Methods, apparatuses, and systems for determining a region of interest of a traffic sign

Abstract

The present application relates to the field of image processing. The present application provides a method for determining a region of interest of a traffic sign, comprising the following steps: reading geographical position information of a traffic sign affecting traffic behavior of a vehicle from a high-precision map, the geographical position information comprising a relative ground height of the traffic sign; reading terrain information of a road section from the high-precision map, the terrain information being limited by regional expression of the vehicle to the traffic sign; acquiring an image of a surrounding environment of the vehicle comprising the traffic sign; determining a region of interest of the traffic sign in the image based at least on the geographical position information and the terrain information of the traffic sign. The present application also provides an apparatus for determining a region of interest of a traffic sign and a system for identifying a state of a traffic sign. The present application aims to solve the problem that a traffic sign cannot be identified on a road section with a slope due to the limitation of regional expression of a high-precision map, effectively improving the accuracy of environmental perception of automatic driving.

Description

Technical Field

[0001] The present invention relates to a method for determining the region of interest of a traffic sign, an apparatus for determining the region of interest of a traffic sign, and a system for identifying the state of a traffic sign. Background Technology

[0002] Environmental perception is a crucial element in achieving autonomous driving. A mainstream solution for accurate traffic facility perception involves using high-precision maps to accurately record the 3D coordinates of most landmarks in the traffic scene. Through calibration parameters and appropriate coordinate system transformations, the bounding boxes of these landmarks can be projected onto the forward-view image from the vehicle's camera. Then, the status of landmarks can be purposefully identified by selecting a Region of Interest (ROI) within the projected area.

[0003] However, due to China's policy requirements regarding the confidentiality of map data and its public use, high-precision maps are strictly limited in their representation of specific information. For example, the absolute elevation information of landmarks cannot be directly represented; only relative height information can be recorded. Under these constraints, the relative height information of a landmark, marked with the height of a fixed point (generally, the default height of the road segment's starting point is 0), cannot be universally applied to the entire road segment. Especially for road segments with large height differences, the region of interest determined by the bounding box based on direct projection may deviate from the actual location of the landmark to be identified, or even be incorrectly located on another object in the traffic scene, thus causing misidentification in environmental perception.

[0004] Currently, existing technologies propose estimating road slope information based on vehicle posture and correcting the projection position of traffic lights based on the linear relationship between road segments with different slopes, thereby adjusting the accurate ROI position. However, the above solutions still have many shortcomings. In particular, this strategy of estimating slope information based on the vehicle's own posture has poor accuracy and is particularly susceptible to interference from various environmental factors, failing to provide reliable correction results in many cases. Summary of the Invention

[0005] The purpose of this invention is to provide a method for determining the region of interest of a traffic sign, an apparatus for determining the region of interest of a traffic sign, and a system for identifying the state of a traffic sign, so as to at least solve some of the problems in the prior art.

[0006] According to a first aspect of the present invention, a method for determining a region of interest for a traffic sign is provided, the method comprising the following steps:

[0007] S1: Read the geographical location information of traffic signs that affect vehicle traffic behavior from a high-precision map, wherein the geographical location information includes at least the relative ground height of the traffic signs;

[0008] S2: Read terrain information from high-precision maps that is limited by regional representation of the road segment from the vehicle to the traffic sign;

[0009] S3: Acquire an image of the vehicle's surrounding environment, including the traffic sign; and

[0010] S4: Based at least on the geographic location information of the traffic sign and the terrain information, determine the region of interest of the traffic sign in the image.

[0011] This invention specifically includes the following technical concept: by considering the terrain information of surrounding road sections when determining the region of interest, coordinate mapping deviations caused by the lack of absolute elevation information can be effectively compensated. Simultaneously, since the terrain information is derived from a high-precision map rather than estimated by the vehicle's own sensors, information reliability is significantly improved, and the software and hardware overhead for environmental perception in autonomous vehicles is reduced. Therefore, considering the limitations of Chinese regulations on map representation elements, the invention fully utilizes prior terrain information from high-precision maps within a limited scope, achieving accurate matching of the region of interest and laying the foundation for reliable decision-making by autonomous vehicles.

[0012] Optionally, step S1 includes: reading the latitude and longitude coordinates and relative ground elevation of the traffic sign from a high-precision map, wherein the coordinates are expressed as point coordinates (x, y, |z|) and / or as polygon bounding box coordinates (x1, y1, |z1|) - (x n , y n , |z n The location information of traffic signs can be read in the form of |).

[0013] In particular, the following technical advantages are achieved: by providing diverse representations of traffic signs in high-precision maps, different accuracy requirements for environmental perception can be met, thus improving the flexibility of the entire solution.

[0014] Optionally, the terrain information subject to regional expression limitations includes: slope classification and / or slope value range, especially percentage range, in particular state bit number form for road segments.

[0015] In particular, the following technical advantages are achieved: by converting road slope information that is not allowed to be expressed in China into a hierarchical representation on a high-precision map, data reliability can be ensured while ensuring national security and not violating regional restrictions on map representation.

[0016] Optionally, step S2 includes: reading a segmented representation of the terrain information of the road segment from a high-precision map, wherein, in the case of segmented representation, corresponding terrain information is assigned to each sub-segment of the road segment.

[0017] In particular, the following technical advantages are achieved: by segmenting terrain information in a high-precision map, for example, the slope direction of the entire road segment can be considered more in detail during the determination of the region of interest, thereby compensating for deviations in coordinate mapping to a certain extent with higher resolution.

[0018] Optionally, step S4 includes: projecting the geographic location information of the traffic sign from the map coordinate system of the high-precision map to the image coordinate system of the image, expanding a window area of ​​a predefined size outward based on the position of the traffic sign in the image coordinate system as the region of interest; and correcting the region of interest in terms of location and / or geometric characteristics according to the read terrain information.

[0019] In particular, the following technical advantages are achieved: based on the principle of similarity, which determines the approximate location of the region of interest in the image, a simple compensation possibility is provided by adjusting the location and / or geometric properties.

[0020] Optionally, the elevation difference between the road surface where the vehicle is located and the road surface where the traffic sign is located can be calculated using the terrain information; the elevation of the region of interest in the image can be adjusted based on the elevation difference information and / or the scaling ratio of the region of interest, especially the vertical scaling factor, can be adjusted.

[0021] In particular, the following technical advantages are achieved: by calculating the height difference information, the adjustment or correction amount of the region of interest can be mathematically visualized, thereby making the subsequent matching process fully controllable.

[0022] Optionally, in order to calculate the height difference information between the road surface where the vehicle is located and the road surface where the traffic sign is located, the road segment from the vehicle to the traffic sign is divided into multiple sub-segments according to the terrain information. For each sub-segment, the height difference information between the starting point and the ending point is calculated, and the height difference information of each sub-segment is accumulated to form the height difference information of the road segment.

[0023] In particular, the following technical advantages are achieved: depending on the required recognition accuracy of traffic signs and the road surface undulations, the entire road segment can be divided into multiple unit slope variation intervals by using a segmented expression of slope grading, thus providing a more flexible solution.

[0024] Optionally, the method further includes: acquiring installation specification information and appearance information of traffic signs in different regions or countries; and determining the region of interest based on the installation specification information and appearance information, particularly determining the shape, size, and aspect ratio of the region of interest.

[0025] In particular, the following technical advantages are achieved: compared to directly determining a fixed-size region of interest centered on the mapping location, by considering the installation specifications and appearance information of traffic signs, more control variables can be introduced in the process of determining the region of interest, thereby further improving the probability that the region of interest accurately covers the traffic signs.

[0026] According to a second aspect of the present invention, an apparatus is provided for determining a region of interest of a traffic sign, the apparatus being configured to perform the method according to a first aspect of the present invention, the apparatus comprising:

[0027] A map reading module is configured to read geographical location information of traffic signs that affect the traffic behavior of vehicles from a high-precision map. The geographical location information includes at least the relative ground elevation of the traffic signs. The map reading module is also configured to read terrain information of the road segment from the vehicle to the traffic signs, which is subject to regional expression limitations, from the high-precision map.

[0028] An image acquisition module is configured to acquire an image of the vehicle's surrounding environment, including the traffic sign; and

[0029] A determination module is configured to determine the region of interest of a traffic sign in the image based at least on the geographical location information of the traffic sign and the terrain information.

[0030] According to a third aspect of the present invention, a system for identifying the state of traffic signs is provided, the system comprising:

[0031] The device according to the second aspect of the invention; and

[0032] An image processing device is configured to continuously monitor a identified region of interest and thereby analyze the status of traffic signs. Attached Figure Description

[0033] The invention will now be described in more detail with reference to the accompanying drawings, which will provide a better understanding of its principles, features, and advantages. The drawings include:

[0034] Figure 1 A flowchart of a method for determining a region of interest according to an exemplary embodiment of the present invention is shown;

[0035] Figure 2 A block diagram of a vehicle according to an exemplary embodiment of the present invention is shown;

[0036] Figure 3 A schematic diagram is shown illustrating the determination of the relative height difference between the road surface where a vehicle is located and the road surface where a traffic sign is located in a sloping section using the method according to the present invention.

[0037] Figure 4 A schematic diagram showing a comparison before and after modifying the region of interest using the method according to the present invention is presented.

[0038] Figure 5 A schematic diagram showing the adjustment of the region of interest using the method according to the invention is shown; and

[0039] Figure 6 A schematic diagram showing the adjustment of the region of interest using the method according to the present invention is shown. Detailed Implementation

[0040] To make the technical problems to be solved, the technical solutions, and the beneficial technical effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and several exemplary embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of protection of this invention.

[0041] Figure 1 A flowchart of a method for determining a region of interest according to an exemplary embodiment of the present invention is shown.

[0042] In step S1, the geographic location information of traffic signs affecting vehicle traffic behavior is read from a high-precision map. Here, this geographic location information includes at least the relative ground elevation of the traffic sign. In the context of this invention, traffic signs can be, for example, traffic signs, traffic lights, or other road surface identifiers that can present traffic information and potentially affect vehicle traffic behavior (e.g., current or future traffic behavior). The geographic location information includes, for example, the latitude and longitude coordinates of the traffic sign in the map coordinate system and its relative elevation information, and can particularly exist in the form of point coordinates or polygon bounding box coordinates.

[0043] In step S2, terrain information of the road segment from the vehicle to the traffic sign, which is subject to regional expression limitations, is read from the high-precision map.

[0044] In the context of this invention, "restricted by regional representation" is understood as follows: due to the confidentiality policies of different countries or regions regarding map data, specific map elements must adhere to regional requirements in their representation, and therefore are often characterized in a way that deviates from their actual values ​​or differs from their conventional physical meaning. For example, terrain information subject to regional representation restrictions includes slope grading, particularly in the form of status bit numbers, for road segments. In general, the actual slope value of a road segment, especially between its starting and ending points, can be calculated based on the quotient of the elevation difference between two points and their horizontal distance. However, due to regional representation restrictions, the slope value of a road cannot be directly displayed on a high-precision map; instead, it needs to be displayed in the form of slope grading (i.e., a specific percentage range covering the actual slope value).

[0045] In the context of this invention, terrain information is understood as information that reflects the shape or contour characteristics of a road, particularly road slope information. Additionally, this terrain information may also include road curvature, cross slope angle, vehicle heading, lane connections, road segment length, etc.

[0046] As an example, step S2 may optionally include steps S21 and S22. This depends on factors such as the required environmental perception accuracy of the autonomous vehicle and computational overhead, particularly the segmented representation of road segment slope information read from a high-precision map. Therefore, in step S21, the road segment from the vehicle to the traffic sign can be divided into sub-segments according to predefined slope intervals. Then, in step S22, the corresponding slope classification is read from the high-precision map for each sub-segment.

[0047] In step S3, an image of the vehicle's surrounding environment, including the traffic sign, is acquired. This image may be provided, for example, by means of at least one image detection device, particularly a forward-facing camera, positioned at the front of the vehicle.

[0048] In step S4, the region of interest for the traffic sign is determined in the image based at least on the geographical location information of the traffic sign and the terrain information.

[0049] In the context of this invention, the region of interest (ROI) is understood as a selected search window within the image provided by the image detection device for traffic sign status recognition. Determining the ROI effectively limits the detection range of the image processing device. For example, if high-precision map data is used to determine the approximate location of a traffic sign in the image, the image processing device can focus on detecting and analyzing the traffic sign at that location. This significantly improves the speed and accuracy of image recognition and effectively saves computational resources.

[0050] As an example, step S4 may optionally include steps S41 to S43. In step S41, with the help of appropriate calibration parameters, the geographical location information of the traffic sign can be projected from the map coordinate system to the image coordinate system through coordinate transformation, and a predefined window area is extended outward from the position of the traffic sign in the image coordinate system as the region of interest. In step S42, the height difference information between the road surface where the vehicle is located and the road surface where the traffic sign is located can be calculated based on the read terrain information, and the position and / or geometric characteristics of the region of interest in the image can be corrected based on the height difference information. In step S43, the shape, size, and aspect ratio of the region of interest can also be further determined based on installation specification information and appearance information to improve the determination of the region of interest.

[0051] Figure 2 A block diagram of a vehicle according to an exemplary embodiment of the present invention is shown.

[0052] Vehicle 1 is, for example, a vehicle that is at least partially autonomous and performs driver assistance functions based on traffic sign status recognition. Exemplarily, vehicle 1 includes a high-precision map module 10 to acquire geographic location information of the vehicle itself and other traffic signs. A camera 30 is also mounted at the front of vehicle 1, for example, to capture real-time images of the vehicle's surrounding environment. Furthermore, vehicle 1 includes a system 20 for recognizing the status of traffic signs, which further includes a device 21 for determining the region of interest of the traffic sign and an image processing device 22.

[0053] Here, device 21 is configured to perform Figure 1 The method described therein. Accordingly, the device 21 includes a map reading module 211 for interacting with the high-precision map module 10, an image acquisition module 212 for communicating with the camera 30, and a region of interest determination module 213. By executing... Figure 1 The corresponding method steps of the illustrated method involve using the determination module 213 of device 21 to mark the region of interest in the image provided by camera 30. Next, the relevant information of the region of interest can be transmitted to image processing device 22, which can then continuously monitor the determined region of interest and analyze the state of traffic signs therein. Exemplarily, the state of traffic signs may include the signal status of traffic lights (go / stop) or the content of traffic signs (e.g., information on traffic-controlled sections, congestion information, distance to ramp exits, etc.).

[0054] Finally, the status of the identified traffic signs can be provided to the control equipment 40 of vehicle 1 to control the actuators of vehicle 1 to take corresponding driver assistance functions.

[0055] Figure 3 A schematic diagram is shown illustrating the determination of the relative height difference between the road surface where a vehicle is located and the road surface where a traffic sign is located in a sloping section using the method according to the present invention.

[0056] like Figure 3 As shown, vehicle 1 is autonomously driving on road segment L using the automatic driving function. Looking along the vehicle's direction of travel, traffic light 2 is located not far ahead. When vehicle 1 enters the influence range 100 of traffic light 2, that is, when vehicle 1's driving behavior may be restricted by the signal state of traffic light 2, vehicle 1 can identify the state of traffic light 2 to determine whether it can continue or needs to take braking or other strategies.

[0057] First, vehicle 1 reads the information from the high-precision map and obtains the geographical location information of traffic light 2 on the high-precision map. Here, this geographical location information is exemplarily expressed in the form of the coordinates of the vertices of the bounding box of traffic light 2 (x1, y1, |z1|), (x2, y2, |z2|), and (x3, y3, |z3|). The height information |z1|, |z2|, and |z3| do not directly display the absolute elevation information of each vertex of the bounding box, but rather represent the relative height with the ground plane0 at the location of traffic light 2 as the reference plane.

[0058] Depend on Figure 3 As can be seen, road segment L exhibits significant variations in elevation. Therefore, for vehicle 1, the relative elevation of traffic light 2 obtained from the high-precision map cannot be universally applied to all possible ground planes (plane1, plane2, plane3, etc.) within road segment L. Consequently, when traffic light 2 is mapped onto an image coordinate system referenced to the plane 1 where vehicle 1 is located, based on this relative elevation information, the mapped position deviates from the actual position of traffic light 2 in the image coordinate system. Therefore, if the region of interest 3 is extended solely based on the position mapped in the image using the relative elevation information of traffic light 2, this region of interest may not cover the actual landmark to be identified, thus leading to the inability to correctly detect the traffic light signal.

[0059] To avoid the problem of traffic light 2 being unrecognized on road sections with steep inclines, vehicle 1 continues to rely on... Figure 2 The device 21 shown reads terrain information for road segment L from a high-precision map. Here, for example, a segmented representation of the slope information of road segment L can be read; Table 1 exemplarily shows a list of slope grades identified in the form of status bits. Table 2 correspondingly shows the slope grades for sub-segments L1, L2, and L3 read from Table 1, each slope grade representing a range of slope values ​​for a sub-segment L1, L2, and L3.

[0060] Table 1

[0061]

[0062] Table 2

[0063]

[0064] like Figure 3 As shown, road segment L encompasses all road segments the local path planning involves, from the vehicle's current location to the traffic light. The endpoints of each sub-segment L1, L2, and L3 are connected to the starting point of the next sub-segment, thus forming a continuous road segment L. Next, the elevation difference between the endpoint and starting point of each sub-segment can be calculated based on the obtained slope classification using the following formula:

[0065]

[0066]

[0067] Among them, h i_min For one of the sub-segments L of road segment L i The lowest possible elevation difference, h, between the starting point and the ending point i_max For one of the sub-segments L of road segment L i The highest possible elevation difference between the starting point and the ending point, Level i_min For one of the sub-segments L of road segment L i The lowest value of the slope grading threshold, Level i_max For one of the sub-segments L of road segment L i The high value of the grading threshold for slope classification, where i is the sub-segment number of L, i=1,2..3.

[0068] For each sub-segment L i After obtaining the corresponding height difference range, the height difference between the road surface where vehicle 1 is located and the road surface where traffic light 2 is located can be calculated in general using the following formula, that is, the height difference range between the start and end points of road segment L can be calculated:

[0069]

[0070]

[0071] Among them, H min H is the lowest possible elevation difference between the start and end points of road segment L. max Let L be the maximum possible elevation difference between the start and end points of road segment L.

[0072] Here, the actual height difference H between road surface plane1 where vehicle 1 is located and road surface plane0 where traffic light 2 is located satisfies the following relationship:

[0073]

[0074] Based on the determined height difference range (H) of road segment L min H max This allows for the correction of the position and / or geometric characteristics of the region of interest (ROI) 3 of the traffic light 2 in the image coordinate system. As an example, a scaling factor, a function of the height difference range, can be introduced along the longitudinal extension of the ROI 3, based on the originally determined size of the ROI 3. As another example, an offset, a function of the height difference range, can be introduced along the height of the ROI 3, based on the originally determined position of the ROI 3. This advantageously takes into account the influence of the terrain factors of road segment L in the determination of the ROI.

[0075] Figure 4 A schematic diagram showing a comparison before and after modifying the region of interest using the method according to the present invention is presented.

[0076] Figure 4 The left side shows schematic diagrams of uncorrected regions of interest (ROIs) 3 for road segments L with different slopes. It can be seen that since the relative height information |z| of traffic light 2 relative to the road surface where traffic light 2 is located is obtained directly from the high-precision map, the determined ROI 3 will not deviate from the actual position of traffic light 2 in the image from the vehicle-mounted camera only when the road segment L from vehicle 1 to traffic light 2 is flat (slope information is negligible). However, if there is a significant height difference between the location of vehicle 1 and traffic light 2, this relative height information |z| of traffic light 2 cannot be well adapted to the entire road segment L, especially not to the road surface where vehicle 1 is currently located. Therefore, if this relative height information |z| is directly superimposed on the current location of vehicle 1 and the ROI 3 is determined based on this, the resulting ROI 3 cannot accurately cover the actual position of traffic light 2.

[0077] Figure 4 The right side shows schematic diagrams of regions of interest 3 after correction using the method according to the invention, for road segments L with different slopes. Here, by taking into account the terrain information of the road segment L between vehicle 1 and traffic light 2 when determining region of interest 3, for example, the determined region of interest 3 can be stretched longitudinally to accommodate different slopes, thereby enabling region of interest 3 to completely surround traffic light 2, thus achieving improved recognition results.

[0078] Figure 5A schematic diagram showing the adjustment of the region of interest using the method according to the present invention is shown.

[0079] Figure 5 The two diagrams at the top illustrate how the shape characteristics of the area of ​​interest can be adjusted for different installation specifications of traffic signs. Figure 5 The top left figure exemplarily shows the bounding box 201 projected onto the image of a horizontally mounted traffic light, and the extended region of interest 301 taking into account this mounting specification. Figure 5 The upper right figure shows the bounding box 202 projected onto the image of a vertically mounted traffic light. It also shows the extended region of interest 302 considering this installation specification. It can be seen that because the installation specifications and appearance information of the traffic light are taken into account when determining the regions of interest 301 and 302, the contours and shapes of the regions of interest 301 and 302 can dynamically adapt to the area of ​​the traffic light in the image. Therefore, it effectively avoids situations where the region of interest cannot completely cover the object to be detected.

[0080] Figure 5 The two lower illustrations show schematic diagrams illustrating the longitudinal scaling of the region of interest for terrain adjustment from vehicle to traffic light. Comparing them respectively... Figure 5 The upper illustration shows that because the geographical location information of traffic signs read from high-precision maps only represents the relative height of the traffic signs, the road slope was not taken into account during coordinate transformation and projection. This causes the bounding boxes 201' and 202' projected onto the corresponding image of the traffic lights to deviate from their actual positions in the image. In this case, it can be addressed as follows: Figure 1 The method described above calculates a corresponding correction factor based on terrain information (such as slope grading) limited by regional representation, in order to adjust for the vertical scaling of the region of interest. Thus, even if the projected bounding boxes 201' and 202' deviate from the actual location of the traffic lights, this adjustment can still ensure that the regions of interest 301' and 302' cover the traffic lights in the image, thereby reducing the risk of misidentification.

[0081] Figure 6 A schematic diagram illustrating the adjustment of the region of interest using the method according to the present invention is shown. Figure 6 The image shows a type of traffic light that is rarely seen in typical traffic scenarios. This traffic light has two indicator lights arranged side by side, one above the other. The larger upper indicator light is used, for example, to direct traffic flow for motor vehicles, while the smaller lower indicator light is used, for example, to indicate traffic flow for pedestrians or other non-motorized vehicles.

[0082] like Figure 6As shown on the left, if the appearance information or installation specifications of the traffic lights are not taken into account when determining the regions of interest 501 and 502, even if the deviations that occur during the traffic light mapping process are compensated for according to the terrain information, the following unreasonable delineation of regions of interest 501 and 502 may still occur: Because regions of interest 501 and 502 are lengthened vertically based on the terrain information, the region of interest 501 of the upper indicator light incorrectly encompasses the lower indicator light. This leads to errors in the subsequent traffic light status recognition process due to the simultaneous existence of two status recognition results.

[0083] exist Figure 6 The right side illustrates an exemplary solution. This is achieved, for example, by setting a safety margin for the longitudinal extension of the regions of interest 501' and 502' corrected based on terrain information. Exemplarily, if the lower edge height of the region of interest 501' of the upper indicator light is determined to be H1_down using the method of the present invention, and the upper edge height of the region of interest 502' of the lower indicator light is simultaneously determined to be H2_up, then to avoid… Figure 6 The unreasonable situation shown on the left also involves a comparison of H1_down and H2_up. Here, if it is determined that H1_down < H2_up, the lower edge height of the region of interest 501' of the upper indicator light and the upper edge height of the region of interest 502' of the lower indicator light are automatically adjusted to satisfy: H1_down = H2_up = 1 / 2(H1_down + H2_up).

[0084] This not only allows us to take the slope effect into account, but also helps to set a reasonable range for the correction amount, avoiding other errors caused by over-correction.

[0085] Although specific embodiments of the invention have been described in detail herein, they are given for illustrative purposes only and should not be construed as limiting the scope of the invention. Various substitutions, alterations, and modifications can be conceived without departing from the spirit and scope of the invention.

Claims

1. A method for determining a region of interest (3) of a traffic sign (2), the method comprising the steps of: S1: Read the geographic location information of traffic signs (2) that affect the traffic behavior of vehicles (1) from a high-precision map. The geographic location information includes at least the relative ground height (|z|) of the traffic signs (2). In the high-precision map, absolute elevation information and actual slope values ​​of road segments are not expressed. S2: Read the terrain information of the road segment (L) from the vehicle (1) to the traffic sign (2) from the high-precision map, which is subject to regional expression limitation. The terrain information subject to regional expression limitation includes: the slope value range in the form of the status bit number of the road segment and / or the slope classification corresponding to the slope value range of the road segment, which covers the actual slope value of the road segment. S3: Acquire an image of the vehicle's surrounding environment, including the traffic sign (2); and S4: Based at least on the geographic location information of the traffic sign (2) and the terrain information, determine the region of interest (3) of the traffic sign (2) in the image. Specifically, the original region of interest for the traffic sign is determined in the image based on the geographical location information of the traffic sign. The height difference range between the road surface where the vehicle is located and the road surface where the traffic sign is located is calculated using the slope value range and / or slope gradation. The originally determined region of interest is stretched longitudinally using a scaling factor that is a function of the height difference range.

2. The method according to claim 1, wherein, Step S1 includes: reading the latitude and longitude coordinates and relative ground elevation (|z|) of the traffic sign (2) from a high-precision map, wherein the coordinates are expressed as point coordinates (x, y, |z|) and / or as polygon bounding box coordinates (x1, y1, |z1|) - (x n , y n , |z n The location information of traffic signs (2) is read in the form of |).

3. The method according to claim 1 or 2, wherein, The slope value range is a percentage range.

4. The method according to claim 1 or 2, wherein, Step S2 includes: reading the segmented representation of the terrain information of the road segment (L) from the high-precision map, wherein, in the case of segmented representation, corresponding terrain information is assigned to each sub-segment (L1, L2, L3) of the road segment (L).

5. The method according to claim 3, wherein, Step S2 includes: reading the segmented representation of the terrain information of the road segment (L) from the high-precision map, wherein, in the case of segmented representation, corresponding terrain information is assigned to each sub-segment (L1, L2, L3) of the road segment (L).

6. The method according to any one of claims 1, 2, and 5, wherein, Step S4 includes: projecting the geographic location information of the traffic sign (2) from the map coordinate system of the high-precision map to the image coordinate system of the image, expanding a window area of ​​a predefined size outward based on the position of the traffic sign (2) in the image coordinate system as the region of interest (3); and correcting the region of interest (3) in terms of position and / or geometric characteristics according to the read terrain information.

7. The method according to claim 3, wherein, Step S4 includes: projecting the geographic location information of the traffic sign (2) from the map coordinate system of the high-precision map to the image coordinate system of the image, expanding a window area of ​​a predefined size outward based on the position of the traffic sign (2) in the image coordinate system as the region of interest (3); and correcting the region of interest (3) in terms of position and / or geometric characteristics according to the read terrain information.

8. The method according to claim 4, wherein, Step S4 includes: projecting the geographic location information of the traffic sign (2) from the map coordinate system of the high-precision map to the image coordinate system of the image, expanding a window area of ​​a predefined size outward based on the position of the traffic sign (2) in the image coordinate system as the region of interest (3); and correcting the region of interest (3) in terms of position and / or geometric characteristics according to the read terrain information.

9. The method according to claim 6, wherein, The elevation difference (H) between the road surface where the vehicle (1) is located and the road surface where the traffic sign (2) is located is calculated using the terrain information; the elevation of the region of interest (3) in the image is adjusted and / or the scaling ratio of the region of interest is adjusted based on the elevation difference (H).

10. The method of claim 9, wherein, In order to calculate the height difference information (H) between the road surface where the vehicle (1) is located and the road surface where the traffic sign (2) is located, the road segment (L) from the vehicle (1) to the traffic sign (2) is divided into multiple sub-segments (L1, L2, L3) according to the terrain information. For each sub-segment (L1, L2, L3), the height difference information (h1, h2, h3) between the starting point and the ending point is calculated. The height difference information (h1, h2, h3) of each sub-segment (L1, L2, L3) is accumulated to form the height difference information (H) of the road segment (L).

11. The method according to any one of claims 1, 2, 5, 7, 8, 9, and 10, wherein, The method further includes: Obtain installation specifications and appearance information for traffic signs (2) in different regions or countries; and Based on the installation specifications and appearance information, the region of interest is determined (3).

12. The method according to claim 9, wherein, Based on the height difference information (H), adjust the height of the region of interest (3) in the image and / or adjust the vertical scaling factor of the region of interest.

13. The method according to claim 11, wherein, Based on the installation specifications and appearance information, determine the shape, size, and aspect ratio of the region of interest (3).

14. An apparatus (21) for determining a region of interest (3) of a traffic sign, the apparatus (21) being used to perform the method according to any one of claims 1 to 13, the apparatus (21) comprising: The map reading module (211) is configured to read the geographic location information of traffic signs (2) that affect the traffic behavior of vehicles (1) from a high-precision map. The geographic location information includes at least the relative ground height (|z|) of the traffic signs. The map reading module (211) is also configured to read the terrain information of the road segment (L) from the vehicle (1) to the traffic signs (2) that is restricted by regional expression from the high-precision map. The image acquisition module (212) is configured to acquire an image of the vehicle’s surrounding environment, including the traffic sign (2); as well as The determination module (213) is configured to determine the region of interest (3) of the traffic sign (2) in the image based at least on the geographic location information of the traffic sign (2) and the terrain information.

15. A system (20) for identifying the state of a traffic sign (2), the system (20) comprising: The device (21) according to claim 14; as well as Image processing device (22) is configured to analyze and process the determined region of interest (3) and identify the state of the traffic sign (2).

Images