WO2023130263A1 - 路面标志识别方法及装置、设备、存储介质、车辆 - Google Patents

路面标志识别方法及装置、设备、存储介质、车辆 Download PDF

Info

Publication number
WO2023130263A1
WO2023130263A1 PCT/CN2022/070325 CN2022070325W WO2023130263A1 WO 2023130263 A1 WO2023130263 A1 WO 2023130263A1 CN 2022070325 W CN2022070325 W CN 2022070325W WO 2023130263 A1 WO2023130263 A1 WO 2023130263A1
Authority
WO
WIPO (PCT)
Prior art keywords
road
sign
redundant
block
information
Prior art date
Application number
PCT/CN2022/070325
Other languages
English (en)
French (fr)
Inventor
王炳蘅
Original Assignee
华为技术有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Priority to PCT/CN2022/070325 priority Critical patent/WO2023130263A1/zh
Priority to CN202280003624.1A priority patent/CN116745820A/zh
Publication of WO2023130263A1 publication Critical patent/WO2023130263A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads

Definitions

  • the present application relates to the field of smart cars, in particular to a road sign recognition method, device, device, storage medium, and vehicle.
  • auxiliary driving functions such as automatic driving and intelligent driving have become necessary functions for vehicles, and whether these functions can be realized depends on the vehicle's perception of road features such as obstacles and road signs.
  • the embodiments of the present application provide a road sign recognition method, device, device, storage medium, and vehicle, which can efficiently and accurately recognize road signs.
  • the first aspect of the present application provides a pavement marking recognition method, including:
  • an identification result of the redundant sign block is obtained, and the identification result includes the first road sign information.
  • redundant road signs on the road surface can be efficiently and accurately identified.
  • the determination of redundant sign blocks in the road image includes: according to pre-configured road sign rules, road sign images and/or color differences of sign blocks in the road image, determining the number of redundant sign blocks in the road image Redundant marker block.
  • determining the redundant sign block in the road image includes: when the sign block of the road image does not conform to the pre-configured road sign rule, the sign block is a redundant sign block; and/ Or, when the shape of the sign block of the road image does not match the shape of the road surface sign image, the sign block is a redundant sign block; and/or, when the color difference of the sign block of the road image is greater than a first preset threshold, the sign block is redundant remaining sign blocks.
  • obtaining the identification result of the redundant sign block according to the reference data and/or the road sign image includes: determining the first road sign information of the redundant sign block according to the reference data; wherein, The reference data includes one or more of the following: high-precision map data, traffic sign information, and navigation map data.
  • the identification result of the redundant marker block can be determined efficiently and accurately by means of one or more types of reference data.
  • obtaining the recognition result of the redundant sign block according to the reference data and/or the road sign image includes: obtaining the redundant sign block by matching the redundant sign block and the road sign image Two or more types of second road sign information, the road sign image corresponding to each second road sign information is included in the image of the redundant sign block; according to the two or more second road sign information of the redundant sign block The sign information is used to determine the first road surface sign information of the redundant sign block.
  • the road sign information of the redundant sign block can be determined efficiently and accurately by means of image matching or the like.
  • the method further includes: generating a safe driving policy according to the second road surface sign information corresponding to the safety road sign. This ensures maximum compliance with traffic regulations and ensures safe driving.
  • determining the first road marking information of the redundant marking block according to two or more types of second road marking information of the redundant marking block includes: Two or more types of second road sign information, using a machine learning algorithm to estimate the reliability of the third road sign information and the third road sign information of the redundant sign block; the third road sign information with the highest reliability is used as the First pavement sign information.
  • the road marking information of the redundant marking blocks can be determined accurately and efficiently by using the machine learning algorithm and reliability.
  • a machine learning algorithm is used to infer the third road sign information and the third road sign information of the redundant sign blocks
  • the reliability of the information includes: calculating the color difference corresponding to the second road sign information by using the image data corresponding to the second road sign information in the redundant sign block;
  • the class center clusters the image data of the redundant sign blocks to obtain two or more third road sign information and image data sets of the third road sign information of the redundant sign blocks; according to the third road sign information An image dataset to determine the credibility of a third pavement marking information.
  • the road marking information of the redundant marking blocks can be accurately and efficiently determined through the combination of color difference and clustering.
  • a machine learning algorithm is used to infer the third road sign information and the third road sign information of the redundant sign blocks.
  • the credibility of the information also includes: before clustering the image data of the redundant sign blocks, removing the second road sign information whose color difference is greater than the second predetermined threshold.
  • determining the first road sign information of the redundant sign block also includes: determining the third road sign information Before the credibility of the information, remove the third pavement marking information that does not meet the pavement marking rules.
  • the first aspect of the present application provides a pavement marking recognition device, including:
  • an acquisition unit configured to acquire a road image, where the road image includes a road surface
  • a redundancy determination unit configured to determine redundant marker blocks of the road image
  • the sign recognition unit is configured to obtain a recognition result of the redundant sign block according to the reference data and/or the road sign image, and the recognition result includes the first road sign information.
  • the redundancy determination unit is specifically configured to: determine the redundant sign of the road image according to the pre-configured road sign rule, the road sign image and/or the color difference of the sign block in the road image piece.
  • the redundancy determining unit is specifically configured to: determine that the sign block is a redundant sign block when the sign block of the road image does not conform to the pre-configured road sign rule; and/or, When the shape of the sign block of the road image does not match the shape of the road surface sign image, it is determined that the sign block is a redundant sign block; and/or, when the color difference of the sign block of the road image is greater than a first preset threshold, it is determined that the sign block is redundant. remaining sign blocks.
  • the sign recognition unit is specifically configured to: determine the first road sign information of the redundant sign block according to the reference data; wherein the reference data includes one or more of the following: high-precision map Data, traffic sign information, navigation map data.
  • the sign recognition unit is specifically configured to: obtain two or more second road sign information of redundant sign blocks by matching redundant sign blocks with road sign images, each The road sign images corresponding to the second road sign information are all included in the image of the redundant sign block; according to two or more than two kinds of second road sign information of the redundant sign block, the first road sign of the redundant sign block is determined information.
  • the road sign recognition device further includes: a safety processing unit, configured to generate a safe driving strategy according to the second road sign information corresponding to the safety road sign.
  • the sign recognition unit is specifically configured to: use a machine learning algorithm to infer the third road sign information of the redundant sign block according to two or more than two kinds of second road sign information of the redundant sign block.
  • the reliability of the road sign information and the third road sign information; the third road sign information with the highest reliability is used as the first road sign information.
  • the sign recognition unit is specifically configured to: use the image data corresponding to the second road sign information in the redundant sign block to calculate the color difference corresponding to the second road sign information; At least two of the second road sign information are the initial cluster centers, and the image data of the redundant sign blocks are clustered to obtain two or more third road sign information and the third road sign information of the redundant sign blocks The image data set; according to the image data set of the third road sign information, determine the credibility of the third road sign information.
  • the sign recognition unit is further configured to: before clustering the image data of redundant sign blocks, remove second road sign information whose color difference is greater than a second predetermined threshold.
  • the sign recognition unit is further configured to: before determining the credibility of the third road sign information, remove third road sign information that does not comply with road sign rules.
  • the third aspect of the present application provides a computing device, including a processor and a memory, the memory stores program instructions, and when the program instructions are executed by the processor, the processor executes the road sign recognition method of the first aspect.
  • a fourth aspect of the present application provides a computer-readable storage medium on which program instructions are stored, wherein the program instructions cause the computer to execute the road sign recognition method of the first aspect when executed by a computer.
  • the fifth aspect of the present application provides a vehicle, including the computing device of the third aspect or the road sign recognition device of the second aspect.
  • a sixth aspect of the present application provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the processor executes the road sign recognition method of the first aspect.
  • the embodiment of the present application first finds the redundant sign blocks of the road image, and then performs targeted processing on the redundant sign blocks according to the reference data and/or the road sign image, which can efficiently and accurately identify the redundant road signs, and solve the problems caused by road signs. Problems such as the difficulty in identifying redundant road signs caused by clutter can effectively improve the accuracy of road feature information and the reliability of vehicle assisted driving functions.
  • FIG. 1 is an example diagram of an application scenario according to Embodiment 1 of the present application.
  • FIG. 2 is an example diagram of another application scenario according to the embodiment of the present application.
  • FIG. 3 is a schematic flow diagram of a method for identifying road signs provided in an embodiment of the present application
  • FIG. 4 is a schematic diagram of a specific implementation flow chart of a road sign recognition method according to an embodiment of the present application
  • FIG. 5 is a schematic diagram of an image of a redundant flag block in an embodiment of the present application.
  • FIG. 6 is a schematic diagram of an image of another redundant flag block according to the embodiment of the present application.
  • FIG. 7 is a schematic structural diagram of a road sign recognition device provided in an embodiment of the present application.
  • Fig. 8 is a schematic diagram of the deployment of the road sign recognition device provided by the embodiment of the present application.
  • FIG. 9 is a schematic structural diagram of a computing device provided by an embodiment of the present application.
  • Road signs traffic signs painted on the road surface, including but not limited to lane lines, guiding arrows, sidewalks, stop lines, mesh lines, diversion lines, speed limit signs, vehicle type limit signs, etc.
  • Safety road signs represented by sidewalks, stop lines, and speed limit signs, road signs that are largely related to driving safety.
  • Traffic regulations and road signs represented by lane lines, guiding arrows, and diversion lines, they have relatively little impact on driving safety, but are largely related to the implementation of traffic rules.
  • the embodiment of the present application provides an improved road sign recognition method, device, equipment, and storage medium, first find out the redundant sign blocks of the road image, and then identify the redundant sign blocks according to the reference data and/or the road sign image Targeted processing can efficiently and accurately identify road signs, solve problems such as the difficulty in identifying redundant road signs caused by mixed road signs, ensure the accuracy of road feature information, and improve the reliability of vehicle assisted driving functions.
  • the embodiments of the present application are applicable to various scenarios, and are especially applicable to scenarios such as cities and suburban roads with chaotic road signs.
  • the embodiments of the present application can be applied to various driving modes, especially to assisted driving modes such as automatic driving and intelligent driving.
  • the “vehicle” in the embodiment of the present application may be of various types.
  • the “vehicle” in this application may be, but not limited to, private cars, commercial vehicles, buses, passenger vehicles, high-speed rail, subways, unmanned vehicles, unmanned aerial vehicles, logistics transport vehicles, unmanned transport vehicles, ships, aircraft, etc.
  • the power type of the "vehicle” can be fuel-driven, pure electric, hydrogen fuel cell-driven, hybrid, etc.
  • Fig. 1 shows an example diagram of a road scene to which the embodiment of the present application is applicable.
  • the leftmost lane of the road should be a straight sign, but the road sign of the leftmost lane is straight ahead and turns left. It can be seen that the road sign of the leftmost lane is painted incorrectly.
  • a road sign may be recognized as going straight, going straight and turning left, turning left, etc., that is, the road sign is a redundant road sign.
  • Fig. 2 shows an example diagram of another road scene to which the embodiment of the present application is applicable.
  • the pedestrian crossing sign that is, the zebra crossing in Figure 2
  • Fig. 3 shows a flow chart of a method for recognizing road signs provided by an embodiment of the present application.
  • the pavement marking recognition method provided by the embodiment of the present application may include the following steps:
  • Step S310 acquiring a road image, which includes a road surface
  • the road image may be, but not limited to, a road image including road signs collected by a vehicle-mounted sensor such as a camera; Road feature images obtained by preprocessing such as road feature extraction, etc.
  • Step S320 determining redundant marker blocks of the road image
  • the redundant sign block can be any sign block that cannot uniquely determine the road sign information, and the sign block can be a region of interest (ROI) of the road image, and the road image ROI contains continuous sign lines.
  • ROI region of interest
  • the redundant sign block may be identified as two or more types of road signs, that is, the road sign information of the redundant sign block may contain two or more types.
  • redundant flag blocks may be fully redundant. For example, due to road maintenance or re-planning, some pavement markings on the road are invalid, but these pavement markings may not be erased in time or may still retain marks due to improper erasure.
  • the flag block is a completely redundant flag block.
  • redundant sign blocks of the road image may be determined according to preconfigured road sign rules, road sign images and/or color differences of sign blocks in the road image.
  • determining the redundant sign blocks of the road image may include any one or more of the following:
  • the sign block is a redundant sign block
  • the internal color difference degree of a single sign block that is, the color difference Diff
  • the color difference Diff can be calculated by the following formula (1):
  • Xmax is the X value of the pixel point with the largest corresponding color component in the marker block
  • Xmin is the X value of the pixel point with the smallest corresponding color component in the marker block
  • X R ⁇ G ⁇ B (red ⁇ green ⁇ blue).
  • the sign block is a redundant sign block
  • the shape similarity between the sign block and each road sign image can be obtained by matching the sign block and the pavement sign image. If the shape similarity is higher than the preset similarity threshold, it can be considered that the sign block matches the road sign image. Therefore, the identification result of the sign block can be obtained, and the identification result can include the road sign information corresponding to all the pavement sign images matching the sign block. If the road sign image matching the sign block is unique, the sign block is not a redundant sign block , if there are two or more types of pavement sign images matching the sign block, the sign block is a redundant sign block.
  • the sign block of the road image does not meet the pre-configured road sign rules, the sign block is a redundant sign block;
  • the recognition result of all the sign blocks in the road image conform to the road sign rules. If the recognition result of a certain sign block does not meet any of the road sign rules, the sign block is a redundant sign block. If a sign block The recognition result of conforms to all pavement marking rules, the sign block is not a redundant sign block, and the recognition result is taken as the final recognition result of the sign block.
  • road sign rules can be set by pre-configuring a blacklist.
  • a blacklist can be pre-configured, and the blacklist can include all road marking rules.
  • the redundant marker blocks can be determined by traversing the blacklist.
  • the above-mentioned item 1), item 2) and item 3) can be executed sequentially according to a predetermined execution order, so as to efficiently and comprehensively filter out the redundancy of the road image flag block.
  • road marking rules can be adjusted according to scene requirements, regional characteristics, traffic rules, etc.
  • Pavement marking rules can be preconfigured. For example, manual configuration, automatic configuration by parsing traffic rule-related data, etc. may be used.
  • the pavement marking rules may include, but are not limited to, one or more of the following: 1) There should be no stop lines within the diversion lines and mesh lines; 2) The distance between guiding arrows and lane lines should be at least 2 meters; 3) There should be no contradictory guiding arrows on the same lane line.
  • Step S330 according to the reference data and/or the road sign image, the identification result of the redundant sign block is obtained, and the identification result includes the first road sign information.
  • the reference data may include one or more of the following: high-precision map data, traffic sign information, and navigation map data.
  • the reference data can be stored in the memory of the vehicle or from the cloud, and the reference data can be updated regularly.
  • step S330 may include: determining the first road sign information of the redundant sign block according to the reference data.
  • determining the road marking information of the redundant marking block according to the reference data may specifically include: determining the road marking information of the redundant marking block according to the first reference data; if the road marking information of the redundant marking block cannot be uniquely determined , continue to determine the road marking information of the redundant marking block according to the second reference data; if the road marking information of the redundant marking block cannot be uniquely determined, continue to determine the road marking information of the redundant marking block according to the third reference data. This is performed until all the reference data are tried or the pavement sign information of the redundant sign block is determined.
  • the first reference data, the second reference data, and the third reference data can be sorted according to priority or other predetermined rules, and executed in sequence according to the sorting of the reference data until the road surface of the redundant marker block can be uniquely determined Flag information or all reference data have been tried.
  • high-precision map data can be used as the highest priority reference data, that is, the first reference data; road traffic sign data can be used as the second highest priority reference data, that is, the second reference data; navigation map data can be used as the priority The lowest-level reference data, that is, the third reference data.
  • the high-precision map data can be used to determine the road sign information of the redundant sign block first, then the road traffic sign data can be used to determine the road sign information of the redundant sign block, and finally the navigation map data can be used to determine the road surface of the redundant sign block Logo information.
  • the sign block 3 should be going straight, not going straight and turning left, and not turning left. Therefore, it can be determined that the first road surface sign information of the sign block 3 is going straight.
  • step S330 may further include: step a1, by matching the redundant sign block with the road sign image, to obtain two or more second road sign information of the redundant sign block, each of the second road sign information The road sign images corresponding to the two road sign information are all included in the image of the redundant sign block; step a2, according to two or more than two kinds of second road sign information of the redundant sign block, determine the first road surface of the redundant sign block Logo information.
  • the road sign recognition method may further include: generating a safe driving policy according to the second road sign information corresponding to a safe road sign.
  • step a2 may include: according to two or more types of second road sign information of the redundant sign block, using a machine learning algorithm to infer the third road sign information of the redundant sign block and the third road sign information Reliability: the third road sign information with the highest reliability is used as the first road sign information.
  • the image data corresponding to the second road sign information in the redundant sign block can be used to calculate the color difference corresponding to the second road sign information; at least two second road sign information with larger color differences are used as initial clustering centers, and the Image data clustering of redundant sign blocks to obtain two or more third road sign information and image data sets of third road sign information of redundant sign blocks; according to the image data set of third road sign information, The credibility of the third pavement marking information is determined.
  • the reliability can be calculated according to the following formula (2):
  • classification refers to the number of categories obtained by clustering, that is, the number of third road sign information obtained by clustering
  • mean R is the mean value of R values of each pixel data in the image data set
  • mean G is the mean value of each pixel data in the image data set
  • B mean value is the mean value of the B value of each pixel data in the image data set.
  • the second road sign information whose color difference is greater than a second predetermined threshold may be removed first.
  • the third road sign information that does not meet the road sign rules may be removed.
  • the specific implementation process of road sign recognition in the embodiment of the present application may include the following steps:
  • Step S401 acquiring a road image, which includes road signs
  • Step S402 extracting sign blocks in the road image
  • the location information of the marker block can be acquired.
  • the position information of the marker block may include but not limited to the position information of all pixels in the image of the marker block, and the position information of predetermined pixels in the image of the marker block (such as pixels on the edge of the image, the center of the image, etc.).
  • Step S403 determine the first road surface marking information of redundant sign blocks and non-redundant sign blocks of the road image, if the road image contains redundant sign blocks, proceed to step S404, if the road image does not contain redundant sign blocks, it can be Jump directly to step S408;
  • each sign block is matched with a known road sign image to obtain an initial recognition result of each sign block, which includes the road sign information corresponding to the road sign image matched with the sign block, which will be described below
  • the road marking information is named initial road marking information.
  • the sign block can uniquely match a certain road sign image, that is, it can be determined that the road sign information of the sign block is the road sign indicated by the road sign image, the sign block is not a redundant sign block, but a non-redundant sign block. If the sign block cannot uniquely match all the pavement sign images, it means that the sign block is a redundant sign block.
  • the initial road marker information of the marker block can be used as the first road marker information of the marker block. If the sign block is a redundant sign block, it is necessary to proceed to the subsequent steps until the first road surface sign information of the redundant sign block is obtained.
  • sign block 1 represents the guiding arrow in the right lane
  • sign block 2 represents the guiding arrow in the middle lane
  • sign block 3 represents the guiding arrow in the left lane:
  • Sign block 1 represents the zebra crossing in Figure 2, including new and old sign lines:
  • the color difference of each marker block can be calculated separately. If the color difference of the marker block is greater than the first predetermined threshold, it indicates that the internal color difference of the marker block is relatively large, and may contain old and new marker lines.
  • the marker block is a redundant marker block , if the color difference of the marker block is less than or equal to the first predetermined threshold, it means that the inner color of the marker line of the marker block is uniform, and the marker block is not a redundant marker block.
  • the sign block it can be judged whether the sign block is a redundant sign block according to the initial recognition result of the sign block and the pre-configured blacklist. If the sign block conforms to all road sign rules in the blacklist, the sign block is not a redundant sign block , if the sign block does not comply with any road sign rule in the blacklist, the sign block is a redundant sign block.
  • the blacklist please refer to the relevant description above, and will not repeat it here.
  • Step S404 determine the first road marker information of the redundant marker block according to the reference data, if all the reference data have been tried but the road marker information of the redundant marker block is not uniquely determined, then proceed to the subsequent step S405, if all the redundant marker blocks The road surface marking information of the remaining marking blocks has been uniquely determined, and the process can directly jump to step S408.
  • the first road sign information determined according to the reference data may be the same as or different from the initial road sign information of the redundant sign block.
  • Step S405 matching the sign block with the road sign image, and determining the road sign information combination of the redundant sign block, the road sign information combination may include two or more second road sign information;
  • the matching rule can be: match the image of the sign block with each road sign image, determine which road sign images are included in the sign block image, and the road sign information corresponding to these road sign images is the second road surface of the sign block
  • the second road sign information of different sign blocks is combined by using the multiplication principle to obtain the road sign information combination of the redundant sign blocks in the road image.
  • the image 51 of the sign block contains two kinds of road sign images of going straight 53 and turning left 52, and the shaded part is the overlapping part of these two kinds of road sign images, that is, the The sign block corresponds to two types of second road sign information, and the combination of the two second road sign information "go straight, turn left" is the combination of road sign information of the sign block.
  • the image 61 of the sign block is in the shape of F, which includes three kinds of road sign images of "go straight 63", “turn left 62" and “go straight and turn left 64", that is, the sign block corresponds to There are three kinds of second road sign information, and the combination of these three kinds of road sign information "go straight, turn left, and go straight and turn left” is the combination of road sign information of the sign block.
  • Step S406 judging whether the second road sign information of the sign block corresponds to a safe road sign, if any one of the second road sign information corresponds to a safe road sign, perform security processing according to the second road sign information corresponding to the safe road sign, and generate a corresponding safety road sign.
  • Driving strategy if all the second road sign information corresponds to other categories, such as road signs with traffic regulations, you can directly jump to step S407;
  • the second road surface sign information corresponding to the sign block 1 is a pedestrian crossing, which means that the location of the sign block 1 may be a pedestrian crossing now or in the past, and the pedestrian crossing belongs to a safe road sign.
  • the corresponding " "Pedestrian crossing" safe driving strategies such as slowing down to pass, pedestrian detection or traffic light detection and other safe driving strategies, and provide these safe driving strategies to the decision-making layer of the automatic driving module, etc., so that the decision-making layer of the automatic driving module controls the vehicle execution Corresponding safe driving strategies to maximize compliance with traffic regulations and ensure safe driving.
  • the corresponding redundant marker block can still proceed to the following steps S407 to S408 to further determine its road surface marker information and add it to the road feature information.
  • Step S407 using a machine learning algorithm to infer the first road surface marking information of the redundant marking block, so as to obtain the final recognition result of the redundant marking block;
  • the final recognition result of the redundant sign block includes the first road surface sign information of the redundant sign block.
  • the aforementioned clustering algorithm may be used to infer the first road sign information of the redundant sign block.
  • Step S408 generate road feature information (road feature) of the road image, and provide the road feature information to subsequent modules, such as an automatic driving module or a path planning, positioning module, etc.
  • the road feature information may include first road surface sign information of each sign block in the road image.
  • the road characteristic information may also include information such as obstacles and other markers other than marking lines.
  • the blacklist screening, reference data matching, pavement sign image matching, security processing, and machine learning algorithm-based speculation are ranked in order of complexity from low to high and accuracy from strong to weak .
  • the priority is executed in order from high to low, not only can identify redundant road signs efficiently and accurately, but also for scenes that cannot match data, scenes without reference data, scenes with reference data but a small amount, scenes with reference data but a large number Various scenes such as multiple scenes, simple road scenes, and complex road scenes can be supported.
  • Fig. 7 shows an exemplary structure of a road sign recognition device provided by an embodiment of the present application.
  • the pavement marking recognition device may include:
  • An acquisition unit 710 configured to acquire a road image, where the road image includes a road surface
  • a redundancy determining unit 720 configured to determine redundant marker blocks of the road image
  • the sign recognition unit 730 is configured to obtain a recognition result of the redundant sign block according to the reference data and/or the road sign image, and the recognition result includes the first road sign information.
  • the redundancy determination unit 720 may be specifically configured to: determine redundant sign blocks of the road image according to preconfigured road sign rules, road sign images and/or color differences of sign blocks in the road image.
  • the redundancy determination unit 720 can be specifically configured to: determine that the sign block is a redundant sign block when the sign block of the road image does not meet the preconfigured road sign rules; and/or, when the sign block of the road image When the shape of the road sign image does not match, the sign block is determined to be a redundant sign block; and/or, when the color difference of the sign block of the road image is greater than a first preset threshold, the sign block is determined to be a redundant sign block.
  • the sign recognition unit 730 can be specifically configured to: determine the first road surface sign information of the redundant sign block according to the reference data; wherein the reference data includes one or more of the following: high-precision map data, traffic sign information , Navigation map data.
  • the sign recognition unit 730 can be specifically configured to: obtain two or more second road sign information of the redundant sign block by matching the redundant sign block with the road sign image, and each second road sign information The corresponding road sign images are included in the image of the redundant sign block; according to two or more than two kinds of second road sign information of the redundant sign block, the first road sign information of the redundant sign block is determined.
  • the road sign recognition device further includes: a safety processing unit 740, configured to generate a safe driving policy according to the second road sign information corresponding to the safe road sign.
  • the sign recognition unit 730 can be specifically configured to: use machine learning algorithms to infer the third road sign information and the third The reliability of the road sign information; the third road sign information with the highest reliability is used as the first road sign information.
  • the sign recognition unit 730 can be specifically configured to: calculate the color difference corresponding to the second road sign information by using the image data corresponding to the second road sign information in the redundant sign block;
  • the road sign information is the initial clustering center, and the image data of the redundant sign blocks are clustered to obtain two or more third road sign information and image data sets of the third road sign information of the redundant sign blocks; according to The image data set of the third road sign information determines the reliability of the third road sign information.
  • the sign identification unit 730 is further configured to: before clustering the image data of redundant sign blocks, remove the second road sign information whose color difference is greater than a second predetermined threshold.
  • the sign recognition unit 730 is further configured to: before determining the credibility of the third road sign information, remove the third road sign information that does not meet the road sign rules.
  • Fig. 8 shows an example diagram of the deployment of the road sign recognition device and its related modules in this embodiment.
  • the road surface sign recognition device can be arranged in the road feature recognition module, and the road feature recognition module is used to perform road feature recognition based on the road perception data (including the image in front of the vehicle, etc.) acquired by the sensor layer of the automatic driving module.
  • the road image above is obtained, the road image enters the road sign recognition device, and the road sign recognition device performs road sign recognition on road features, including identification of redundant sign blocks, safety processing of safety-level sign blocks, etc., and finally obtains road features
  • the information is output to the decision-making layer of the automatic driving module, so that the decision-making layer of the automatic driving module can perform decisions such as positioning, planning, control, etc., and can also provide a safe driving strategy (not shown in the figure) to the automatic driving module, so that the automatic driving module
  • the decision-making layer executes safe processing decisions according to the safe driving strategy.
  • Fig. 8 is only an example. In practical applications, the deployment of the pavement sign recognition device and its interaction with other modules can be flexibly adjusted according to actual needs, and its specific implementation is not limited to the one shown in FIG. 8 .
  • FIG. 9 is a schematic structural diagram of a computing device 900 provided by an embodiment of the present application.
  • the computing device 900 includes: one or more processors 910 , one or more memories 920 .
  • the processor 910 may be connected to the memory 920 .
  • the memory 920 can be used to store the program codes and data. Therefore, the memory 920 may be a storage unit inside the processor 910, or an external storage unit independent of the processor 910, or may include a storage unit inside the processor 910 and an external storage unit independent of the processor 910. part.
  • computing device 900 may also include a communication interface 930 . It should be understood that the communication interface 930 in the computing device 900 shown in FIG. 9 can be used to communicate with other devices.
  • computing device 900 may further include a bus.
  • the memory 920 and the communication interface 930 may be connected to the processor 910 through a bus.
  • the processor 910 may be a central processing unit (central processing unit, CPU).
  • the processor can also be other general-purpose processors, digital signal processors (digital signal processors, DSPs), application specific integrated circuits (application specific integrated circuits, ASICs), field programmable gate arrays (field programmable gate arrays, FPGAs) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
  • a general-purpose processor may be a microprocessor, or the processor may be any conventional processor, or the like.
  • the processor 910 adopts one or more integrated circuits for executing related programs, so as to implement the technical solutions provided by the embodiments of the present application.
  • the memory 920 may include read-only memory and random-access memory, and provides instructions and data to the processor 910 .
  • a portion of processor 910 may also include non-volatile random access memory.
  • processor 910 may also store device type information.
  • the processor 910 executes the computer-executable instructions in the memory 920 to perform the operation steps of the above-mentioned method for recognizing road signs.
  • the computing device 900 may correspond to a corresponding body executing the methods according to the various embodiments of the present application, and the above-mentioned and other operations and/or functions of the modules in the computing device 900 are for realizing the present invention For the sake of brevity, the corresponding processes of the methods in the embodiments are not repeated here.
  • the computing device 900 may be implemented as a functional unit in a chip, an independent chip, a functional unit of a vehicle-mounted terminal device, or an independent vehicle-mounted terminal device.
  • the computing device 900 may be a functional unit in a car machine, a cockpit domain controller (cockpit domain controller, CDC), or a mobile data center/multi-domain controller (Mobile Data Center/Multi-Domain Controller, MDC). /module, the embodiment of the present application does not limit the shape and deployment mode of the computing device 900 .
  • the embodiment of the present application also provides a vehicle, which may include the computing device, the road sign recognition device, the computer storage medium or the computer program product provided in the embodiment of the present application.
  • the embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the program is run by a processor, the processor executes the above-mentioned method for recognizing road signs.
  • the computer readable medium may be a computer readable signal medium or a computer readable storage medium.
  • a computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, semiconductor system, device, or device, or any combination thereof.
  • Computer-readable storage media include: electrical connections with one or more conductors, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable Read memory, optical fiber, portable compact disk read only memory, optical storage devices, magnetic storage devices, or any suitable combination of the above.
  • the embodiment of the present application further provides a computer program product, which includes a computer program, and when the computer program is run by a processor, the processor is executed to execute the above-mentioned method for recognizing road signs.
  • the programming language of the computer program product may be one or more, and the programming language may include but not limited to object-oriented programming languages such as Java and C++, conventional procedural programming languages such as "C" language, etc. language.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)
  • Traffic Control Systems (AREA)

Abstract

本申请涉及智能汽车领域，尤其涉及一种路面标志识别方法及装置、设备、存储介质、车辆。本申请一实施例中，结合参考数据和/或路面标志图像针对道路图像的冗余标志块进行识别，能够高效准确地识别冗余的路面标志，解决因路面标志混杂引起的冗余路面标志识别困难等问题，提升车辆的自动驾驶、智能驾驶等功能的可靠性。

Description

路面标志识别方法及装置、设备、存储介质、车辆 技术领域
本申请涉及智能汽车领域，尤其涉及一种路面标志识别方法及装置、设备、存储介质、车辆。
背景技术
随着汽车行业的高速发展，自动驾驶、智能驾驶等辅助驾驶功能已成为车辆的必备功能，而这些功能能否实现依赖于车辆对诸如障碍物、路面标志等道路特征的感知。
因道路的路面标志数量巨大、种类繁多、年久褪色、擦除不当、新旧交叠、更新频繁等混杂情况，车辆难以高效准确地识别道路上的路面标志，从而影响车辆辅助驾驶功能的实现，甚至引起辅助驾驶功能的运行错误。
发明内容
为解决上述技术问题，本申请实施例提供了一种路面标志识别方法及装置、设备、存储介质、车辆，能够高效准确地识别路面标志。
本申请第一方面提供了一种路面标志识别方法，包括：
获取道路图像，道路图像中包含路面；
确定道路图像的冗余标志块；
根据参考数据和/或路面标志图像，得到冗余标志块的识别结果，识别结果包含第一路面标志信息。
由此，结合参考数据和/或路面标志图像，即可高效准确地识别道路路面的冗余路面标志。
作为第一方面的一种可能的实现方式，确定道路图像中的冗余标志块，包括：根据预配置的路面标志规则、路面标志图像和/或道路图像中标志块的色差，确定道路图像的冗余标志块。
由此，可高效、准确并全面的找出道路图像中的冗余标志块。
作为第一方面的一种可能的实现方式，确定道路图像中的冗余标志块，包括：在道路图像的标志块不符合预配置的路面标志规则时，标志块为冗余标志块；和/或，在道路图像的标志块与路面标志图像的形状不匹配时，标志块为冗余标志块；和/或，在道路图像的标志块的色差大于第一预设阈值时，标志块为冗余标志块。
由此，可根据需要采用多种手段高效、准确并全面的找出道路图像中的冗余标志块。
作为第一方面的一种可能的实现方式，根据参考数据和/或路面标志图像，得到冗余标志块的识别结果，包括：根据参考数据确定冗余标志块的第一路面标志信息；其中，参考数据包括如下之一或多项：高精地图数据、交通指示牌信息、导航地图数据。
由此，可借由一种或多种参考数据高效准确确定冗余标志块的识别结果。
作为第一方面的一种可能的实现方式，根据参考数据和/或路面标志图像，得到冗余标志块的识别结果，包括：通过匹配冗余标志块与路面标志图像，得到冗余标志块的两种或两种以上第二路面标志信息，每个第二路面标志信息对应的路面标志图像均包含在冗余标志块的图像中；根据冗余标志块的两种或两种以上第二路面标志信息，确定冗余标志块的第一路面标志信息。
由此，可通过图像匹配等方式高效准确地确定冗余标志块的路面标志信息。
作为第一方面的一种可能的实现方式，方法还包括：根据对应安全路标的第二路面标志信息，生成安全驾驶策略。由此，可最大程度地遵守交规和确保安全驾驶。
作为第一方面的一种可能的实现方式，根据冗余标志块的两种或两种以上第二路面标志信息，确定冗余标志块的第一路面标志信息，包括：根据冗余标志块的两种或两种以上第二路面标志信息，采用机器学习算法推测冗余标志块的第三路面标志信息和第三路面标志信息的可信度；以可信度最高的第三路面标志信息作为第一路面标志信息。
由此，可利用机器学习算法与可信度，准确高效地确定冗余标志块的路面标志信息。
作为第一方面的一种可能的实现方式，根据冗余标志块的两种或两种以上第二路面标志信息，采用机器学习算法推测冗余标志块的第三路面标志信息和第三路面标志信息的可信度，包括：利用冗余标志块中对应第二路面标志信息的图像数据，计算第二路面标志信息对应的色差；以色差较大的至少两个第二路面标志信息为初始聚类中心，对冗余标志块的图像数据聚类，以获得冗余标志块的两个或两个以上第三路面标志信息和第三路面标志信息的图像数据集；根据第三路面标志信息的图像数据集，确定第三路面标志信息的可信度。
由此，可通过色差和聚类的结合准确高效地确定冗余标志块的路面标志信息。
作为第一方面的一种可能的实现方式，根据冗余标志块的两种或两种以上第二路面标志信息，采用机器学习算法推测冗余标志块的第三路面标志信息和第三路面标志信息的可信度，还包括：对冗余标志块的图像数据聚类之前，去除色差大于第二预定阈值的第二路面标志信息。
由此，可进一步提高冗余标志块的路面标志信息的准确性，同时降低复杂度，提高效率。
作为第一方面的一种可能的实现方式，根据冗余标志块的两种或两种以上第二路面标志信息，确定冗余标志块的第一路面标志信息，还包括：确定第三路面标志信息的可信度之前，去除不符合路面标志规则的第三路面标志信息。
由此，可进一步提高冗余标志块的路面标志信息的准确性，同时降低复杂度，提高效率。
本申请第一方面提供了一种路面标志识别装置，包括：
获取单元，用于获取道路图像，道路图像中包含路面；
冗余确定单元，用于确定道路图像的冗余标志块；
标志识别单元，用于根据参考数据和/或路面标志图像，得到冗余标志块的识别结果，识别结果包含第一路面标志信息。
作为第二方面的一种可能的实现方式，冗余确定单元，具体用于：根据预配置的路面标志规则、路面标志图像和/或道路图像中标志块的色差，确定道路图像的冗余标志块。
作为第二方面的一种可能的实现方式，冗余确定单元，具体用于：在道路图像的标志块不符合预配置的路面标志规则时，确定标志块为冗余标志块；和/或，在道路图像的标志块与路面标志图像的形状不匹配时，确定标志块为冗余标志块；和/或，在道路图像的标志块的色差大于第一预设阈值时，确定标志块为冗余标志块。
作为第二方面的一种可能的实现方式，标志识别单元，具体用于：根据参考数据确定冗余标志块的第一路面标志信息；其中，参考数据包括如下之一或多项：高精地图数据、交通指示牌信息、导航地图数据。
作为第二方面的一种可能的实现方式，标志识别单元，具体用于：通过匹配冗余标志块与路面标志图像，得到冗余标志块的两种或两种以上第二路面标志信息，每个第二路面标志信息对应的路面标志图像均包含在冗余标志块的图像中；根据冗余标志块的两种或两种以上第二路面标志信息，确定冗余标志块的第一路面标志信息。
作为第二方面的一种可能的实现方式，路面标志识别装置还包括：安全处理单元，用于根据对应安全路标的第二路面标志信息，生成安全驾驶策略。
作为第二方面的一种可能的实现方式，标志识别单元，具体用于：根据冗余标志块的两种或两种以上第二路面标志信息，采用机器学习算法推测冗余标志块的第三路面标志信息和第三路面标志信息的可信度；以可信度最高的第三路面标志信息作为第一路面标志信息。
作为第二方面的一种可能的实现方式，标志识别单元，具体用于：利用冗余标志块中对应第二路面标志信息的图像数据，计算第二路面标志信息对应的色差；以色差较大的至少两个第二路面标志信息为初始聚类中心，对冗余标志块的图像数据聚类，以获得冗余标志块的两个或两个以上第三路面标志信息和第三路面标志信息的图像数据集；根据第三路面标志信息的图像数据集，确定第三路面标志信息的可信度。
作为第二方面的一种可能的实现方式，标志识别单元，还用于：对冗余标志块的图像数据聚类之前，去除色差大于第二预定阈值的第二路面标志信息。
作为第二方面的一种可能的实现方式，标志识别单元，还用于：确定第三路面标志信息的可信度之前，去除不符合路面标志规则的第三路面标志信息。
本申请第三方面提供了一种计算设备，包括处理器和存储器，存储器存储有程序指令，程序指令当被处理器执行时使得处理器执行第一方面的路面标志识别方法。
本申请第四方面提供了一种计算机可读存储介质，其上存储有程序指令，其特征在于，程序指令当被计算机执行时使得计算机执行第一方面的路面标志识别方法。
本申请第五方面提供了一种车辆，包括第三方面的计算设备或者第二方面的路面标志识别装置。
本申请第六方面提供了一种计算机程序产品，其包括计算机程序，计算机程序在被处理器运行时使得该处理器执行第一方面的路面标志识别方法。
本申请实施例先找出道路图像的冗余标志块，再根据参考数据和/或路面标志图像对冗余标志块执行针对性处理，能够高效准确地识别冗余的路面标志，解决因路面标志混杂引起的冗余路面标志识别困难等问题，可有效提高道路特征信息的准确性和车辆辅助驾驶功能的可靠性。
本申请的这些和其它方面在以下(多个)实施例的描述中会更加简明易懂。
附图说明
图1是本申请实施例一应用场景的示例图；
图2是本申请实施例又一应用场景的示例图；
图3是本申请实施例提供的路面标志识别方法的流程示意图；
图4是本申请实施例路面标志识别方法的具体实现流程示意图；
图5是本申请实施例冗余标志块的图像示意图；
图6是本申请实施例又一冗余标志块的图像示意图；
图7是本申请实施例提供的路面标志识别装置的结构示意图；
图8是本申请实施例提供的路面标志识别装置的部署示意图；
图9是本申请实施例提供的计算设备的结构示意图。
具体实施方式
术语解释：
路面标志：涂画在道路表面的交通标志，包括但不限于车道线、引导箭头、人行道、停车线、网状线、导流线、限制速度标志、限制车型标志等。
安全路标：以人行道、停车线、限速标志为代表的、很大程度上关乎行车安全的路面标志。
交规路标：以车道线、引导箭头、导流线为代表的、对行车安全影响相对较小、但很大程度上关乎交通规则执行的路面标志。
因道路的路面标志数量巨大、种类繁多、年久褪色、擦除不当、新旧交叠、更新频繁等混杂情况，常造成车辆识别到的路面标志出现冗余。目前，车辆仅可识别简单的冗余路面标志，针对冗余的路面标志，主要存在如下问题：1)缺乏系统性的处理流程；2)针对不同道路、不同路况，冗余路面标志的识别处理缺乏一致性，难以产品化；3)只能处理例如新旧车道线等简单冗余的路面标志，无法处理存在例如本文图1、图2、图5和图6等复杂冗余的路面标志；4)对于包含复杂冗余路面标志的场景，难以制定安全、符合交规的行车策略。
鉴于此，本申请实施例提供一种改进的路面标志识别方法及装置、设备、存储介质，先找出道路图像的冗余标志块，再根据参考数据和/或路面标志图像对冗余标志块执行针对性处理，能够高效准确地识别路面标志，解决因路面标志混杂引起的冗余路面标志识别困难等问题，保证了道路特征信息的准确性，提升了车辆辅助驾驶功能的可靠性。
本申请实施例可适用于各种场景，尤其适用于路面标志混乱的城市、城郊道路等场景。本申请实施例可适用于各种行车模式，尤其适用于自动驾驶、智能驾驶等辅助 驾驶模式。
本申请实施例中的“车辆”可以是各种类型。例如，本申请的“车辆”可以是但不限于私人汽车、商用汽车、公共汽车、客运车、高铁、地铁、无人车、无人机、物流运输车、无人运输车、船只、飞行器等动力驱动的设备，“车辆”的动力类型可以为燃油驱动、纯电动、氢燃料电池驱动、混合动力等。
图1示出了本申请实施例适用的一道路场景示例图。参见图1的道路及其交通指示牌可知，道路的最左侧车道应为直行标志，但最左侧车道的路面标志却是直行左转，可见最左侧车道的路面标志涂画错误，该路面标志可能被识别为直行、直行左转、左转等多种路标，也即，该路面标志为冗余路面标志。
图2示出了本申请实施例适用的另一道路场景的示例图。参见图2所示，人行横道标志(即图2中的斑马线)存在旧标志线擦除不当的问题，新旧标志线共同存在，该路面标志可能被识别为人行横道、无效标志或其他路标，也就是说，该路面标志也是冗余路面标志。
通过本申请实施例可实现图1、图2所示场景中冗余路面标志的高效准确识别。
下面详细说明本申请实施例的具体实施方式。
图3示出了本申请实施例提供的路面标志识别方法的流程图。参见图3所示，本申请实施例提供的路面标志识别方法可以包括如下步骤：
步骤S310，获取道路图像，道路图像中包含路面；
一些实施例中，道路图像可以是但不限于由例如摄像头等车载传感器采集的包含路面标志的道路图像、对道路图像执行诸如标志物(包含路面标志、交通指示牌、交通信号灯等)、障碍物等道路特征提取等预处理得到的道路特征图像等。
步骤S320，确定道路图像的冗余标志块；
冗余标志块可以是任何无法唯一确定路面标志信息的标志块，标志块可以为道路图像的感兴趣区域(region of interest，ROI)，该道路图像ROI中包含连续标志线。
一些示例中，冗余标志块可能被识别为两种或两种以上路面标志，也即，冗余标志块的路面标志信息可能会包含两种或两种以上。
一些示例中，冗余标志块可能是完全冗余的。例如，因道路维修或重新规划，道路上的部分路面标志作废，但这些路面标志可能不会及时擦除或者因擦除不当仍保留有印记，这样的路面标志即为无效标志，对应该路面标志的标志块即是完全冗余的标志块。
一些实施例中，可以根据预配置的路面标志规则、路面标志图像和/或道路图像中标志块的色差，确定道路图像的冗余标志块。
一些实施例中，确定道路图像的冗余标志块可以包括如下任意一项或多项：
1)在道路图像的标志块的色差大于第一预设阈值时，标志块为冗余标志块；
一些示例中，单一标志块的内部颜色差异度，即色差Diff，可以通过下式(1)公式计算得到：
Diff＝(Rmax-Rmin)+(Gmax-Gmin)+(Bmax-Bmin)     (1)
其中，Xmax为标志块内相应颜色分量最大的像素点的X值，Xmin为标志块内相 应颜色分量最的像素点的X值，X＝R\G\B(红\绿\蓝)。
2)在道路图像的标志块与路面标志图像的形状不匹配时，标志块为冗余标志块；
这里，可以通过匹配标志块与路面标志图像，得到标志块与各个路面标志图像的形状相似度，如果形状相似度高于预设的相似度阈值，可以认为标志块与该路面标志图像匹配，由此，可以获得标志块的识别结果，该识别结果中可以包括与标志块匹配的所有路面标志图像对应的路面标志信息，如果与标志块匹配的路面标志图像唯一，该标志块不是冗余标志块，如果与标志块匹配的路面标志图像有两种或两种以上，该标志块为冗余标志块。
3)在道路图像的标志块不符合预配置的路面标志规则时，标志块为冗余标志块；
这里，判断道路图像中所有标志块的识别结果是否符合路面标志规则，如果某个标志块的识别结果不符合任意一项路面标志规则，该标志块即为冗余标志块，如果某个标志块的识别结果符合所有的路面标志规则，该标志块不是冗余标志块，以该识别结果作为标志块的最终识别结果。
一些实施例中，可以通过预先配置黑名单的方式设置路面标志规则。具体地，可以预先配置黑名单，该黑名单可以包括所有的路面标志规则，在第3)项中可通过遍历黑名单实现冗余标志块的确定。
一些实施例中，对于道路图像中的标志块，可以按照预定的执行顺序依次执行上述的第1)项、第2)项和第3)项，以便高效且全面地筛选出道路图像的冗余标志块。
具体应用中，可以根据场景需求、地域特征、交通规则等调整路面标志规则。路面标志规则可以预先配置。例如，可采用例如人工配置、通过解析交通规则相关数据自动配置等方式。
一些实施例中，路面标志规则可以包括但不限于如下之一或多项：1)导流线和网状线内不应有停车线；2)引导箭头、车道线之间的间距应至少为2米；3)同一条车道线上不应出现相矛盾的引导箭头。
步骤S330，根据参考数据和/或路面标志图像，得到冗余标志块的识别结果，识别结果包含第一路面标志信息。
一些实施例中，参考数据可以包括如下之一或多项：高精地图数据、交通指示牌信息、导航地图数据。具体应用中，参考数据可存储于车辆的存储器中、也可来自于云端，并且参考数据可定期更新。
一些实施例中，步骤S330可以包括：根据参考数据确定冗余标志块的第一路面标志信息。
一些实施例中，根据参考数据确定冗余标志块的路面标志信息，具体可以包括：根据第一参考数据，确定冗余标志块的路面标志信息；若无法唯一确定冗余标志块的路面标志信息，可根据第二参考数据继续确定冗余标志块的路面标志信息；若仍无法唯一确定冗余标志块的路面标志信息，可继续根据第三参考数据确定冗余标志块的路面标志信息。如此执行，直到所有的参考数据均已尝试或者冗余标志块的路面标志信息确定。
一些实施例中，第一参考数据、第二参考数据和第三参考数据可按照优先级排序或者按照其他的预定规则排序，按照参考数据的排序依次执行，直到能够唯一确定冗 余标志块的路面标志信息或者所有的参考数据均已尝试。
举例来说，高精地图数据可以作为优先级最高的参考数据，即第一参考数据；道路交通指示牌数据可以作为优先级次高的参考数据，即第二参考数据；导航地图数据可作为优先级最低的参考数据，即第三参考数据。具体应用中，可先使用高精地图数据确定冗余标志块的路面标志信息，再使用道路交通指示牌数据确定冗余标志块的路面标志信息，最后使用导航地图数据确定冗余标志块的路面标志信息。
以图1为例，根据交通指示牌可知，标志块3应是直行，非直行左转、非左转，因此，可以确定标志块3的第一路面标志信息为直行。
一些实施例中，步骤S330还可包括：步骤a1，通过匹配所述冗余标志块与所述路面标志图像，得到冗余标志块的两种或两种以上第二路面标志信息，每个第二路面标志信息对应的路面标志图像均包含在冗余标志块的图像中；步骤a2，根据冗余标志块的两种或两种以上第二路面标志信息，确定冗余标志块的第一路面标志信息。
一些实施例中，路面标志识别方法还可包括：根据对应安全路标的所述第二路面标志信息，生成安全驾驶策略。
一些实施例中，步骤a2可以包括：根据冗余标志块的两种或两种以上第二路面标志信息，采用机器学习算法推测冗余标志块的第三路面标志信息和第三路面标志信息的可信度；以可信度最高的第三路面标志信息作为第一路面标志信息。具体地，可以利用冗余标志块中对应第二路面标志信息的图像数据，计算第二路面标志信息对应的色差；以色差较大的至少两个第二路面标志信息为初始聚类中心，对冗余标志块的图像数据聚类，以获得冗余标志块的两个或两个以上第三路面标志信息和第三路面标志信息的图像数据集；根据第三路面标志信息的图像数据集，确定第三路面标志信息的可信度。
一些实施例中，可按照下式(2)计算可信度：
可信度＝(R均值+G均值+B均值)/log(size(分类)+1)    (2)
其中，分类是指聚类得到的类别数量，也即聚类得到的第三路面标志信息的数量，R均值为图像数据集中各像素数据的R值的均值，G均值为图像数据集中各像素数据的G值的均值，B均值为图像数据集中各像素数据的B值的均值。
一些实施例中，对冗余标志块的图像数据聚类之前，可以先去除色差大于第二预定阈值的第二路面标志信息。
一些实施例中，还可以在确定第三路面标志信息的可信度之前，去除不符合路面标志规则的第三路面标志信息。
下面结合示例详细说明本申请实施例路面标志识别方法的具体实现过程。
参见图4，本申请实施例路面标志识别的具体实现流程可以包括如下步骤：
步骤S401，获取道路图像，道路图像中包含路面标志；
步骤S402，提取道路图像中的标志块；
这里，提取标志块的同时，可以获取标志块的位置信息。一些示例中，标志块的位置信息可以包括但不限于标志块的图像中所有像素点的位置信息、标志块的图像中预定像素点(例如图像边缘、图像中心等像素点)的位置信息。
步骤S403，确定道路图像的冗余标志块和非冗余标志块的第一路面标志信息，如果道路图像中包含冗余标志块，可继续步骤S404，如果道路图像中不包含冗余标志块可直接跳转至步骤S408；
具体地，分别将各个标志块与已知路面标志图像匹配，获得每个标志块的初始识别结果，该初始识别结果中包括与标志块匹配的路面标志图像所对应的路面标志信息，下文将该路面标志信息命名为初始路面标志信息。
这里，如果标志块能够与某个路面标志图像唯一匹配，也即，可以确定该标志块的路面标志信息为该路面标志图像指示的路面标志，该标志块不是冗余标志块，而是非冗余标志块，如果标志块与所有路面标志图像均无法唯一匹配，说明该标志块是冗余标志块。
如果标志块不是冗余标志块，即标志块为非冗余标志块时，可以标志块的初始路面标志信息作为标志块的第一路面标志信息。如果标志块是冗余标志块，则需进入后续步骤的处理，直到得到该冗余标志块的第一路面标志信息。
以图1为例，道路图像的初始识别结果如下表1所示。其中，标志块1代表右侧车道内的引导箭头，标志块2代表中间车道内的引导箭头，标志块3代表左侧车道内的引导箭头：
Figure PCTCN2022070325-appb-000001
表1
以图2为例，道路图像的初始识别结果如下表2所示，标志块1代表图2中的斑马线，包括新标志线和旧标志线：
Figure PCTCN2022070325-appb-000002
表2
一些示例中，可以分别计算各个标志块的色差，如果标志块的色差大于第一预定阈值，说明该标志块的内部颜色差异较大，可能包含了新旧标志线，该标志块为冗余标志块，如果标志块的色差小于或等于第一预定阈值，说明该标志块的标志线内部颜色均匀，该标志块不是冗余标志块。
一些示例中，可以根据标志块的初始识别结果和预先配置的黑名单，判断标志块是否为冗余标志块，如果标志块符合黑名单中的所有路面标志规则，该标志块不是冗余标志块，如果标志块不符合黑名单中的任意一项路面标志规则，该标志块即为冗余标志块。关于黑名单，可参见前文相关描述，此处不再赘述。
步骤S404，根据参考数据确定冗余标志块的第一路面标志信息，如果所有的参考数据均已尝试但仍有冗余标志块的路面标志信息没有唯一确定，则继续后续步骤S405，若所有冗余标志块的路面标志信息均已唯一确定，可以直接跳转至步骤S408。
这里，根据参考数据确定的第一路面标志信息与冗余标志块的初始路面标志信息可能相同、也可以不同。
步骤S405，将标志块与路面标志图像匹配，确定冗余标志块的路面标志信息组合，该路面标志信息组合可以包括两个或两个以上第二路面标志信息；
具体地，匹配规则可以为：将标志块的图像和各个路面标志图像匹配，确定标志块的图像包含了哪些路面标志图像，这些路面标志图像对应的路面标志信息即为该标志块的第二路面标志信息，使用乘法原理将不同标志块的第二路面标志信息进行组合即可得到道路图像中冗余标志块的路面标志信息组合。
参见图5的冗余标志块图像示例，该标志块的图像51中包含了直行53和左转52这两种路面标志图像，阴影部分为这两种路面标志图像的重叠部分，也即，该标志块对应2种第二路面标志信息，这两种第二路面标志信息的组合“直行、左转”即为标志块的路面标志信息组合。
参见图6的示例，该标志块的图像61为F形状，其包括了“直行63”、“左转62”和“直行左转64”这三种路面标志图像，也即，该标志块对应3种第二路面标志信息，这三种路面标志信息的组合“直行、左转和直行左转”即为标志块的路面标志信息组合。
步骤S406，判断标志块的第二路面标志信息是否对应安全路标，若任意一项第二路面标志信息对应安全路标，则根据对应安全路标的第二路面标志信息，执行安全处理，生成相应的安全行驶策略，若所有的第二路面标志信息均对应了其他类别，例如交规路标，可以直接跳转至步骤S407；
以图2的场景为例，标志块1对应的第二路面标志信息为人行横道，代表标志块1所在的位置现在或历史上可能是人行横道，人行横道属于安全路标，从安全角度出发，可以生成对应“人行横道”的安全行驶策略，例如减速通过、进行行人检测或者进行红绿灯检测等安全行驶策略，并将这些安全行驶策略提供给例如自动驾驶模块的决策层等，以便自动驾驶模块的决策层控制车辆执行相应的安全行驶策略，以最大程度地遵守交规和确保安全驾驶。
执行安全处理后，相应冗余标志块仍可继续后续步骤S407～步骤S408，进一步确定其路面标志信息并添加到道路特征信息中。
步骤S407，利用机器学习算法推测冗余标志块的第一路面标志信息，以得到冗余标志块的最终识别结果；
这里，冗余标志块的最终识别结果包括冗余标志块的第一路面标志信息。具体地，可以采用前文的聚类算法推测冗余标志块的第一路面标志信息。
步骤S408，生成道路图像的道路特征信息(road feature)，并将道路特征信息提供给后续模块，例如自动驾驶模块或路径规划、定位模块等。
这里，道路特征信息可以包括道路图像中各个标志块的第一路面标志信息。此外，道路特征信息还可包括诸如障碍物、除标志线之外的其他标志物的信息。
图4的示例性实现流程中，经过包含路面标志规则的黑名单筛选、参考数据匹配、路面标志图像匹配、安全处理、基于机器学习算法的推测等一系列处理，不仅各种路面标志的冗余形态均可完成处理，而且安全合规，同时还可高效、准确地排除冗余路面标志，可以有效提高道路特征信息的准确性和车辆辅助驾驶功能的可靠性。
具体应用中，图4的示例性实现流程中，该黑名单筛选、参考数据匹配、路面标志图像匹配、安全处理、基于机器学习算法的推测按照复杂度由低到高、准确性由强到弱、优先级由高到低的顺序执行，不仅能够高效准确地识别冗余路面标志，而且对于无法匹配数据的场景、无参考数据的场景、有参考数据但数量少的场景、有参考数据但数量多的场景、简单道路场景、复杂道路场景等各种场景均可支持。
图7示出了本申请实施例提供的一种路面标志识别装置的示例性结构。参见图7所示，路面标志识别装置可以包括：
获取单元710，用于获取道路图像，道路图像中包含路面；
冗余确定单元720，用于确定道路图像的冗余标志块；
标志识别单元730，用于根据参考数据和/或路面标志图像，得到冗余标志块的识别结果，识别结果包含第一路面标志信息。
一些实施例中，冗余确定单元720，具体可用于：根据预配置的路面标志规则、路面标志图像和/或道路图像中标志块的色差，确定道路图像的冗余标志块。
一些实施例中，冗余确定单元720，具体可用于：在道路图像的标志块不符合预配置的路面标志规则时，确定标志块为冗余标志块；和/或，在道路图像的标志块与路面标志图像的形状不匹配时，确定标志块为冗余标志块；和/或，在道路图像的标志块的色差大于第一预设阈值时，确定标志块为冗余标志块。
一些实施例中，标志识别单元730，具体可用于：根据参考数据确定冗余标志块的第一路面标志信息；其中，参考数据包括如下之一或多项：高精地图数据、交通指示牌信息、导航地图数据。
一些实施例中，标志识别单元730，具体可用于：通过匹配冗余标志块与路面标志图像，得到冗余标志块的两种或两种以上第二路面标志信息，每个第二路面标志信息对应的路面标志图像均包含在冗余标志块的图像中；根据冗余标志块的两种或两种以上第二路面标志信息，确定冗余标志块的第一路面标志信息。
一些实施例中，路面标志识别装置还包括：安全处理单元740，用于根据对应安全路标的第二路面标志信息，生成安全驾驶策略。
一些实施例中，标志识别单元730，具体可用于：根据冗余标志块的两种或两种以上第二路面标志信息，采用机器学习算法推测冗余标志块的第三路面标志信息和第三路面标志信息的可信度；以可信度最高的第三路面标志信息作为第一路面标志信息。
一些实施例中，标志识别单元730，具体可用于：利用冗余标志块中对应第二路面标志信息的图像数据，计算第二路面标志信息对应的色差；以色差较大的至少两个第二路面标志信息为初始聚类中心，对冗余标志块的图像数据聚类，以获得冗余标志块的两个或两个以上第三路面标志信息和第三路面标志信息的图像数据集；根据第三路面标志信息的图像数据集，确定第三路面标志信息的可信度。
一些实施例中，标志识别单元730，还可用于：对冗余标志块的图像数据聚类之 前，去除色差大于第二预定阈值的第二路面标志信息。
一些实施例中，标志识别单元730，还可用于：确定第三路面标志信息的可信度之前，去除不符合路面标志规则的第三路面标志信息。
图8示出了本实施例实施例中路面标志识别装置的部署及其相关模块的示例图。参见图8所示，路面标志识别装置可以设置于道路特征识别模块中，道路特征识别模块用于基于自动驾驶模块的传感器层获取到的道路感知数据(包括车辆前方图像等)执行道路特征识别从而得到前文的道路图像，道路图像进入路面标志识别装置，路面标志识别装置对道路特征执行路面标志的识别，包括冗余标志块的识别、安全级标志块的安全处理等，将最终获得的道路特征信息输出给自动驾驶模块的决策层，以便自动驾驶模块的决策层执行诸如定位、规划、控制等决策，同时还可向自动驾驶模块提供安全驾驶策略(图中未示出)，以便自动驾驶模块的决策层根据该安全驾驶策略执行安全处理的决策。
需要说明的是，图8仅为示例。实际应用中，路面标志识别装置的部署及其与其他模块之间的交互，可根据实际需求灵活调整，其具体实施方式不限于图8所示的方式。
图9是本申请实施例提供的一种计算设备900的结构性示意性图。该计算设备900包括：一个或多个处理器910、一个或多个存储器920。
其中，该处理器910可以与存储器920连接。该存储器920可以用于存储该程序代码和数据。因此，该存储器920可以是处理器910内部的存储单元，也可以是与处理器910独立的外部存储单元，还可以是包括处理器910内部的存储单元和与处理器910独立的外部存储单元的部件。
可选地，计算设备900还可包括通信接口930。应理解，图9所示的计算设备900中的通信接口930可以用于与其他设备之间进行通信。
可选的，计算设备900还可以包括总线。其中，存储器920、通信接口930可以通过总线与处理器910连接。
应理解，在本申请实施例中，该处理器910可以采用中央处理单元(central processing unit，CPU)。该处理器还可以是其它通用处理器、数字信号处理器(digital signal processor，DSP)、专用集成电路(application specific integrated circuit，ASIC)、现场可编程门阵列(field programmable gate Array，FPGA)或者其它可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。或者该处理器910采用一个或多个集成电路，用于执行相关程序，以实现本申请实施例所提供的技术方案。
该存储器920可以包括只读存储器和随机存取存储器，并向处理器910提供指令和数据。处理器910的一部分还可以包括非易失性随机存取存储器。例如，处理器910还可以存储设备类型的信息。
在计算设备900运行时，处理器910执行存储器920中的计算机执行指令执行上述路面标志识别方法的操作步骤。
应理解，根据本申请实施例的计算设备900可以对应于执行根据本申请各实施例的方法中的相应主体，并且计算设备900中的各个模块的上述和其它操作和/或功能分 别为了实现本实施例各方法的相应流程，为了简洁，在此不再赘述。
实际应用中，计算设备900可实现为芯片中的一个功能单元、独立的芯片、车载终端设备的一个功能单元或独立的车载终端设备。一些实施例中，计算设备900可以是车机、座舱域控制器(cockpit domain controller，CDC)、移动数据中心/多域控制器(Mobile Data Center/Multi-Domain Controller，MDC)中的一个功能单元/模块，本申请实施例对计算设备900的形态和部署方式不做限定。
本申请实施例还提供一种车辆，该车辆可包括本申请实施例提供的计算设备、路面标志识别装置、计算机存储介质或计算机程序产品。
本申请实施例还提供了一种计算机可读存储介质，其上存储有计算机程序，该程序被处理器运行时使得处理器执行上述的路面标志识别方法。这里，计算机可读介质可以是计算机可读信号介质或者计算机可读存储介质。计算机可读存储介质例如可以是，但不限于电、磁、光、电磁、红外线、半导体的系统、装置或器件，或者任意以上的组合。计算机可读存储介质的更具体的例子(非穷举的列表)包括：具有一个或多个导线的电连接、便携式计算机磁盘、硬盘、随机存取存储器、只读存储器、可擦式可编程只读存储器、光纤、便携式紧凑磁盘只读存储器、光存储器件、磁存储器件或者上述的任意合适的组合。
本申请实施例还提供了一种计算机程序产品，其包括计算机程序，所述计算机程序在被处理器运行时使得该处理器执行上述的路面标志识别方法。这里，计算机程序产品的程序设计语言可以是一种或多种，该程序设计语言可以包括但不限于诸如Java、C++等面向对象的程序设计语言、诸如“C”语言等的常规过程式程序设计语言。
注意，上述仅为本申请的较佳实施例及所运用的技术原理。本领域技术人员会理解，本申请不限于这里所述的特定实施例，对本领域技术人员来说能够进行各种明显的变化、重新调整和替代而不会脱离本申请的保护范围。因此，虽然通过以上实施例对本申请进行了较为详细的说明，但是本申请不仅仅限于以上实施例，在不脱离本申请的构思的情况下，还可以包括更多其他等效实施例，均属于本申请的保护范畴。

Claims (24)

  1. 一种路面标志识别方法，其特征在于，包括：
    获取道路图像，所述道路图像中包含路面；
    确定所述道路图像的冗余标志块；
    根据参考数据和/或路面标志图像，得到所述冗余标志块的识别结果，所述识别结果包含第一路面标志信息。
  2. 根据权利要求1所述的路面标志识别方法，其特征在于，所述确定所述道路图像中的冗余标志块，包括：根据预配置的路面标志规则、所述路面标志图像和/或所述道路图像中标志块的色差，确定所述道路图像的冗余标志块。
  3. 根据权利要求1或2所述的路面标志识别方法，其特征在于，所述确定所述道路图像中的冗余标志块，包括：
    在所述道路图像的标志块不符合预配置的路面标志规则时，所述标志块为所述冗余标志块；和/或，
    在所述道路图像的标志块与所述路面标志图像的形状不匹配时，所述标志块为所述冗余标志块；和/或，
    在所述道路图像的标志块的色差大于第一预设阈值时，所述标志块为所述冗余标志块。
  4. 根据权利要求1-3任一项所述的路面标志识别方法，其特征在于，所述根据参考数据和/或路面标志图像，得到所述冗余标志块的识别结果，包括：
    根据所述参考数据确定所述冗余标志块的第一路面标志信息；
    其中，所述参考数据包括如下之一或多项：高精地图数据、交通指示牌信息、导航地图数据。
  5. 根据权利要求1-4任一项所述的路面标志识别方法，其特征在于，所述根据参考数据和/或路面标志图像，得到所述冗余标志块的识别结果，包括：
    通过匹配所述冗余标志块与所述路面标志图像，得到所述冗余标志块的两种或两种以上第二路面标志信息，每个所述第二路面标志信息对应的路面标志图像均包含在所述冗余标志块的图像中；
    根据所述冗余标志块的两种或两种以上第二路面标志信息，确定所述冗余标志块的第一路面标志信息。
  6. 根据权利要求5所述的路面标志识别方法，其特征在于，所述方法还包括：根据对应安全路标的所述第二路面标志信息，生成安全驾驶策略。
  7. 根据权利要求5所述的路面标志识别方法，其特征在于，所述根据所述冗余标 志块的两种或两种以上第二路面标志信息，确定所述冗余标志块的第一路面标志信息，包括：
    根据所述冗余标志块的两种或两种以上第二路面标志信息，采用机器学习算法推测所述冗余标志块的第三路面标志信息和所述第三路面标志信息的可信度；
    以可信度最高的所述第三路面标志信息作为所述第一路面标志信息。
  8. 根据权利要求7所述的路面标志识别方法，其特征在于，所述根据所述冗余标志块的两种或两种以上第二路面标志信息，采用机器学习算法推测所述冗余标志块的第三路面标志信息和所述第三路面标志信息的可信度，包括：
    利用所述冗余标志块中对应所述第二路面标志信息的图像数据，计算所述第二路面标志信息对应的色差；
    以所述色差较大的至少两个所述第二路面标志信息为初始聚类中心，对所述冗余标志块的图像数据聚类，以获得所述冗余标志块的两个或两个以上第三路面标志信息和所述第三路面标志信息的图像数据集；
    根据所述第三路面标志信息的图像数据集，确定所述第三路面标志信息的可信度。
  9. 根据权利要求8所述的路面标志识别方法，其特征在于，所述根据所述冗余标志块的两种或两种以上第二路面标志信息，采用机器学习算法推测所述冗余标志块的第三路面标志信息和所述第三路面标志信息的可信度，还包括：
    对所述冗余标志块的图像数据聚类之前，去除所述色差大于第二预定阈值的所述第二路面标志信息。
  10. 根据权利要求7-9任一项所述的路面标志识别方法，其特征在于，所述根据所述冗余标志块的两种或两种以上第二路面标志信息，确定所述冗余标志块的第一路面标志信息，还包括：确定所述第三路面标志信息的可信度之前，去除不符合所述路面标志规则的第三路面标志信息。
  11. 一种路面标志识别装置，其特征在于，包括：
    获取单元，用于获取道路图像，所述道路图像中包含路面；
    冗余确定单元，用于确定所述道路图像的冗余标志块；
    标志识别单元，用于根据参考数据和/或路面标志图像，得到所述冗余标志块的识别结果，所述识别结果包含第一路面标志信息。
  12. 根据权利要求11所述的路面标志识别装置，其特征在于，所述冗余确定单元，具体用于：根据预配置的路面标志规则、所述路面标志图像和/或所述道路图像中标志块的色差，确定所述道路图像的冗余标志块。
  13. 根据权利要求11或12所述的路面标志识别装置，其特征在于，所述所述冗余确定单元，具体用于：
    在所述道路图像的标志块不符合预配置的路面标志规则时，确定所述标志块为所述冗余标志块；和/或，
    在所述道路图像的标志块与所述路面标志图像的形状不匹配时，确定所述标志块为所述冗余标志块；和/或，
    在所述道路图像的标志块的色差大于第一预设阈值时，确定所述标志块为所述冗余标志块。
  14. 根据权利要求11-13任一项所述的路面标志识别装置，其特征在于，所述标志识别单元，具体用于：根据所述参考数据确定所述冗余标志块的第一路面标志信息；
    其中，所述参考数据包括如下之一或多项：高精地图数据、交通指示牌信息、导航地图数据。
  15. 根据权利要求11-14任一项所述的路面标志识别装置，其特征在于，所述标志识别单元，具体用于：
    通过匹配所述冗余标志块与所述路面标志图像，得到所述冗余标志块的两种或两种以上第二路面标志信息，每个所述第二路面标志信息对应的路面标志图像均包含在所述冗余标志块的图像中；
    根据所述冗余标志块的两种或两种以上第二路面标志信息，确定所述冗余标志块的第一路面标志信息。
  16. 根据权利要求15所述的路面标志识别装置，其特征在于，所述路面标志识别装置还包括：
    安全处理单元，用于根据对应安全路标的所述第二路面标志信息，生成安全驾驶策略。
  17. 根据权利要求15所述的路面标志识别装置，其特征在于，所述标志识别单元，具体用于：
    根据所述冗余标志块的两种或两种以上第二路面标志信息，采用机器学习算法推测所述冗余标志块的第三路面标志信息和所述第三路面标志信息的可信度；
    以可信度最高的所述第三路面标志信息作为所述第一路面标志信息。
  18. 根据权利要求17所述的路面标志识别装置，其特征在于，所述标志识别单元，具体用于：
    利用所述冗余标志块中对应所述第二路面标志信息的图像数据，计算所述第二路面标志信息对应的色差；
    以所述色差较大的至少两个所述第二路面标志信息为初始聚类中心，对所述冗余标志块的图像数据聚类，以获得所述冗余标志块的两个或两个以上第三路面标志信息和所述第三路面标志信息的图像数据集；
    根据所述第三路面标志信息的图像数据集，确定所述第三路面标志信息的可信度。
  19. 根据权利要求18所述的路面标志识别装置，其特征在于，所述标志识别单元，还用于：对所述冗余标志块的图像数据聚类之前，去除所述色差大于第二预定阈值的所述第二路面标志信息。
  20. 根据权利要求17-19任一项所述的路面标志识别装置，其特征在于，所述标志识别单元，还用于：确定所述第三路面标志信息的可信度之前，去除不符合所述路面标志规则的第三路面标志信息。
  21. 一种计算设备，其特征在于，包括处理器和存储器，所述存储器存储有程序指令，所述程序指令当被所述处理器执行时使得所述处理器执行如权利要求1-10任一项所述的方法。
  22. 一种车辆，其特征在于，包括如权利要求21所述的计算设备或如权利要求11-20任一项所述的路面标志识别装置。
  23. 一种计算机可读存储介质，其上存储有程序指令，其特征在于，所述程序指令当被计算机执行时使得所述计算机执行如权利要求1-10任一项所述的方法。
  24. 一种计算机程序产品，其包括计算机程序，其特征在于，所述计算机程序在被处理器运行时使得该处理器执行如权利要求1-10任一项所述的方法。
PCT/CN2022/070325 2022-01-05 2022-01-05 路面标志识别方法及装置、设备、存储介质、车辆 WO2023130263A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2022/070325 WO2023130263A1 (zh) 2022-01-05 2022-01-05 路面标志识别方法及装置、设备、存储介质、车辆
CN202280003624.1A CN116745820A (zh) 2022-01-05 2022-01-05 路面标志识别方法及装置、设备、存储介质、车辆

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/070325 WO2023130263A1 (zh) 2022-01-05 2022-01-05 路面标志识别方法及装置、设备、存储介质、车辆

Publications (1)

Publication Number Publication Date
WO2023130263A1 true WO2023130263A1 (zh) 2023-07-13

Family

ID=87072843

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/070325 WO2023130263A1 (zh) 2022-01-05 2022-01-05 路面标志识别方法及装置、设备、存储介质、车辆

Country Status (2)

Country Link
CN (1) CN116745820A (zh)
WO (1) WO2023130263A1 (zh)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104361350A (zh) * 2014-10-28 2015-02-18 奇瑞汽车股份有限公司 一种交通标识识别系统
CN107924617A (zh) * 2015-08-19 2018-04-17 索尼公司 用于为自主车辆确定导航信息的系统和方法
CN108564874A (zh) * 2018-05-07 2018-09-21 腾讯大地通途（北京）科技有限公司 地面标志提取的方法、模型训练的方法、设备及存储介质
CN108572644A (zh) * 2017-03-07 2018-09-25 德尔福技术公司 用于自动车辆的变道系统
US20200042806A1 (en) * 2018-07-31 2020-02-06 Here Global B.V. Method and system for unsupervised learning of road signs using vehicle sensor data and map data

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104361350A (zh) * 2014-10-28 2015-02-18 奇瑞汽车股份有限公司 一种交通标识识别系统
CN107924617A (zh) * 2015-08-19 2018-04-17 索尼公司 用于为自主车辆确定导航信息的系统和方法
CN108572644A (zh) * 2017-03-07 2018-09-25 德尔福技术公司 用于自动车辆的变道系统
CN108564874A (zh) * 2018-05-07 2018-09-21 腾讯大地通途（北京）科技有限公司 地面标志提取的方法、模型训练的方法、设备及存储介质
US20200042806A1 (en) * 2018-07-31 2020-02-06 Here Global B.V. Method and system for unsupervised learning of road signs using vehicle sensor data and map data

Also Published As

Publication number Publication date
CN116745820A (zh) 2023-09-12

Similar Documents

Publication Publication Date Title
CN110163176B (zh) 车道线变化位置识别方法、装置、设备和介质
CN110443225B (zh) 一种基于特征像素统计的虚实车道线识别方法及其装置
CN111046709B (zh) 车辆车道级定位方法、系统、车辆及存储介质
CN110763246A (zh) 自动驾驶车辆路径规划方法、装置、车辆及存储介质
US10185880B2 (en) Method and apparatus for augmenting a training data set
TWI493513B (zh) 車道偏離警示系統與車道辨識裝置以及相關方法
CN108279016B (zh) Had地图的平滑处理方法和装置、导航系统及自动驾驶系统
CN114394088B (zh) 泊车循迹轨迹生成方法、装置、电子设备及存储介质
CN107328423B (zh) 基于地图数据的弯道识别方法及其系统
CN110008891B (zh) 一种行人检测定位方法、装置、车载计算设备及存储介质
Rateke et al. Passive vision region-based road detection: A literature review
WO2024012211A1 (zh) 自动驾驶环境感知方法、介质及车辆
TWI549102B (zh) 車道偏離警示系統與車道辨識裝置以及相關方法
CN113740874A (zh) 一种道路边沿检测方法、电子设备及车辆
CN114092909A (zh) 一种车道线提取方法、装置、车辆及存储介质
WO2023130263A1 (zh) 路面标志识别方法及装置、设备、存储介质、车辆
CN110784680B (zh) 一种车辆定位方法、装置、车辆和存储介质
CN112699773A (zh) 交通灯识别方法、装置及电子设备
Jung et al. Intelligent Hybrid Fusion Algorithm with Vision Patterns for Generation of Precise Digital Road Maps in Self-driving Vehicles.
Gao et al. An automatic verification method for vehicle line-pressing violation based on CNN and geometric projection
CN110539748A (zh) 基于环视的拥堵跟车系统和终端
CN105416290A (zh) 检测车距的方法和装置
Borkar et al. Detecting lane markers in complex urban environments
TWI762887B (zh) 交通安全管控方法、車載裝置及可讀儲存介質
CN114119951A (zh) 车信标注的方法、装置、设备以及存储介质

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 202280003624.1

Country of ref document: CN

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22917740

Country of ref document: EP

Kind code of ref document: A1