A turning blind area detection method and device

Abstract

The embodiment of the application discloses a kind of corner blind area detection method and device.The method comprises: through the image collector of vehicle that is arranged on the target vehicle that is driven on target road, image acquisition is carried out to the front of vehicle driving, and the convex mirror area is determined by identifying the collected image;The position of the vehicle radar arranged on the target vehicle is determined, and the emission direction of the detection light emitted by the vehicle radar to the convex mirror area is determined;According to the position of the vehicle radar, the emission direction of the detection light and the size information of the convex mirror, the reflection point information of the target object reflected by the detection light passing through the convex mirror is determined, and the three-dimensional point cloud model of the target object is established according to the reflection point information.The above-mentioned scheme can make the detection light of vehicle radar be reflected by the convex mirror, automatically detect the blind area caused by physical obstruction at the corner, and solve the problem that the convex mirror is greatly affected by ambient light by relying on human eye observation.

Description

Technical Field

[0001] This application relates to the field of road traffic safety technology, and in particular to a method and device for detecting blind spots when turning. Background Technology

[0002] In modern cities, parking lots and underground garages are being used more and more frequently. However, the environment inside underground garages is complex and changeable, especially at corners where blind spots often exist, increasing the risk of vehicle collisions. When vehicles meet at corners, drivers have difficulty accurately judging the position and speed of oncoming vehicles due to obstructed vision, leading to frequent collisions.

[0003] Traditional solutions mainly include increasing lighting, setting up warning signs, and installing surveillance cameras. However, these methods have the following limitations: Insufficient lighting: Increasing lighting can improve the visual environment, but it cannot eliminate physical obstructions. Warning signs: Warning signs only serve as a reminder and cannot provide real-time vehicle location information. Surveillance cameras: Although they can record the accident scene, they cannot provide the information needed by the driver in real time, and their installation and maintenance costs are high. Convex traffic mirrors: They are greatly affected by angle, lighting, and the driver's subjective perception; if not careful, they can still easily cause accidents. Summary of the Invention

[0004] This application provides a method and apparatus for detecting blind spots at turns, which can automatically detect blind spots at turns and is not easily affected by ambient light.

[0005] According to one aspect of this application, a method for detecting blind spots at turns is provided, wherein a convex mirror is disposed at the turn of a first road and a second road, the convex mirror being disposed at the outer vertex of the turn and facing the inner side of the turn; the method includes:

[0006] The onboard image acquisition device installed on the target vehicle traveling on the target road acquires images of the area in front of the vehicle, and identifies the convex mirror region from the acquired images; wherein the target road is either the first road or the second road;

[0007] The location of the vehicle-mounted radar installed on the target vehicle is determined, and the emission direction of the detection light emitted by the vehicle-mounted radar toward the convex mirror area is determined;

[0008] Based on the location of the vehicle-mounted radar, the emission direction of the detection light beam, and the size information of the convex mirror, the reflection point information of the detection light beam after being reflected by the convex mirror and detecting the target object and being reflected by the target object is determined, and a three-dimensional point cloud model of the target object is established based on the reflection point information.

[0009] According to one aspect of this application, a turning blind spot detection device is provided, the device comprising:

[0010] The convex mirror region determination module is used to acquire images of the area in front of the vehicle by means of an on-board image acquisition device installed on the target vehicle traveling on the target road, and to identify and determine the convex mirror region from the acquired images; wherein the target road is either the first road or the second road.

[0011] The emission direction determination module is used to determine the position of the vehicle-mounted radar installed on the target vehicle, and to determine the emission direction of the detection light emitted by the vehicle-mounted radar toward the convex mirror area;

[0012] The three-dimensional point cloud model determination module is used to determine the reflection point information of the detection light after being reflected by the convex mirror and detected by the target object, based on the position of the vehicle radar, the emission direction of the detection light and the size information of the convex mirror, and to establish a three-dimensional point cloud model of the target object based on the reflection point information.

[0013] According to another aspect of this application, an electronic device is provided, the electronic device comprising:

[0014] At least one processor; and

[0015] A memory that is communicatively connected to at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by at least one processor, such that the at least one processor is able to perform the turning blind spot detection method of any embodiment of the present application.

[0017] According to another aspect of this application, a computer-readable storage medium is provided, which stores computer instructions for causing a processor to execute and implement the turning blind spot detection method of any embodiment of this application.

[0018] The technical solution of this application embodiment involves acquiring images of the area in front of a target vehicle traveling on a target road using an onboard image acquisition device. The acquired images are then used to identify and determine the convex mirror region. The target road is either the first road or the second road. The position of the onboard radar installed on the target vehicle is determined, as is the emission direction of the detection light emitted by the radar towards the convex mirror region. Based on the position of the onboard radar, the emission direction of the detection light, and the size information of the convex mirror, the reflection point information of the detection light after reflection by the convex mirror is determined, indicating that the detection light has detected the target object and is reflected by the target object. A three-dimensional point cloud model of the target object is then established based on the reflection point information. In this solution, the detection light from the onboard radar, after reflection by the convex mirror, detects blind spots at turns, enabling automatic detection of blind spots caused by physical obstructions. Furthermore, it is less affected by ambient light, improving the accuracy and reliability of blind spot detection.

[0019] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent from the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart illustrating a blind spot detection method for turning, provided in an embodiment of this application;

[0022] Figure 2 A schematic diagram of a turning point provided in an embodiment of this application;

[0023] Figure 3 This is a first schematic diagram of vehicle-mounted radar detection provided in an embodiment of this application;

[0024] Figure 4 This is a second schematic diagram of vehicle-mounted radar detection provided in an embodiment of this application;

[0025] Figure 5 A flowchart of a turning blind spot detection method is provided in another embodiment of this application;

[0026] Figure 6 A flowchart of a turning blind spot detection method provided in another embodiment of this application;

[0027] Figure 7 This is a schematic diagram of the structure of a turning blind spot detection device provided in an embodiment of this application;

[0028] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0029] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0030] It should be noted that the terms "first," "second," "third," "fourth," "actual," "preset," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0031] Figure 1 This is a flowchart illustrating a turning blind spot detection method provided in an embodiment of this application. This embodiment is applicable to situations requiring blind spot detection at turns. The method can be executed by a turning blind spot detection device, which can be implemented in hardware and / or software and can be configured in an electronic device. Figure 1 As shown, the method includes:

[0032] S110. The vehicle-mounted image acquisition device installed on the target vehicle traveling on the target road acquires images of the area in front of the vehicle, and identifies the convex mirror area by recognizing the acquired images; wherein, the target road is either the first road or the second road.

[0033] The first road and the second road can be two roads that intersect. At the intersection of the first road and the second road, there is a bend. A convex mirror is installed at the bend, positioned at the outer apex of the bend and facing inwards towards the bend. Figure 2 As shown. The target road can be any road between the first road and the second road. That is, the target vehicle can be a vehicle traveling from the first road to the second road via a turn, or a vehicle traveling from the second road to the first road via a turn. The onboard image acquisition device on the target vehicle can be a dashcam or other device capable of capturing images of the area ahead of the vehicle. The onboard image acquisition device can capture images of the area ahead of the vehicle at a preset frequency, obtain the captured images, and transmit the captured images to the blind spot detection device.

[0034] In this embodiment, after the target vehicle acquires images via an onboard image acquisition device during its operation, the acquired images are identified to determine the convex mirror region. Specifically, the acquired images can be converted to grayscale to reduce computational complexity. Gaussian blurring can be applied to reduce noise. Edge detection is then performed on the preprocessed acquired images to obtain the edges. The edges are classified to identify straight edges and curved edges. The edges are matched with pre-defined convex mirror edge features to detect convex mirrors in the image. Alternatively, a machine learning model or deep learning model can be used to identify convex mirrors in the acquired images. After identifying the convex mirrors in the acquired images, based on their position and size in the acquired images, combined with the position of the onboard image acquisition device and image acquisition parameters, the convex mirror region in the actual scene is determined, including the position and size of the convex mirror.

[0035] S120. Determine the position of the vehicle-mounted radar installed on the target vehicle, and determine the emission direction of the detection light emitted by the vehicle-mounted radar toward the convex mirror area.

[0036] The target vehicle is equipped with vehicle-mounted radar, which can emit detection beams to detect surrounding targets.

[0037] For example, the position of the vehicle-mounted radar can be determined based on the positioning device on the target vehicle. Since the direction of the detection light emitted by the vehicle-mounted radar is known, and the convex mirror area is known, the emission direction of the detection light emitted by the vehicle-mounted radar towards the convex mirror area can be determined. Specifically, the vehicle-mounted radar can emit detection light towards the front of the target vehicle, and the detection light can form a cone-shaped light beam with the vehicle-mounted radar as the apex, such as... Figure 3As shown, the cone-shaped light beam has portions that are emitted into the convex mirror area and portions that are emitted outside the convex mirror. The portion emitted outside the convex mirror directly detects targets ahead of the vehicle, while the portion emitted into the convex mirror area can detect targets on another road after a turn from the target road. Based on the direction of the cone-shaped light beam and the convex mirror area, the emission direction of the detection light emitted into the convex mirror area can be determined.

[0038] In this embodiment of the application, after identifying and determining the convex mirror region from the acquired image, the method further includes:

[0039] Based on the vehicle-mounted radar emitting detection light beams towards the front of the vehicle at initial power and initial frequency, the power and / or frequency of the detection light beams emitted by the vehicle-mounted radar towards the convex mirror area are increased.

[0040] For example, in the initial state, the vehicle-mounted radar emits a detection beam at an initial power and frequency towards the direction in front of the vehicle to detect objects within its field of vision. In this case, targets within the visible range can already be detected. After identifying the convex mirror area, enhanced detection beams can be selectively emitted towards that area to improve detection of blind spots that are only visible through reflection from the convex mirror. Specifically, based on the initial power and frequency of the detection beams emitted by the vehicle-mounted radar towards the convex mirror area, the power and / or frequency of the detection beams can be increased. This allows for enhanced detection of blind spots through reflection from the convex mirror using higher-energy and / or higher-frequency detection beams, improving the detection strength and accuracy of blind spots.

[0041] S130. Based on the position of the vehicle-mounted radar, the emission direction of the detection light beam, and the size information of the convex mirror, determine the reflection point information of the detection light beam after being reflected by the convex mirror, where the target object is detected and reflected by the target object, and establish a three-dimensional point cloud model of the target object based on the reflection point information.

[0042] The dimensional information of the convex mirror can be pre-stored information about the convex mirror at that location. This dimensional information may include, for example, the center and radius of curvature of the convex mirror.

[0043] For example, such as Figure 4As shown, assuming the target road is the first road, the detection light emitted by the vehicle-mounted radar is reflected by a convex mirror and illuminates the second road. If a target object exists on the second road, the detection light will be reflected by the target object, illuminate the convex mirror, and then reflected again by the convex mirror before being received by the vehicle-mounted radar. The optical path is reversible. The angle of incidence of the detection light on the convex mirror is equal to the angle of reflection. Based on the position of the vehicle-mounted radar, the emission direction of the detection light emitted towards the convex mirror, and the size of the convex mirror, the reflection point information of the detection light reflected by the target object can be calculated. The reflection point information reflects the position information of the reflection point of the detection light reflected by the target object. This reflection point is a point on the surface of the target object, representing a point on the target object. A three-dimensional point cloud model of the target object can be built based on the reflection point information, reflecting the position of the target object. The three-dimensional point cloud model of the target object can be presented to the driver in the target vehicle, so that the driver can clearly understand the situation of blind spots at turns and adjust the driving status of the target vehicle in time.

[0044] In this embodiment, only the process of establishing a three-dimensional point cloud model of the target object detected by the detection light after reflection by the convex mirror is specifically described. In fact, the vehicle radar will also emit detection light to areas outside the convex mirror area, which can directly detect the target object within the detectable range and establish a three-dimensional point cloud model of the directly detected target object. The two three-dimensional point cloud models can be merged to automatically obtain a three-dimensional point cloud model of the target object within the visible range and a three-dimensional point cloud model of the target object in the blind spot at the turning point, which are fed back to the driver of the target vehicle for driving reference.

[0045] The process of constructing a 3D point cloud model of the target object involves selecting a suitable data structure to store the reflection point information, such as using a NumPy array, and integrating the reflection point information into the data structure. The reflection point information is then processed, including filtering, noise reduction, and deduplication. Finally, Open3D is used for point cloud visualization to form a 3D point cloud model of the target object.

[0046] The technical solution of this application embodiment involves acquiring images of the area in front of a target vehicle traveling on a target road using an onboard image acquisition device. The acquired images are then used to identify and determine the convex mirror region. The target road is either the first road or the second road. The position of the onboard radar installed on the target vehicle is determined, as is the emission direction of the detection light emitted by the radar towards the convex mirror region. Based on the position of the onboard radar, the emission direction of the detection light, and the size information of the convex mirror, the reflection point information of the detection light after reflection by the convex mirror is determined, indicating that the detection light has detected the target object and is reflected by the target object. A three-dimensional point cloud model of the target object is then established based on the reflection point information. In this solution, the detection light from the onboard radar, after reflection by the convex mirror, detects blind spots at turns, enabling automatic detection of blind spots caused by physical obstructions. Furthermore, it is less affected by ambient light, improving the accuracy and reliability of blind spot detection.

[0047] Figure 5 This is a flowchart illustrating a blind spot detection method for turning, provided as another embodiment of this application. This embodiment is an optimization based on the above embodiment; solutions not described in detail in this embodiment are found in the above embodiment. Figure 5 As shown, the method in this embodiment of the application specifically includes the following steps:

[0048] S210. The vehicle-mounted image acquisition device installed on the target vehicle traveling on the target road acquires images of the area in front of the vehicle, and identifies the convex mirror area by recognizing the acquired images; wherein, the target road is either the first road or the second road.

[0049] S220. Determine the position of the vehicle-mounted radar installed on the target vehicle, and determine the emission direction of the detection light emitted by the vehicle-mounted radar toward the convex mirror area.

[0050] S230. Based on the position of the vehicle-mounted radar and the emission direction of the detection light, the equation of the detection light is expressed to obtain the first equation.

[0051] For example, the equation for the detection ray emitted by the vehicle-mounted radar can be expressed based on the position of the vehicle-mounted radar and the emission direction of the detection ray, resulting in the first equation. For instance, assuming the position of the lidar is P... Lidar =(x Lidar ,y Lidar ,z Lidar The direction of the detection light beam is d. Lidar =(d xLidar ,d yLidar ,d zLidar If the first equation for the probe ray is: P(t) = PLidar +t*d Lidar ; where t is a parameter representing the distance the probe light extends along the emission direction.

[0052] S240. Based on the size information of the convex mirror, the equation of the convex mirror is expressed to obtain the second equation.

[0053] For example, the equation of a convex mirror can be expressed based on the mirror's dimensions. The convex mirror can be approximated as a sphere, and its equation can be expressed using the equation of a sphere. For instance, suppose the mirror's dimensions include the center C of its surface. Mirror =(x Mirror ,y Mirror ,z Mirror If the radius of curvature R of the convex mirror is given, then the second equation can be expressed as (xx Mirror ) 2 +(yy Mirror ) 2 +(zz Mirror ) 2 =R 2 , where P I ntersection=(x,y,z) represents the intersection point of the convex mirror and the probe ray.

[0054] S250. Determine the intersection information of the probe ray and the convex mirror according to the first equation and the second equation.

[0055] For example, the intersection point information of the probe ray and the convex mirror can be determined based on the first and second equations. The first equation can be substituted into the second equation to obtain the unknown quantity t, and then P can be solved. Intersection Obtain the intersection point information P of the probe ray and the convex mirror. Intersection Substituting the first equation into the second equation, we get:

[0056] (x Lidar +t*d xLidar -x Mirror ) 2 +(y Lidar +t*d yLidar -

[0057] y Mirror ) 2 +(z Lidar +t*d zLidar -z Mirror ) 2 =R 2 ;

[0058] Solving for t yields in b = 2*(d) xLidar (x Lidar -x Mirror )+d yLidar (y Lidar -y Mirror )+d zLidar (z Lidar -z Mirror )); c = (x Lidar -x Mirror ) 2 +(y Lidar -y Mirror ) 2 +(z Lidar -z Mirror ) 2 -R 2 Solving for t and the intersection point P yields the answer. Intersection .

[0059] S260. Determine the normal vector at the intersection point based on the intersection point information and the size information of the convex mirror.

[0060] For example, the normal vector at the intersection point can be determined based on the intersection point information and the size information of the convex mirror, so as to determine the trajectory of the probe light after reflection by the convex mirror according to the reflection principle of the convex mirror. The vector from the intersection point to the center of the convex mirror surface is... Normal vector is for The length of the module.

[0061] S270. Based on the normal vector and the emission direction of the probe ray, determine the reflection point information, and establish a three-dimensional point cloud model of the target object based on the reflection point information.

[0062] For example, the reflection point information can be determined based on the normal vector and the emission direction of the probe ray. Specifically, the reflection direction can be represented as: d Reflection =d Lidar -2*(d Lidar *d Normal )*d Normal

[0063] Where d Lidar *d Normal It is the dot product of the emission direction of the detection beam from the vehicle-mounted radar and the normal vector. Assuming the distance between the reflection point and the intersection point is t', then the reflection point P... Reflected =P Intersection +t′×d Reflection Furthermore, a three-dimensional point cloud model of the target object is established based on the reflection location information.

[0064] In this embodiment of the application, establishing a three-dimensional point cloud model of the target object based on the reflection point information includes:

[0065] The azimuth and elevation angles of the reflection point relative to the vehicle-mounted radar are determined based on the reflection point information.

[0066] Based on the azimuth angle, the elevation angle, and the target distance of the reflection point relative to the vehicle-mounted radar, a three-dimensional point cloud model of the target object relative to the target vehicle is established.

[0067] For example, the position of the vehicle-mounted radar reflects the position of the target vehicle. A three-dimensional point cloud model of the target object relative to the target vehicle can be established based on the position of the reflection point relative to the vehicle-mounted radar, thereby clarifying the relative positional relationship between the target object and the target vehicle. The azimuth and elevation angles of the reflection point relative to the vehicle-mounted radar can be determined based on the reflection point information. Based on the azimuth, elevation angles, and the target distance of the reflection point relative to the vehicle-mounted radar, a three-dimensional point cloud model of the target object relative to the target vehicle can be established. The azimuth angle can be expressed as θ = arctan2(P yReflected -P yLidar P xReflected -P xLidar The pitch angle can be expressed as:

[0068]

[0069] In this embodiment of the application, the process of determining the target distance includes:

[0070] Based on the round-trip time between the vehicle-mounted radar emitting a detection beam and receiving a reflected beam, and the propagation speed of the emitted beam, the sum of the first distance from the vehicle-mounted radar to the intersection point and the second distance from the intersection point to the reflection point is determined; wherein, the intersection point is the intersection point of the detection beam and the convex mirror;

[0071] Based on the location of the vehicle-mounted radar and the location of the intersection, a first distance from the vehicle-mounted radar to the intersection is determined, and a second distance is determined based on the first distance and the location of the intersection.

[0072] The target distance is determined based on the first distance, the second distance, and the reflection angle of the probe light on the convex mirror.

[0073] For example, the target distance is the distance between the intersection point and the reflection point. The sum of the first distance from the vehicle radar to the intersection point and the second distance from the intersection point to the transmission point can be determined based on the round-trip time from the vehicle radar emitting the detection light to receiving it, and the propagation speed of the emitted light. Assuming the propagation speed of the detection light is c and the round-trip time is T, then... Where, dall Let d be the distance sum. all = d1 + d2. d1 is the first distance, and d2 is the second distance. The first distance d1 can be calculated based on the location of the vehicle-mounted radar and the location of the intersection point. Therefore, the second distance d2 can be solved based on the above formula. The angle between the incident and reflected rays of the detection ray and the reflected ray can be determined based on the incident and reflected ray directions of the convex mirror. Using the first distance, the second distance, and the angle, the distance between the vehicle-mounted radar and the reflection point, i.e., the distance between the target vehicle and the target object, can be calculated using the law of cosines.

[0074] In this embodiment of the application, establishing a three-dimensional point cloud model of the target object based on the reflection point information includes:

[0075] Based on road network information or road information in images captured by vehicle-mounted image acquisition devices, determine the traffic range on another road; wherein, the other road is a road other than the target road between the first road and the second road.

[0076] The reflection points detected by the vehicle-mounted radar that are located within the passage range are identified, and a three-dimensional point cloud model of the target object is established based on the reflection point information of the reflection points located within the passage range.

[0077] For example, in the area detected by the probe light reflected by the convex mirror, there may be barriers such as curves or boundary barriers that do not belong to the target objects on the first or second road. Therefore, these barriers can be removed, and a 3D point cloud model can be built for the target objects located on the other road. The traffic range on the other road can be determined based on road network information or road information reflected in images captured by the vehicle-mounted image acquisition device; that is, the range excluding barriers, boundary barriers, and other areas used only for vehicle passage. For the reflection points detected by the vehicle-mounted radar, the traffic range of the other road can be used to filter and determine the reflection points within that range. A 3D point cloud model of the target object can be built based on the reflection point information of these points, thereby accurately removing target objects that do not belong to the other road and accurately reflecting target objects located on the other road that can affect the movement of target vehicles.

[0078] This application provides a method for detecting blind spots during turns. Based on the position of the vehicle-mounted radar and the direction of the detection light emission, the detection light is represented by the first equation. The convex mirror is represented by the second equation based on its size information. The intersection point information of the detection light and the convex mirror is determined based on the first and second equations, and the normal vector at the intersection point is determined. Then, based on the reflection principle of the convex mirror on the detection light, the reflection point information is determined, accurately determining the location of the reflection point, thereby reflecting the location of the target object. This method automatically detects target objects in the blind spot at turns without requiring the driver to actively check, and is less affected by ambient light, reducing misjudgments, improving the accuracy of target object detection in blind spots, and reducing traffic accidents.

[0079] Figure 6 This is a flowchart illustrating a blind spot detection method for turning, provided as another embodiment of this application. This embodiment is an optimization based on the above embodiments; solutions not described in detail in this embodiment are found in the above embodiments. Figure 6 As shown, the method in this embodiment of the application specifically includes the following steps:

[0080] S310. The on-board image acquisition device installed on the target vehicle traveling on the target road acquires images of the area in front of the vehicle, and identifies the convex mirror area by recognizing the acquired images; wherein, the target road is either the first road or the second road.

[0081] S320. Determine the position of the vehicle-mounted radar installed on the target vehicle, and determine the emission direction of the detection light emitted by the vehicle-mounted radar toward the convex mirror area.

[0082] S330. Based on the position of the vehicle-mounted radar, the emission direction of the detection light beam, and the size information of the convex mirror, determine the reflection point information of the detection light beam after being reflected by the convex mirror, where the target object is detected and reflected by the target object, and establish a three-dimensional point cloud model of the target object based on the reflection point information.

[0083] S340. The surface point cloud data of the convex mirror is obtained through the detection of the vehicle-mounted radar.

[0084] For example, during the detection process of the vehicle radar emitting a detection ray, the intersection point of the detection ray and the convex mirror is calculated, which is a point on the surface of the convex mirror. The data of the intersection point is used as the surface point cloud data of the convex mirror.

[0085] S350. Substitute the surface point cloud data into the quadratic surface model, solve for the parameters of the quadratic surface model, and obtain the convex mirror model.

[0086] In practice, convex mirrors may become distorted during application and are not standard spheres. A quadratic surface model can be pre-defined, and the surface point cloud data can be substituted into this model to solve for its parameters, thus obtaining the convex mirror model. For example, the surface point cloud data of a convex mirror includes {(x... i ,y i ,z i ,)},i=1,2,3…,n, where n is the number of intersection points. The default quadratic surface model is: Ax 2 +By 2 +Cz 2 +Dxy+Exz+Fyz+Gx+Hy+Iz+J=0. Substituting the surface point cloud data into the preset quadratic surface model, we get: Where e i This is the error term, because surface point cloud data contains noise and other factors, these points may not fall exactly on the theoretical quadric surface model. The quadric surface model can be written in matrix form as Ma = e, where...

[0087] The least squares method can be used to solve this problem. The goal is to find 'a' that minimizes the sum of squares of the errors, which means... Minimum. According to the principle of least squares, the optimal solution for a is... Since the actual calculation of e is unknown, it can be solved... (Here, 0 is obtained because it is the ideal case where the error is minimized) That is, the coefficient estimates of the quadratic surface.

[0088] In this embodiment of the application, the method further includes:

[0089] The surface point cloud data of the convex mirror, which serves as the validation set, is substituted into the convex mirror model to obtain the validation values.

[0090] The model error is determined based on the verification value, and the convex mirror model is optimized based on the model error.

[0091] For example, in the surface point cloud data of a convex mirror, a portion is reserved not for fitting the pre-defined quadratic surface model, but instead used as a validation set to validate the convex mirror model. The surface point cloud data of the convex mirror used as the validation set is substituted into the convex mirror model to obtain validation values. The model error is determined based on the validation values, and the convex mirror model is optimized based on the model error to reduce the error. For example, the surface point cloud data used as the validation set is P. j (x i, y i ,z i Substituting this into the convex mirror model yields... Define model error as If the model error exceeds the preset error threshold, the amount of surface point cloud data of the convex mirror can be increased, data preprocessing can be optimized (such as more thorough filtering), or a more complex fitting method (such as weighted least squares method, which allocates weights according to the reliability of the surface point cloud data) can be used to optimize the convex mirror model.

[0092] In this embodiment of the application, the method further includes:

[0093] The vehicle-mounted radar emits a test beam parallel to the optical axis of the convex mirror towards the convex mirror area, and the reflection direction vector of the test beam reflected by the convex mirror is determined based on the position of the vehicle-mounted radar, the emission direction of the test beam, and the convex mirror model.

[0094] An error function is constructed based on the reflection direction vector and the focal information of the convex mirror;

[0095] The conjugate gradient method is used to iteratively optimize the parameters of the quadratic surface model based on the error function, thereby obtaining the final convex mirror model.

[0096] For example, since the reflected light rays parallel to the optical axis of the convex mirror pass through the focal point of the convex mirror after reflection, if the parameters in the convex mirror model are accurate, the reflected light rays obtained from the convex mirror model should also pass through the focal point. A test light ray parallel to the optical axis of the convex mirror can be emitted into the convex mirror area using an onboard radar. Based on the position of the onboard radar, the emission direction of the test light ray, and the convex mirror model, the reflection direction vector of the test light ray reflected by the convex mirror can be determined. Based on the reflection direction vector and the focal point information of the convex mirror, an error function is constructed and optimized to obtain accurate parameters for the convex mirror model.

[0097] Specifically, the intersection point of the test ray and the convex mirror is determined based on the same principle as in the above embodiments. At the intersection point, the normal vector is calculated according to the convex mirror model. Solve for the reflection direction vector using the normal vector. The focal point of the convex mirror is F(x) F ,y F ,z F The error function is: in The magnitude of the vector difference is calculated using the vector magnitude formula:

[0098]

[0099] weight ω iThe error function can be determined based on the intensity of the test light at the intersection point, and it is positively correlated with the intensity of the test light. The conjugate gradient method can be used to optimize the error function.

[0100] The conjugate gradient method is a commonly used iterative optimization algorithm for solving problems of the form .

[0101] ( The quadratic function minimization problem (representing the parameter vector of a convex mirror model, including the coordinates of the focal position and the coefficients of the quadratic surface, etc.) has the advantages of fast convergence speed and small storage requirements.

[0102] Initialization: Given an initial guess of the convex mirror model parameter vector. Typically, the initially estimated focal position and the fitted quadratic surface coefficients can be used as initial values. The initial error function value is then calculated. Initialize the search direction vector in For the error function in The gradient vector at point A is calculated using the partial derivative formula:

[0103] For the above error function E, the partial derivatives with respect to the coefficients A, B, C...J of the convex mirror model are used to obtain the components of the gradient vector.

[0104] Error function The gradient vector represents the function at a certain point. The direction of greatest rate of change. Specifically, each component of the gradient vector is the partial derivative of the error function with respect to the corresponding parameter. For example, if there is a multivariate function... Its gradient vector

[0105] In the current problem, Including the convex mirror model parameters (A, B, C…J), so the components of the gradient vector This refers to the partial derivatives of the error function with respect to these variables. Initialization of the search direction vector and the negative gradient direction: In the initialization phase of the conjugate gradient method, a search direction vector needs to be determined. This vector will guide the search in the parameter space for a direction that reduces the error function.

[0106] choose This is based on the following reasons: gradient vector Pointing error function in In the direction of fastest growth, then This points in the direction of the fastest decrease in the error function. Therefore, using this negative gradient direction as the initial search direction, we hope to quickly reduce the value of the error function in the first step of the iteration.

[0107] This is the initial guess of the convex mirror model parameter vector, which can be the quadratic surface coefficients obtained based on some prior knowledge or preliminary estimation. Through calculation... And take its negative direction as This establishes an initial search direction for the iterative process of the conjugate gradient method, which is expected to rapidly reduce the error function.

[0108] Iteration steps:

[0109] In the k-th iteration:

[0110] Calculate step size α k : By using a linear search method, To find the minimum value, common linear search methods include exact line search or Wolfe conditional line search. Here is a brief introduction to exact line search, which solves for:

[0111]

[0112] The solution can be found by differentiating and setting the derivative to zero; in practical calculations, numerical methods can be used for approximation. Step size α k This determines the direction of the search in each iteration. The distance traveled. Choosing an appropriate step size is crucial for quickly converging to the minimum of the error function.

[0113] Continue iterating:

[0114] a. Update the model parameter vector:

[0115] b. Calculate the new error function value

[0116] c. Update the search direction vector:

[0117]

[0118]

[0119] Explanation of the meaning of each parameter:

[0120] α k It represents the step size in the k-th iteration. This parameter controls the search direction. The distance moved upwards. By finding a suitable α k Hoping to find a new point in this direction. Make the error function As small as possible. For example, if α k If α is too large, it may exceed the minimum point of the error function; if α k If it's too small, the convergence speed will be very slow.

[0121] and : It is the model parameter vector at the k-th iteration, which contains the currently estimated convex mirror model parameters (A, B, C…J).

[0122] It is the model parameter vector updated after the k-th iteration. It is obtained by... Based on the search Directional movement α k The step size is obtained, that is The purpose of this update process is to make the error function value corresponding to the new parameter vector smaller, thereby gradually approaching the optimal parameter value that minimizes the error function.

[0123] This is the search direction vector in the iteration. In each iteration of the conjugate gradient method, the search direction vector guides the search for the direction of descent of the error function in the parameter space. This vector is calculated using the conjugate relation and gradient information mentioned earlier, and is used during the initialization phase. Subsequent updates will be made based on gradient information and conjugate properties during the iteration process, with the aim of effectively guiding the search direction to find the minimum value of the error function.

[0124] The error function to be minimized measures the difference between the model's prediction (e.g., the deviation between the endpoint and focal point of reflected light) and the actual expectation (convergence near the focal point). In each iteration, the value of the error function needs to be evaluated (by changing) under different parameter combinations to determine how to adjust the parameters to reduce the error.

[0125] Convergence condition judgment:

[0126] The iteration stops when a preset convergence condition is met, for example, when the difference between the error function values ​​obtained from two consecutive iterations is less than a certain threshold (e.g., 10). -6 ), that is, |E k+1 -E k |<∈, or when the magnitude of the gradient vector is less than a certain value, i.e. At this point, it is believed that the convex mirror model parameters that minimize the error function have been found, thus obtaining the precise focal position and focal length.

[0127] S360. Based on the position of the vehicle-mounted radar, the emission direction of the detection light, and the convex mirror model, determine the reflection point information and optimize the three-dimensional point cloud model.

[0128] For example, after determining the convex mirror model, the size information of the convex mirror can be determined based on the convex mirror model, and the reflection point information can be solved according to the above scheme. The three-dimensional point cloud model can be optimized based on the latest solved reflection point information to make the position of the target object more accurate.

[0129] This application provides a method for detecting blind spots during turns. It can fit the surface point cloud data actually detected on the surface of a convex mirror to obtain an actual convex mirror model. This solves the problem that the surface of a convex mirror is not a standard sphere due to distortion during use, and there are errors in calculating the emission point information based on the sphere. By solving for an accurate convex mirror model, the reflection point information can be solved more accurately, accurately reflecting the position of the target object and providing an effective reference for the driver of the target vehicle.

[0130] Figure 7 This is a schematic diagram of a turning blind spot detection device provided in an embodiment of this application. This device can execute the turning blind spot detection method provided in any embodiment of this application, and has the corresponding functional modules and beneficial effects for executing the method. Figure 7 As shown, a convex mirror is installed at the bend between the first road and the second road. The convex mirror is positioned at the outer vertex of the bend, facing inwards towards the bend. The device includes:

[0131] The convex mirror region determination module 410 is used to acquire images of the area in front of the vehicle by means of an on-board image acquisition device installed on the target vehicle traveling on the target road, and to identify and determine the convex mirror region from the acquired images; wherein the target road is either the first road or the second road.

[0132] The emission direction determination module 420 is used to determine the position of the vehicle-mounted radar installed on the target vehicle and to determine the emission direction of the detection light emitted by the vehicle-mounted radar toward the convex mirror area.

[0133] The three-dimensional point cloud model determination module 430 is used to determine the reflection point information of the detection light after being reflected by the convex mirror and detected by the target object, based on the position of the vehicle radar, the emission direction of the detection light and the size information of the convex mirror, and to establish a three-dimensional point cloud model of the target object based on the reflection point information.

[0134] In this embodiment of the application, the device further includes:

[0135] An adjustment module is used to increase the power and / or frequency of the detection light emitted by the vehicle-mounted radar toward the convex mirror area, based on the initial power and initial frequency of the detection light emitted toward the front of the vehicle by the vehicle-mounted radar.

[0136] In this embodiment, the three-dimensional point cloud model determination module 430 determines the reflection point information of the target object detected by the detection light after reflection by the convex mirror, based on the position of the vehicle-mounted radar, the emission direction of the detection light, and the size information of the convex mirror, including:

[0137] Based on the position of the vehicle-mounted radar and the emission direction of the detection beam, the equation of the detection beam is expressed to obtain the first equation;

[0138] The equation of the convex mirror is expressed based on the size information of the convex mirror to obtain the second equation;

[0139] The intersection information of the probe ray and the convex mirror is determined based on the first equation and the second equation.

[0140] The normal vector at the intersection point is determined based on the intersection point information and the size information of the convex mirror.

[0141] The reflection point information is determined based on the normal vector and the emission direction of the probe ray.

[0142] In this embodiment of the application, the three-dimensional point cloud model determination module 430 establishes a three-dimensional point cloud model of the target object based on the reflection point information, including:

[0143] The azimuth and elevation angles of the reflection point relative to the vehicle-mounted radar are determined based on the reflection point information.

[0144] Based on the azimuth angle, the elevation angle, and the target distance of the reflection point relative to the vehicle-mounted radar, a three-dimensional point cloud model of the target object relative to the target vehicle is established.

[0145] In this embodiment of the application, the apparatus further includes a target distance determination module, used for:

[0146] Based on the round-trip time between the vehicle-mounted radar emitting a detection beam and receiving a reflected beam, and the propagation speed of the emitted beam, the sum of the first distance from the vehicle-mounted radar to the intersection point and the second distance from the intersection point to the reflection point is determined; wherein, the intersection point is the intersection point of the detection beam and the convex mirror;

[0147] Based on the location of the vehicle-mounted radar and the location of the intersection, a first distance from the vehicle-mounted radar to the intersection is determined, and a second distance is determined based on the first distance and the location of the intersection.

[0148] The target distance is determined based on the first distance, the second distance, and the reflection angle of the probe light on the convex mirror.

[0149] In this embodiment of the application, the three-dimensional point cloud model determination module 430 establishes a three-dimensional point cloud model of the target object based on the reflection point information, including:

[0150] Based on road network information or road information in images captured by vehicle-mounted image acquisition devices, determine the traffic range on another road; wherein, the other road is a road other than the target road between the first road and the second road.

[0151] The reflection points detected by the vehicle-mounted radar that are located within the passage range are identified, and a three-dimensional point cloud model of the target object is established based on the reflection point information of the reflection points located within the passage range.

[0152] In this embodiment of the application, the device further includes:

[0153] A surface point cloud data determination module is used to obtain the surface point cloud data of the convex mirror through the detection of the vehicle-mounted radar;

[0154] The convex mirror model construction module is used to substitute the surface point cloud data into the quadratic surface model, solve the parameters of the quadratic surface model, and obtain the convex mirror model.

[0155] The 3D point cloud model optimization module is used to optimize the 3D point cloud model by determining the reflection point information based on the position of the vehicle-mounted radar, the emission direction of the detection light, and the convex mirror model.

[0156] In this embodiment of the application, the device further includes:

[0157] The verification value determination module is used to substitute the surface point cloud data of the convex mirror, which serves as the verification set, into the convex mirror model to obtain the verification value.

[0158] The convex mirror model optimization module is used to determine the model error based on the verification value and optimize the convex mirror model based on the model error.

[0159] In this embodiment of the application, the device further includes:

[0160] The reflection direction vector determination module is used to emit a test light beam parallel to the optical axis of the convex mirror towards the convex mirror area through the vehicle-mounted radar, and determine the reflection direction vector of the test light beam reflected by the convex mirror based on the position of the vehicle-mounted radar, the emission direction of the test light beam, and the convex mirror model.

[0161] An error function construction module is used to construct an error function based on the reflection direction vector and the focal information of the convex mirror.

[0162] The parameter optimization module is used to iteratively optimize the parameters of the quadratic surface model using the conjugate gradient method based on the error function, thereby obtaining the final convex mirror model.

[0163] The turning blind spot detection device provided in this application embodiment can execute the turning blind spot detection method provided in any embodiment of this application, and has the corresponding functional modules and beneficial effects of the method execution.

[0164] Figure 8 A schematic diagram of an electronic device 10, which can be used to implement embodiments of this application, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the application described and / or claimed herein.

[0165] like Figure 8 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0166] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of monitors, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless cornering detection transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0167] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the blind spot detection method for cornering.

[0168] In some embodiments, the turning blind spot detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the turning blind spot detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the turning blind spot detection method by any other suitable means (e.g., by means of firmware).

[0169] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0170] Computer programs used to implement the methods of this application may be written in any combination of one or more programming languages. These computer programs may be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable blind spot detection device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0171] In the context of this application, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0172] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0173] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0174] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0175] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired information of the technical solution of this application can be achieved, and this is not limited herein.

[0176] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for detecting blind spots during turns, characterized in that, A convex mirror is installed at the bend between the first road and the second road, the convex mirror being positioned at the outer vertex of the bend and facing inwards towards the bend; the method includes: The onboard image acquisition device installed on the target vehicle traveling on the target road acquires images of the area in front of the vehicle, and identifies the convex mirror region from the acquired images; wherein the target road is either the first road or the second road; The location of the vehicle-mounted radar installed on the target vehicle is determined, and the emission direction of the detection light emitted by the vehicle-mounted radar toward the convex mirror area is determined; Based on the location of the vehicle-mounted radar, the emission direction of the detection light beam, and the size information of the convex mirror, the reflection point information of the detection light beam after being reflected by the convex mirror and detecting the target object and being reflected by the target object is determined, and a three-dimensional point cloud model of the target object is established based on the reflection point information.

2. The method according to claim 1, characterized in that, After identifying and determining the convex mirror region from the acquired image, the method further includes: Based on the vehicle-mounted radar emitting detection light beams towards the front of the vehicle at initial power and initial frequency, the power and / or frequency of the detection light beams emitted by the vehicle-mounted radar towards the convex mirror area are increased.

3. The method according to claim 1, characterized in that, Based on the location of the vehicle-mounted radar, the emission direction of the detection light beam, and the size information of the convex mirror, the information of the reflection point of the target object detected by the detection light beam after reflection by the convex mirror is determined, including: Based on the position of the vehicle-mounted radar and the emission direction of the detection ray, the equation of the detection ray is expressed to obtain the first equation; The equation of the convex mirror is expressed based on the size information of the convex mirror to obtain the second equation; The intersection information of the probe ray and the convex mirror is determined based on the first equation and the second equation. The normal vector at the intersection point is determined based on the intersection point information and the size information of the convex mirror. The reflection point information is determined based on the normal vector and the emission direction of the probe ray.

4. The method according to claim 1, characterized in that, Based on the reflection point information, a three-dimensional point cloud model of the target object is established, including: The azimuth and elevation angles of the reflection point relative to the vehicle-mounted radar are determined based on the reflection point information. Based on the azimuth angle, the elevation angle, and the target distance of the reflection point relative to the vehicle-mounted radar, a three-dimensional point cloud model of the target object relative to the target vehicle is established.

5. The method according to claim 4, characterized in that, The process of determining the target distance includes: Based on the round-trip time between the vehicle-mounted radar emitting a detection beam and receiving a reflected beam, and the propagation speed of the emitted beam, the sum of the first distance from the vehicle-mounted radar to the intersection point and the second distance from the intersection point to the reflection point is determined; wherein, the intersection point is the intersection point of the detection beam and the convex mirror; Based on the location of the vehicle-mounted radar and the location of the intersection, a first distance from the vehicle-mounted radar to the intersection is determined, and a second distance is determined based on the first distance and the location of the intersection. The target distance is determined based on the first distance, the second distance, and the reflection angle of the probe light on the convex mirror.

6. The method according to claim 1, characterized in that, Based on the reflection point information, a three-dimensional point cloud model of the target object is established, including: Based on road network information or road information in images captured by vehicle-mounted image acquisition devices, determine the traffic range on another road; wherein, the other road is a road other than the target road between the first road and the second road. The reflection points detected by the vehicle-mounted radar that are located within the passage range are identified, and a three-dimensional point cloud model of the target object is established based on the reflection point information of the reflection points located within the passage range.

7. The method according to claim 1, characterized in that, The method further includes: The surface point cloud data of the convex mirror is obtained through the detection of the vehicle-mounted radar; The surface point cloud data is substituted into the quadratic surface model, and the parameters of the quadratic surface model are solved to obtain the convex mirror model. The reflection point information is determined based on the location of the vehicle-mounted radar, the emission direction of the detection light, and the convex mirror model, and the three-dimensional point cloud model is optimized.

8. The method according to claim 7, characterized in that, The method further includes: The surface point cloud data of the convex mirror, which serves as the validation set, is substituted into the convex mirror model to obtain the validation values. The model error is determined based on the verification value, and the convex mirror model is optimized based on the model error.

9. The method according to claim 7, characterized in that, The method further includes: The vehicle-mounted radar emits a test beam parallel to the optical axis of the convex mirror towards the convex mirror area, and the reflection direction vector of the test beam reflected by the convex mirror is determined based on the position of the vehicle-mounted radar, the emission direction of the test beam, and the convex mirror model. An error function is constructed based on the reflection direction vector and the focal information of the convex mirror; The conjugate gradient method is used to iteratively optimize the parameters of the quadratic surface model based on the error function, thereby obtaining the final convex mirror model.

10. A turning blind spot detection device, characterized in that, A convex mirror is installed at the bend between the first road and the second road. The convex mirror is positioned at the outer vertex of the bend, facing inwards towards the bend. The device includes: The convex mirror region determination module is used to acquire images of the area in front of the vehicle by means of an on-board image acquisition device installed on the target vehicle traveling on the target road, and to identify and determine the convex mirror region from the acquired images; wherein the target road is either the first road or the second road. The emission direction determination module is used to determine the position of the vehicle-mounted radar installed on the target vehicle, and to determine the emission direction of the detection light emitted by the vehicle-mounted radar toward the convex mirror area; The three-dimensional point cloud model determination module is used to determine the reflection point information of the detection light after being reflected by the convex mirror and detected by the target object, based on the position of the vehicle radar, the emission direction of the detection light and the size information of the convex mirror, and to establish a three-dimensional point cloud model of the target object based on the reflection point information.

Images