external parameters.
8. The method of claim 7, wherein the method further comprises:
and if the number of the first space points is larger than the number of the second space points, adjusting parameters of the first regression analysis network, adopting the first optimized external parameters to update the preset initial external parameters, and returning to execute the preset initial external parameters and the preset internal parameters based on the image acquisition equipment to match the point cloud data with the environment image.
9. A method for calibrating equipment parameters, which is applied to a vehicle-mounted computing platform, the method comprising:
acquiring an environment image and point cloud data corresponding to a target vehicle, wherein the environment image and the point cloud data are images and space points acquired for the target environment respectively;
Based on a preset initial external parameter and a preset internal parameter of the image acquisition equipment, matching the point cloud data with the environment image, and determining a first external parameter;
based on a first regression analysis network, matching the point cloud data with the environment image according to the preset initial external parameters and the preset internal parameters, and determining a second external parameter;
determining a first optimized external parameter according to the first external parameter and the second external parameter;
the first optimized external parameters are sent to a cloud computing platform, so that the cloud computing platform matches the point cloud data with the environment image based on a second regression analysis network according to the first optimized external parameters and the preset internal parameters, after a second optimized external parameters are determined, matches the point cloud data with the environment image according to the second optimized external parameters and the preset internal parameters, determines a third optimized external parameters, and sends the third optimized external parameters to the vehicle-mounted computing platform;
and receiving the third optimized external parameters sent by the vehicle-mounted computing platform, and calibrating the third optimized external parameters as target external parameters of the image acquisition equipment, wherein the target external parameters represent the attitude information of the image acquisition equipment relative to a radar.
10. A device parameter calibration apparatus, the apparatus comprising:
the first data acquisition module is used for acquiring an environment image and point cloud data corresponding to a target vehicle, wherein the environment image and the point cloud data are images and space points acquired for the target environment respectively;
the first matching module is used for matching the point cloud data with the environment image based on a preset initial external parameter and a preset internal parameter of the image acquisition equipment and determining a first external parameter;
the second matching module is used for matching the point cloud data with the environment image based on the first regression analysis network and according to the preset initial external parameters and the preset internal parameters, and determining a second external parameter;
the first parameter calibration module is used for determining a first optimized external parameter according to the first external parameter and the second external parameter, and calibrating the first optimized external parameter as a target external parameter of the image acquisition equipment, wherein the target external parameter represents the attitude information of the image acquisition equipment relative to a radar.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310449374.0A CN116168090B (en) | 2023-04-24 | 2023-04-24 | Equipment parameter calibration method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310449374.0A CN116168090B (en) | 2023-04-24 | 2023-04-24 | Equipment parameter calibration method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116168090A true CN116168090A (en) | 2023-05-26 |
CN116168090B CN116168090B (en) | 2023-08-22 |
Family
ID=86413572
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310449374.0A Active CN116168090B (en) | 2023-04-24 | 2023-04-24 | Equipment parameter calibration method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116168090B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117557660A (en) * | 2024-01-09 | 2024-02-13 | 北京集度科技有限公司 | Data processing method and device, electronic equipment and vehicle |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109949371A (en) * | 2019-03-18 | 2019-06-28 | 北京智行者科技有限公司 | A kind of scaling method for laser radar and camera data |
CN110782497A (en) * | 2019-09-06 | 2020-02-11 | 腾讯科技(深圳)有限公司 | Method and device for calibrating external parameters of camera |
CN111308448A (en) * | 2018-12-10 | 2020-06-19 | 杭州海康威视数字技术股份有限公司 | Image acquisition equipment and radar external parameter determination method and device |
-
2023
- 2023-04-24 CN CN202310449374.0A patent/CN116168090B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111308448A (en) * | 2018-12-10 | 2020-06-19 | 杭州海康威视数字技术股份有限公司 | Image acquisition equipment and radar external parameter determination method and device |
CN109949371A (en) * | 2019-03-18 | 2019-06-28 | 北京智行者科技有限公司 | A kind of scaling method for laser radar and camera data |
CN110782497A (en) * | 2019-09-06 | 2020-02-11 | 腾讯科技(深圳)有限公司 | Method and device for calibrating external parameters of camera |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117557660A (en) * | 2024-01-09 | 2024-02-13 | 北京集度科技有限公司 | Data processing method and device, electronic equipment and vehicle |
CN117557660B (en) * | 2024-01-09 | 2024-04-12 | 北京集度科技有限公司 | Data processing method and device, electronic equipment and vehicle |
Also Published As
Publication number | Publication date |
---|---|
CN116168090B (en) | 2023-08-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10852139B2 (en) | Positioning method, positioning device, and robot | |
CN107481292B (en) | Attitude error estimation method and device for vehicle-mounted camera | |
CN113674421B (en) | 3D target detection method, model training method, related device and electronic equipment | |
CN112560680A (en) | Lane line processing method and device, electronic device and storage medium | |
CN116168090B (en) | Equipment parameter calibration method and device | |
CN114036253B (en) | High-precision map data processing method, device, electronic equipment and medium | |
EP4124878A2 (en) | Method and apparatus for calibrating lidar and positioning device and storage medium | |
CN113436233A (en) | Registration method and device of automatic driving vehicle, electronic equipment and vehicle | |
CN115690739B (en) | Multi-sensor fusion obstacle presence detection method and automatic driving vehicle | |
CN114312843B (en) | Method and device for determining information | |
CN113177980B (en) | Target object speed determining method and device for automatic driving and electronic equipment | |
CN115656989A (en) | External parameter calibration method and device, electronic equipment and storage medium | |
CN113127489B (en) | Data table updating method and device | |
CN109691185B (en) | Positioning method, positioning device, terminal and readable storage medium | |
CN113554712A (en) | Registration method and device of automatic driving vehicle, electronic equipment and vehicle | |
CN113219505A (en) | Method, device and equipment for acquiring GPS coordinates for vehicle-road cooperative tunnel scene | |
CN115235527B (en) | Sensor external parameter calibration method and device and electronic equipment | |
CN114419564B (en) | Vehicle pose detection method, device, equipment, medium and automatic driving vehicle | |
CN115147561A (en) | Pose graph generation method, high-precision map generation method and device | |
CN115575931A (en) | Calibration method, calibration device, electronic equipment and storage medium | |
CN111929585B (en) | Battery charge state calculating device, method, server and medium | |
CN110244710B (en) | Automatic tracing method, device, storage medium and electronic equipment | |
CN115096328B (en) | Positioning method and device of vehicle, electronic equipment and storage medium | |
CN114399555B (en) | Data online calibration method and device, electronic equipment and computer readable medium | |
CN114219907B (en) | Three-dimensional map generation method, device, equipment 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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |