Vehicle-mounted visor control method and device, vehicle and medium

Abstract

The invention discloses a vehicle-mounted visor control method and device, a vehicle and a medium, and relates to the technical field of automobiles. The method comprises the steps of obtaining a road image of a vehicle at a first moment, and performing recognition processing on the road image to obtain a first coordinate of a light source and a light source type; obtaining eyeball coordinates of the driver at the first moment; determining a first shading area of the shading plate based on the eyeball coordinates and the first coordinates of the light source; and based on a target shading strategy matched with the light source type, carrying out shading processing on the first shading area of the shading plate. According to different light source types, a proper shading strategy is adopted to accurately shade the shading area of the shading plate so as to adapt to different driving scenes.

Description

Technical Field

[0001] This disclosure relates to the field of automotive technology, and in particular to a method, device, vehicle, and medium for controlling a vehicle-mounted sunshade. Background Technology

[0002] While driving, strong light (such as direct sunlight or oncoming vehicle headlights) can cause the driver's pupils to constrict sharply and the retinal photoreceptor cells to become saturated, resulting in glare. At this time, the driver's ability to distinguish details in and around the bright light area is significantly reduced, and may even cause temporary "blindness," which is an important cause of traffic accidents.

[0003] In existing technologies, to address the aforementioned problems, one approach is to physically block direct sunlight using traditional mechanical sunshades. However, this method covers a large area behind the sunshade, creating a fixed blind spot for the driver and posing a significant safety hazard. Another approach is to use photosensors to detect ambient light intensity and then apply the sunshade to the area to be shaded based on the light intensity. However, this method applies the shading to all light sources, failing to effectively distinguish between "harmful light sources" (such as the sun and high beams) and "important light sources" (such as traffic lights), thus increasing safety risks.

[0004] Therefore, how to effectively identify the types of light sources in different driving scenarios and select appropriate shading strategies to accurately shade the shading area of ​​the shading plate for different types of light sources has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of the above problems, this disclosure provides a method, device, vehicle, and medium for controlling a vehicle-mounted sunshade to overcome or at least partially solve the above problems. The technical solution is as follows: A method for controlling a vehicle-mounted sunshade includes: acquiring a road image of the vehicle at a first moment, and performing recognition processing on the road image to obtain the first coordinates and type of a light source; acquiring the eye coordinates of the driver at the first moment; determining a first shading area of ​​the sunshade based on the eye coordinates and the first coordinates of the light source; and performing shading processing on the first shading area of ​​the sunshade based on a target shading strategy matching the type of the light source.

[0006] This disclosure provides a vehicle-mounted sunshade control method. It acquires road images and identifies the coordinates and types of light sources in the current driving scene from these images. This facilitates effective shading of light sources in the current driving scene based on their type and coordinates. Next, it acquires the driver's eye coordinates and, based on these coordinates and the light source coordinates, determines the local area on the sunshade that needs shading. This ensures the shading area is perfectly aligned with the glare in the driver's line of sight. By selecting only the shading area, the light transmittance of other areas of the sunshade is preserved to the maximum extent, eliminating the risk of blind spots and safety hazards associated with traditional techniques. Then, it selects a shading strategy matching the light source type and applies the shading to the required local area. Different shading strategies are selected for different light source types to optimize the shading effect of the sunshade, thus enabling precise shading of the required local area for different light source types.

[0007] Optionally, the road image is processed to obtain the first coordinates and type of the light source, including: extracting features from the road image to obtain image features; performing target detection on the image features to obtain the first coordinates of the light source; performing prediction processing on the image features corresponding to the first coordinates to obtain the predicted probability of the light source corresponding to different types; and taking the type corresponding to the highest predicted probability as the type of the light source.

[0008] In this embodiment, feature extraction of road images can automatically filter out irrelevant information to reduce the amount of data in subsequent processing, thereby obtaining image features. Subsequently, target detection is performed on the image features to accurately locate the spatial position of the light source, thus obtaining the first coordinate of the light source. Based on the image features corresponding to the light source at the first coordinate, the type of the light source is predicted, and the predicted probability of the light source corresponding to different types is obtained. By focusing on the analysis of local features in the light source area, interference from background information can be avoided, improving the accuracy of type prediction, thereby obtaining the probability that the current light source belongs to different light source types. Then, the type corresponding to the highest predicted probability is taken as the light source type of the current light source, which can accurately identify the type of the light source. Moreover, the above method can avoid the missed detection / false detection problem of traditional rule-based recognition, and at the same time realize the classification of light source types, which can provide a basis for subsequent selection of different shading strategies for differentiated processing based on the light source type.

[0009] Optionally, obtaining the driver's eye coordinates at the first moment includes: obtaining a facial image of the driver at the first moment, and performing key point detection on the facial image to obtain the two-dimensional coordinates of multiple key points; obtaining a standard three-dimensional head model corresponding to the multiple key points, and performing pose estimation on the standard three-dimensional head model based on the two-dimensional coordinates of the multiple key points to obtain an optimized three-dimensional head model; and extracting the driver's eye coordinates from the optimized three-dimensional head model.

[0010] In this embodiment, by acquiring a driver's facial image and performing key point detection on the facial image, the two-dimensional coordinates of multiple key points are obtained. This provides accurate two-dimensional constraints for subsequent 3D modeling. Furthermore, key point detection can filter noise through the geometric relationships between feature points, ensuring stable output of key coordinates even in non-frontal facial postures such as the driver's side profile or head tilt. Subsequently, posture estimation is performed on a standard 3D head model based on the two-dimensional key points to obtain a 3D head model adapted to the current driver. This provides a data source for subsequent eye coordinate extraction, improving the accuracy of data acquisition. Afterward, the positions of the driver's actual eyes in the camera coordinate system can be directly extracted from the optimized 3D head model adapted to the current driver, allowing for subsequent gaze tracking based on the acquired eye coordinates.

[0011] Optionally, obtaining the driver's eye coordinates at the first moment includes: obtaining the driver's historical vehicle configuration information; constructing a three-dimensional human body model corresponding to the driver based on the historical vehicle configuration information; and extracting the driver's eye coordinates from the three-dimensional human body model.

[0012] In this embodiment, by reading the driver's relevant configuration information in the vehicle, the driver's height and other body information are inferred from the obtained configuration information to generate a three-dimensional human body model adapted to the current driver. This provides a basis for subsequent extraction of the driver's eye coordinates. Compared with real-time three-dimensional modeling, the current solution can significantly reduce the real-time computing pressure and effectively save hardware resources. Subsequently, based on the connection relationship of various structures in the three-dimensional human body model and the model's positioning in the vehicle coordinate system, the eye coordinates are obtained, and subsequent gaze tracking is performed based on the obtained eye coordinates.

[0013] Optionally, determining the first light-blocking area of ​​the light-shielding plate based on the eye coordinates and the first coordinates of the light source includes: determining the first direction vector of the light source based on the first coordinates of the light source; generating the driver's line-of-sight equation based on the eye coordinates and the first direction vector; obtaining the plane equation where the light-shielding plate is located, and taking the preset area where the line-of-sight equation intersects the plane equation as the first light-blocking area of ​​the light-shielding plate.

[0014] In this embodiment, a first direction vector is determined based on the first coordinates of the light source, thereby determining the position information of the light source in three-dimensional space. A spatial relationship is established based on the three-dimensional spatial position information and the eye coordinates. Subsequently, a three-dimensional line-of-sight equation is constructed based on the eye coordinates and the first direction vector to accurately simulate the driver's line-of-sight direction, enabling deep integration of light source information and driver information, laying the foundation for subsequent determination of the shading point on the sun visor. Then, the plane equation of the sun visor is obtained, and by solving the intersection of the plane equation and the line-of-sight equation, the three-dimensional line-of-sight path is precisely aligned with the plane of the sun visor, ensuring that the shading area is exactly located on the line connecting the driver's line of sight and the light source, thus achieving precise shading. Furthermore, a preset area is set to take into account the errors in the actual scene, ensuring that even with a small offset, glare can still be effectively covered, effectively avoiding the instability of point shading. In addition, the shading is applied to a local area around the intersection of the line of sight rather than the entire sun visor, so as to preserve the shading of other areas to the greatest extent, fundamentally eliminating the problem of fixed blind spots on the sun visor.

[0015] Optionally, determining the first direction vector of the light source based on the first coordinates of the light source includes: normalizing the first coordinates of the light source based on the camera's internal parameters to obtain the second coordinates of the light source; constructing a second direction vector corresponding to the second coordinates; and performing coordinate transformation on the second direction vector to obtain the first direction vector of the light source.

[0016] In this embodiment, the first coordinates of the light source are normalized using the camera intrinsic parameter matrix to eliminate the offset between the origin of the pixel coordinate system and the optical center of the camera, achieving "decentralization." Furthermore, the normalization process eliminates the influence of camera physical characteristics on pixel coordinates, converting the image coordinates into dimensionless normalized coordinates related to the angle between the camera's optical axis and the pixel coordinates. Subsequently, a three-dimensional direction vector (second direction vector) in the camera coordinate system is constructed based on the normalized coordinates to fully describe the spatial orientation of the light source relative to the camera. Then, the second direction vector undergoes coordinate transformation from the camera coordinate system to the vehicle coordinate system, enabling unified calculations of the light source direction with the driver's eye coordinates and the plane of the light shield, laying the foundation for subsequent tracking and determination of the light-shielding area.

[0017] Optionally, the method further includes: predicting the trajectory of the light source at a second time based on the first coordinates of the light source at the first time and the type of the light source, to obtain the third coordinates of the light source, wherein the second time is after the first time; determining the second shading area of ​​the light shield based on the third coordinates of the light source and the eye coordinates of the driver at the second time; and performing shading processing on the second shading area of ​​the light shield based on the target shading strategy.

[0018] In this embodiment, the position of the light source at the next moment in the image is predicted based on the position and type of the light source at historical moments. This allows the shading area to move according to the predicted trajectory, achieving dynamic tracking and keeping the movement of the shading area synchronized with the movement of the light source. Furthermore, by predicting the movement trajectory, the movement of the shading area can be made more continuous and natural, avoiding abrupt adjustments that could cause visual discomfort to the driver. Then, based on the target shading strategy matched to the light source type, the shading area of ​​the shading plate is shaded to ensure that the shading is completed before the light source enters the driver's line of sight. This effectively eliminates the problem of shading lag. Moreover, the trajectory prediction can quickly estimate the direction of light source movement based on historical images, shortening the system response time and thus improving driving safety.

[0019] A vehicle-mounted sunshade control device, the device comprising: The recognition module is used to acquire the road image of the vehicle at the first moment, and to perform recognition processing on the road image to obtain the first coordinates of the light source and the type of the light source; The acquisition module is used to acquire the driver's eye coordinates at the first moment; The determining module is used to determine the first light-blocking area of ​​the light-blocking plate based on the eye coordinates and the first coordinates of the light source; A light-shielding module is used to perform light-shielding treatment on the first light-shielding area of ​​the light-shielding plate based on a target light-shielding strategy that matches the type of light source.

[0020] A vehicle includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform: acquiring a road image of the vehicle at a first moment, and performing recognition processing on the road image to obtain a first coordinate and a type of a light source; acquiring the eye coordinates of the driver at the first moment; determining a first shading area of ​​a light shield based on the eye coordinates and the first coordinates of the light source; and performing shading processing on the first shading area of ​​the light shield based on a target shading strategy matching the type of the light source.

[0021] A computer-readable storage medium stores computer-executable instructions, the computer-executable instructions being configured to: acquire a road image of a vehicle at a first moment, and perform recognition processing on the road image to obtain a first coordinate and a type of light source; acquire the eye coordinates of the driver at the first moment; determine a first shading area of ​​a light shield based on the eye coordinates and the first coordinates of the light source; and perform shading processing on the first shading area of ​​the light shield based on a target shading strategy matching the type of light source. Attached Figure Description

[0022] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a vehicle-mounted sunshade control method provided in an embodiment of this application; Figure 2 This is a structural schematic diagram of a vehicle provided in an embodiment of this application. Detailed Implementation

[0023] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0024] While driving, strong light (such as direct sunlight or oncoming vehicle headlights) can cause the driver's pupils to constrict sharply and the retinal photoreceptor cells to become saturated, resulting in glare. At this time, the driver's ability to distinguish details in and around the bright light area is significantly reduced, and may even cause temporary "blindness," which is an important cause of traffic accidents.

[0025] Existing vehicle sun visors or automatic anti-glare systems have the following two main drawbacks: 1) While effectively preventing glare, it may excessively obstruct the driver's view, thus creating blind spots; 2) It cannot intelligently distinguish different types of light sources (such as the sun, vehicle headlights, and traffic lights), and it is difficult to provide precise and personalized anti-glare protection based on the position and line of sight of different drivers.

[0026] In existing technologies, some vehicles use traditional mechanical sun visors. While these physically block direct sunlight, they inevitably obscure a large area behind the visor, preventing drivers from observing road conditions and creating fixed blind spots that pose significant safety hazards. Secondly, some vehicles use a global automatic anti-glare system that uses photosensors to detect ambient light intensity. When the intensity exceeds a threshold, it controls the entire visor's color-changing area to darken uniformly. However, this system cannot effectively distinguish between "harmful light sources" (such as the sun and high beams) and "important light sources" (such as traffic lights), and the global darkening also sacrifices a significant amount of effective visibility. Furthermore, some vehicles use fixed-position local dimming solutions, blocking light at a fixed location on the visor. However, this does not consider the differences in vision caused by different drivers' heights and sitting postures. For example, for shorter drivers, a fixed-position visor may not block glare, while for taller drivers, it may be off-target, resulting in unstable anti-glare effects and a poor user experience.

[0027] Therefore, this application provides a method for controlling a vehicle-mounted sunshade. Figure 1 This is a flowchart illustrating a vehicle-mounted sunshade control method provided in an embodiment of this application. The method can be applied to different types of vehicles. The process can be executed by a computing device in the corresponding field (e.g., a controller installed in the vehicle, a vehicle-mounted system, or a server located in the cloud). Certain input parameters or intermediate results in the process can be manually adjusted to help improve accuracy.

[0028] This application provides a method for controlling a vehicle-mounted sun visor. It should be noted that the executing entity in these embodiments can be a server or any terminal device with data processing capabilities. For example, the server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. The terminal device can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, vehicle terminal, etc., but is not limited to these.

[0029] like Figure 1 As shown, this disclosure provides a method for controlling a vehicle-mounted sunshade, including: Step 101: Obtain the road image of the vehicle at the first moment.

[0030] It should be noted that the first moment refers to the starting point of the system's anti-glare control; the road image of the vehicle can be obtained using different on-board image acquisition devices, such as forward-view cameras, surround-view cameras, smart cockpit cameras, etc., without specific limitations here.

[0031] As an example, suppose a vehicle is driving on a city street at night, at 21:00 on October 23, 2024. The scenario is a rural road without streetlights, and an oncoming vehicle has its high beams on. At this time, the system uses a forward-facing camera to capture a real-time image of the road ahead.

[0032] Step 102: Perform recognition processing on the road image to obtain the first coordinates of the light source and the type of the light source.

[0033] It should be noted that the above recognition process can be implemented using deep learning object detection algorithms (such as YOLO, SSD, RetinaNet, etc.), traditional computer vision algorithms (such as detection algorithms based on brightness thresholds and geometric features), or a combination of deep learning and traditional computer vision methods; no specific limitations are made here. The first coordinate of the light source is used to characterize the position information of the light source, and the type of light source can include sunlight, oncoming high beams, traffic lights, etc.

[0034] In some embodiments, step 102 described above can be implemented as follows: extracting features from the road image to obtain image features; performing target detection on the image features to obtain the first coordinates of the light source; performing prediction processing on the type of the light source based on the image features corresponding to the first coordinates to obtain the prediction probability of the light source corresponding to different types; and taking the type corresponding to the highest prediction probability as the type of the light source.

[0035] In this way, by extracting features from road images, irrelevant information can be automatically filtered out to reduce the amount of data in subsequent processing, thereby obtaining image features. Subsequently, target detection is performed on the image features to accurately locate the spatial position of the light source, thus obtaining the first coordinate of the light source. Based on the image features corresponding to the light source at the first coordinate, the type of the light source is predicted, and the predicted probability of the light source corresponding to different types is obtained. By focusing on the analysis of local features in the light source area, the interference of background information can be avoided, improving the accuracy of type prediction, thereby obtaining the probability that the current light source belongs to different light source types. Then, the type corresponding to the highest predicted probability is taken as the light source type of the current light source, which can accurately identify the type of the light source. Moreover, the above method can avoid the missed detection / false detection problem of traditional rule-based recognition, and at the same time realize the classification of light source types, which can provide a basis for subsequent differentiated processing by selecting different shading strategies according to the light source type.

[0036] As an example, suppose the image captured by the forward-facing camera includes the sun and oncoming high beams. The sun is located in the upper right corner of the image, a circular highlighted area with a blurred halo around the edges. The oncoming high beams are located slightly to the left of the center of the image, two symmetrical small rectangular highlighted areas. The background of the image consists of the sky, road, and trees. First, feature extraction is performed on the road image. This can be achieved using a CNN backbone network. The first few convolutional layers can extract edge (e.g., the circular outline of the sun, the rectangular edges of the high beams, etc.), color (e.g., the warm yellow of the sun, the pure white of the high beams, etc.), and texture (e.g., the gradient halo around the sun's edge, the uniform brightness of the high beams, etc.) features, thus obtaining low-dimensional features containing edge and color information. These features are then processed through intermediate convolutional layers. Low-dimensional features are combined to identify local semantics (e.g., circular highlighted areas, paired small rectangular highlighted areas, light spots in the sky background, etc.), thus obtaining mid-dimensional features containing local shape and contextual information. These are then fused with global context through subsequent convolutional layers to obtain high-dimensional features containing global semantics for object detection and classification. Subsequently, anchor boxes of different sizes (e.g., 10×10, 30×30) are pre-defined on the high-dimensional features to match the light source regions in the features. The matched anchor boxes are then fine-tuned to accurately predict the bounding box coordinates of the light sources. Furthermore, the center coordinates of each light source can be determined based on the bounding box coordinates; for example, the bounding box A coordinates of the sun are [x_min=1200, Given y_min=100, x_max=1400, y_max=300], we can obtain the first coordinates of the center point of the solar light source as (u1=1300, v1=200). The coordinates of the two bounding boxes B and C of the opposing high beam are [x_min=500, y_min=450, x_max=530, y_max=480] and [x_min=670, y_min=450, x_max=700, y_max=480], respectively. Therefore, the first coordinates of the center points of the two light sources of the opposing high beam are (u2=515, v2=465) and (u3=685, v2=465, v3=685, v2=465, v3=480 ... (v3=465) Simultaneously, it predicts the light source type based on the coordinates within the bounding boxes. Each bounding box outputs the predicted probability of belonging to different light source types. For example, the output probability of bounding box A is [Sun: 0.97, Oncoming High Beam: 0.02, Traffic Light: 0.01], the output probability of bounding box B is [Sun: 0.03, Oncoming High Beam: 0.95, Traffic Light: 0.02], and the output probability of bounding box C is [Sun: 0.02, Oncoming High Beam: 0.96, Traffic Light: 0.02]. Finally, the type corresponding to the highest probability is selected as the light source type, that is, the light source type of bounding box A is the sun, and the light source types of bounding boxes B and C are both oncoming high beams.

[0037] Step 103: Obtain the driver's eye coordinates at the first moment.

[0038] It should be noted that the driver's facial image can be acquired in real time and the driver's eye coordinates can be extracted from the facial image. Alternatively, a three-dimensional human body model can be constructed based on the driver's historical vehicle configuration information (such as vehicle seat angle, steering wheel height, etc.) and the driver's eye coordinates can be extracted from the three-dimensional human body model.

[0039] In some embodiments, step 103 described above can be implemented as follows: acquiring a facial image of the driver at the first moment, and performing key point detection on the facial image to obtain the two-dimensional coordinates of multiple key points; acquiring a standard three-dimensional head model corresponding to the multiple key points, and performing pose estimation on the standard three-dimensional head model based on the two-dimensional coordinates of the multiple key points to obtain an optimized three-dimensional head model; and extracting the eye coordinates of the driver from the optimized three-dimensional head model.

[0040] Thus, by acquiring the driver's facial image and performing keypoint detection, the two-dimensional coordinates of multiple keypoints are obtained. This provides precise two-dimensional constraints for subsequent 3D modeling. Furthermore, keypoint detection filters noise through the geometric relationships between feature points, ensuring stable output of key coordinates even in non-frontal poses such as the driver's side profile or head tilt. Subsequently, pose estimation is performed on a standard 3D head model based on the two-dimensional keypoints to obtain a 3D head model adapted to the current driver. This provides a data source for subsequent eye coordinate extraction, improving the accuracy of data acquisition. Afterward, the positions of the driver's actual eyes in the camera coordinate system can be directly extracted from the optimized 3D head model adapted to the current driver, enabling subsequent gaze tracking based on the acquired eye coordinates.

[0041] As an example, suppose a driver's facial image is captured at a specific moment by an in-vehicle camera. This image includes the driver's face, head slightly tilted back (approximately 5°), eyes open, looking directly at the road ahead, and no obstructions (no mask / sunglasses). First, the facial image is preprocessed (e.g., background removal, resizing, grayscale normalization) to obtain the processed image. Next, keypoint detection is performed on the facial image to locate 68 key feature points (e.g., eye keypoints: pupil center, inner corner of the eye, outer corner of the eye; facial contour points: jawline, tip of the nose, corners of the mouth, etc.), thus obtaining the two-dimensional coordinates of these 68 keypoints. Then, the system's built-in general-purpose 3D head model is called, which can include average facial geometry (i.e., the three-dimensional relative coordinates of the preset 68 keypoints) and triangular mesh topology (i.e., facial skin). The three-dimensional mesh structure of the skeleton ensures that the spatial relationship between key points conforms to the laws of human anatomy. Based on the two-dimensional coordinates of the key points and the 3D head model, the rotation matrix R (describing the head's pitch, roll, and yaw angles) and translation vector t (describing the head's position in the camera coordinate system) of the 3D model are solved. Here, the head posture can be roughly estimated based on the 2D positions of the root of the nose and the eyes. The key points of the 3D model are projected onto the 2D image through the initial R and t, and the Euclidean distance between them and the actual detection points is calculated. Then, R and t are adjusted through the Levenberg-Marquardt (LM) algorithm to minimize the reprojection error, thereby obtaining the optimized three-dimensional head model. Finally, the driver's eye coordinates are extracted from the eye area of ​​the optimized three-dimensional head model.

[0042] In some embodiments, step 103 described above can be implemented by: obtaining the driver's historical vehicle configuration information; constructing a three-dimensional human body model corresponding to the driver based on the historical vehicle configuration information; and extracting the driver's eye coordinates from the three-dimensional human body model.

[0043] Thus, by reading the driver's relevant configuration information in the vehicle, the driver's height and other body information can be inferred from the obtained configuration information to generate a 3D human body model adapted to the current driver. This provides a basis for the subsequent extraction of the driver's eye coordinates. Moreover, compared with real-time 3D modeling, the current solution can significantly reduce the real-time computing pressure and effectively save hardware resources. Subsequently, based on the connection relationship of various structures in the 3D human body model, and combined with the model's positioning in the vehicle coordinate system, the eye coordinates are obtained, and subsequent gaze tracking is performed based on the obtained eye coordinates.

[0044] It should be noted that historical vehicle configuration information refers to the configuration information that the driver customizes according to their own needs when driving the vehicle, such as the seat position, steering wheel height, and rearview mirror position, etc., and is not specifically limited here.

[0045] As an example, assuming a user logs into their account through the vehicle's infotainment system, the system reads the user's historical configuration data from the cloud or local storage. The system's built-in mapping model then estimates information such as the driver's height based on this historical configuration information. For instance, the driver's height can be estimated using the formula "seat fore-aft ratio = 0.4 × height + constant." Simultaneously, the system calculates the dimensions of various body parts based on a standard human proportion model. For example, the standard sitting height (from seat surface to head) for a 175cm tall person is 950mm, and the eye height (from seat surface to eyeball) is 65% of the sitting height. Next, the system obtains the coordinates of the vehicle seat's reference point M and calculates the vertical height of the eyeball relative to reference point M based on the seat height and the standard eye height ratio. Based on the seat fore-aft distance and steering wheel position, the system estimates the driver's torso length, thus deducing the horizontal distance from the eyeball to reference point M. Simultaneously, the system retrieves a standard 3D human model corresponding to the driver's height from the database and adjusts the model's posture according to the calculated driver's human body parameters, resulting in a 3D human model adapted to the current driver. Finally, the system extracts the current driver's eye coordinates from the eyeball portion of the human model adapted to the current driver.

[0046] Step 104: Based on the eye coordinates and the first coordinates of the light source, determine the first light-blocking area of ​​the light-blocking plate.

[0047] It should be noted that the light-shielding plate can use an electrochromic element array (which relies on electrochemical reactions to reversibly change the light absorption characteristics of the material itself through the insertion and extraction of ions in the material layer, so as to achieve a change from transparent to dark), or it can use a liquid crystal matrix panel (which controls the transmission or scattering of light by changing the alignment direction of liquid crystal molecules by applying voltage, without changing the absorption of light itself). No specific limitation is made here.

[0048] In some embodiments, step 104 described above can be implemented as follows: determining a first direction vector of the light source based on the first coordinates of the light source; generating the driver's line-of-sight equation based on the eye coordinates and the first direction vector; obtaining the plane equation where the light shield is located, and using a preset area where the line-of-sight equation intersects the plane equation as the first light-shielding area of ​​the light shield.

[0049] Thus, the first direction vector is determined based on the first coordinate of the light source, thereby determining the position information of the light source in three-dimensional space, and establishing a spatial relationship based on the three-dimensional spatial position information and the eye coordinates. Subsequently, based on the eye coordinates and the first direction vector, a three-dimensional line-of-sight equation is constructed to accurately simulate the driver's line-of-sight direction, enabling deep integration of light source information and driver information, laying the foundation for subsequently determining the shading point on the sun visor. Then, the plane equation of the sun visor is obtained, and by solving the intersection of the plane equation and the line-of-sight equation, the three-dimensional line-of-sight path is precisely aligned with the plane of the sun visor, ensuring that the shading area is exactly located on the line connecting the driver's line of sight and the light source, so as to achieve precise shading. Furthermore, the preset area takes into account the errors in the actual scene, ensuring that even if a small offset is achieved, glare can still be effectively covered, which can effectively avoid the instability of point shading. In addition, the shading is applied to a local area around the intersection of the line of sight rather than the entire sun visor, so as to preserve the shading of other areas to the greatest extent, fundamentally eliminating the problem of fixed blind spots of the sun visor.

[0050] It should be noted that the direction vector is a three-dimensional vector used to represent the direction of a straight line in space. The first direction vector of the light source is a unit vector indicating the direction of the light source and is used to describe the spatial orientation of the light source relative to the vehicle. The line-of-sight equation is a three-dimensional parametric equation that can represent a straight line extending from the driver's eye coordinates along the direction vector of the light source and is used to characterize the driver's line-of-sight direction.

[0051] As an example, assuming the first coordinate O of the light source is (1500, 1300), the system converts the two-dimensional pixel coordinates into directions in the vehicle coordinate system using camera calibration parameters (intrinsic parameters include focal length and principal point, extrinsic parameters include camera mounting angle), thus obtaining the first direction vector. Then, starting from the eye coordinate E, the straight line in the direction of the first direction vector is solved, which yields the driver's line of sight equation. After that, the equation of the plane where the sun visor is located is obtained (the equation of the plane where the sun visor is located can be solved according to the preset vehicle parameters), and the intersection point C of the equation and the sun visor is solved, thus obtaining the precise position on the sun visor where light needs to be blocked. Considering that the light source has a certain size (approximately 0.5° in diameter) and the driver's head may shake slightly, the system can set the light-blocking area as a circular area with a diameter d centered at the intersection point as the area to be blocked, i.e., the first light-blocking area.

[0052] In some embodiments, the above-described determination of the first direction vector of the light source based on the first coordinates of the light source can be achieved by: normalizing the first coordinates of the light source based on the camera's internal parameters to obtain the second coordinates of the light source; constructing a second direction vector corresponding to the second coordinates, and performing coordinate transformation on the second direction vector to obtain the first direction vector of the light source.

[0053] Thus, by using the camera intrinsic parameter matrix, the first coordinate of the light source is normalized to eliminate the offset between the origin of the pixel coordinate system and the optical center of the camera, achieving "decentralization." Furthermore, the normalization process can also eliminate the influence of the camera's physical characteristics on the pixel coordinates, converting the image coordinates into dimensionless normalized coordinates related to the angle between the camera's optical axis and the pixel coordinates. Subsequently, a three-dimensional direction vector (second direction vector) in the camera coordinate system is constructed based on the normalized coordinates to fully describe the spatial orientation of the light source relative to the camera. Then, the second direction vector is transformed from the camera coordinate system to the vehicle coordinate system, so that the direction of the light source can be uniformly calculated with the driver's eye coordinates and the plane of the light shield, laying the foundation for subsequent tracking and determination of the light shielding area.

[0054] As an example, assume the camera's intrinsic focal length is fx = 1000 pixels, fy = 1000 pixels, and the principal point is cx = 960 pixels, cy = 540 pixels. Here, focal length refers to the distance from the camera's optical center to the imaging plane, and the principal point is the projection of the camera's optical axis onto the imaging plane. First, the first coordinate (800, 500) is normalized according to the camera's intrinsic parameters to eliminate the influence of the camera's focal length and principal point position on the light source's two-dimensional pixel coordinates. This includes horizontal and vertical normalization, resulting in the output second coordinate (-0.16, -0.04), meaning the light source is located 0.16 focal units to the left and 0.04 focal units below the center on the camera's imaging plane. Then, the second coordinate is expanded into a three-dimensional direction vector, resulting in the second direction vector (-0.16, -0.14, ...). 1) The second direction vector represents the direction that starts from the camera light point, extends 0.16 units to the left, 0.04 units down, and 1 unit forward, and finally points to the light source. Finally, according to the rotation matrix of the camera extrinsic parameters, the second direction vector in the camera coordinate system is converted into the first direction vector in the vehicle coordinate system. Here, due to the tilt and deflection of the camera during installation, there is an angle between its forward direction and the direction directly in front of the vehicle. Through the rotation matrix, the second direction vector (-0.16, -0.04, 1) is rotated around the X and Y axes of the vehicle coordinate system to counteract the influence of the camera installation angle.

[0055] Step 105: Based on the target shading strategy that matches the type of light source, perform shading treatment on the first shading area of ​​the shading plate.

[0056] It should be noted that different types of light sources have different light intensities, so different shading strategies are used to adapt to different types of light sources. For example, a medium gray scale and stable size shading strategy can be used for the sun, a dark and fast-response shading strategy can be used for oncoming high beams, and a non-coloring strategy can be used for traffic lights.

[0057] As an example, assuming the light source type is an oncoming high beam, the target shading strategy matched with the oncoming high beam is obtained by matching with a shading area gray value of 80% and a response speed of less than 50ms. Kalman filtering is enabled to predict the light source trajectory, and the shading area adjusts its position in real time as the light source moves. The diameter of the shading area is 4cm. Then, the corresponding area on the shading plate is controlled to change color according to the target shading strategy.

[0058] In another example, assuming the light source is sunlight, the target shading strategy that matches sunlight is obtained by matching the gray value of the shading area to 50%, the response speed to be less than 100ms, dynamic tracking to be disabled, and the diameter of the shading area to be 8cm. Then, the corresponding area on the shading plate is controlled to change color according to the target shading strategy.

[0059] In another example, assuming the light source is a traffic light, and the target shading strategy matched with the traffic light is to not perform any shading treatment, then the shading plate is controlled not to perform any color change treatment according to the target shading strategy.

[0060] In some embodiments, the following processing may also be performed: based on the first coordinates of the light source at the first moment and the type of the light source, the trajectory of the light source at the second moment is predicted to obtain the third coordinates of the light source, wherein the second moment is after the first moment; based on the third coordinates of the light source and the eye coordinates of the driver at the second moment, the second shading area of ​​the light shield is determined; and based on the target shading strategy, the second shading area of ​​the light shield is shaded.

[0061] In this way, the position of the light source at the next moment in the image is predicted based on the position and type of the light source at historical moments. This allows the shading area to move according to the predicted trajectory, achieving dynamic tracking and keeping the movement of the shading area synchronized with the movement of the light source. Furthermore, by predicting the movement trajectory, the movement of the shading area can be made more continuous and natural, avoiding abrupt adjustments that could cause visual discomfort to the driver. Then, based on the target shading strategy matched to the type of light source, the shading area of ​​the shading plate is shaded to ensure that the shading is completed before the light source enters the driver's line of sight. This can effectively eliminate the problem of shading lag. Moreover, trajectory prediction can quickly estimate the direction of light source movement based on historical images, shortening the system response time and thus improving driving safety.

[0062] It should be noted that methods for predicting the trajectory of a light source can include Kalman filtering, constant velocity / constant acceleration models, deep learning, optical flow methods, etc., and no specific limitations are made here.

[0063] As an example, suppose a vehicle is traveling on a highway at night (speed 100 km / h), and an oncoming vehicle with its high beams on (relative speed approximately 200 km / h) appears in the opposite lane. The forward-facing camera identifies the pixel coordinates of the oncoming high beams at time t0 as (700, 520). Since the oncoming vehicle is traveling in a straight line, a uniform linear motion model is used to predict its trajectory. The relative speed is 200 km / h = 55.6 m / s. Within 100 ms (the second time t1 = t0 + 100 ms), the light source moves laterally in the image. The distance of this lateral movement is calculated based on the camera's field of view (e.g., 1° field of view corresponds to 30 pixels): 55.6 m / s × 0.1 s = 5.56 m, corresponding to a lateral displacement of approximately 150 pixels (moving towards the image center). Therefore, the predicted pixel coordinates of the light source at time t1 are (850, 520). Then, using the eye coordinates at the second time t1 as the starting point, combined with the predicted third coordinates (850, 520...), the trajectory is predicted... Based on the direction vector corresponding to 520), a new line-of-sight equation is constructed, and the intersection point of the line of sight on the sunshade at time t1 is solved. The area with a preset area centered on the intersection point is used as the second sunshade area. The second sunshade area is then shaded according to the target sunshade strategy (e.g., 80% grayscale of the sunshade area, 50ms response speed, dynamic tracking mode enabled, and 4cm diameter). In other words, from the driver's perspective, the strong light of the high beam is always covered by a dark sunshade area, and the sunshade area slides smoothly with the movement of the light source. Meanwhile, other areas of the sunshade remain transparent, so as not to affect the observation of the oncoming lane and road markings.

[0064] The following will describe an exemplary application of the embodiments of this application in a real-world application scenario.

[0065] In some embodiments, to address the problems of severe field-of-view obstruction, limited processing strategies, and inability to adapt to different drivers' gaze patterns in existing anti-glare systems, this application achieves intelligent control of the vehicle's sun visor by integrating forward-looking AI image recognition, in-vehicle driver gaze positioning, and independent zone control electrochromic technology. This is achieved by intelligently identifying the type and forward position of the glare source, tracking the driver's eye spatial coordinates in real time, determining the obstruction point on the sun visor through geometric calculations, and finally applying precise and differentiated local darkening only to that point. This results in personalized and source-specific anti-glare, allowing for precise shading of the visor's shading area based on different light source types, adapting to various driving scenarios.

[0066] In some embodiments, the entire processing flow of this application mainly consists of four parts: a forward-view image acquisition module, a driver gaze positioning module, a data processing and control unit, and a local dimming execution unit. The forward-view image acquisition module is an existing vehicle-mounted forward-view camera located inside the windshield of the vehicle, which can capture road scene information. The driver gaze positioning module is an existing driver monitoring system camera, which can capture driver facial images. The data processing and control unit is a vehicle domain controller or a dedicated ECU, which has built-in AI recognition algorithms and geometric calculation modules. The local dimming execution unit is a dimming film composed of an array of electrochromic elements installed in the position of a traditional sunshade. This array can be divided into M×N units that can be independently controlled by circuitry.

[0067] In some embodiments, the system first transmits a video stream to the control unit via a forward-facing camera, runs a lightweight AI model, and identifies and locates strong light sources (such as the sun, oncoming high beams, and taillights in the same direction) in the image in real time, outputting their pixel coordinates in the forward-facing camera image. Simultaneously, the driver monitoring system camera captures the driver's face in real time, calculating the three-dimensional coordinates of the driver's eyes within the vehicle space using facial key point detection and pupil localization algorithms. Subsequently, the system establishes a unified vehicle coordinate system, transforming the pixel coordinates of the glare sources identified by the forward-facing camera into this coordinate system using camera calibration parameters to obtain the direction vector of the glare source. Based on the driver's eye coordinates and the glare source direction vector, the system calculates the line-of-sight equation from the eyes to the glare source and calculates the intersection coordinates of this line of sight with the physical plane of the electrochromic array (sunshade). This intersection point is the point where local darkening is required. Precise shading position; Here, the system can also indirectly calculate the approximate driver's eye position based on the different user's seat, steering wheel, and rearview mirror positions using a preset human body model, which can effectively reduce computational costs; Then, the control unit applies different dimming strategies to the calculated precise shading position according to the type of light source identified by AI, and independently controls the electrochromic element array in zones through circuitry. For example, for sunlight, a specific voltage is applied to the intersection area to form a medium grayscale, stable-sized shading area; for oncoming high beams, another specific voltage is applied to the intersection area to form a dark, fast-responding shading area, and a motion prediction algorithm is activated to make the shading area move along the predicted trajectory to achieve dynamic tracking; In addition, the system only applies voltage to a specific range of shading areas around the intersection to darken them, while the electrochromic elements in other areas remain in a high-transmittance state because they are not energized.

[0068] In some embodiments, the vehicle-mounted glare control method provided in this application completely eliminates glare while preserving the driver's overall field of vision to the greatest extent, essentially eliminating blind spots and achieving personalized and precise protection tailored to each individual, greatly improving driving safety. By employing AI image recognition technology to distinguish light source types, a differentiated and intelligent anti-glare strategy is achieved. At the same time, based on the driver's gaze positioning and geometric projection calculation of the driver monitoring system, the core problem of "visual alignment" is solved, enabling local dimming to adapt to different drivers. Furthermore, by using an array of electrochromic elements with independent zone control as actuators, precise glare blocking is achieved, effectively preserving the driver's overall field of vision and improving driving safety.

[0069] The following describes an exemplary structure of the vehicle sunshade control device provided in the embodiments of this application as a software module. In some embodiments, the software module in the vehicle sunshade control device may include: an identification module, an acquisition module, a determination module, and a prediction module.

[0070] The system includes a recognition module for acquiring a road image of the vehicle at a first moment and performing recognition processing on the road image to obtain the first coordinates and type of the light source; an acquisition module for acquiring the eye coordinates of the driver at the first moment; a determination module for determining the first shading area of ​​the light shield based on the eye coordinates and the first coordinates of the light source; and a shading module for shading the first shading area of ​​the light shield based on a target shading strategy matching the type of the light source.

[0071] In some embodiments, the recognition module is further configured to extract features from the road image to obtain image features; perform target detection on the image features to obtain the first coordinates of the light source; perform prediction processing on the type of the light source based on the image features corresponding to the first coordinates to obtain the prediction probability of the light source corresponding to different types; and take the type corresponding to the highest prediction probability as the type of the light source.

[0072] In some embodiments, the acquisition module is further configured to acquire a facial image of the driver at the first moment, and perform key point detection on the facial image to obtain the two-dimensional coordinates of multiple key points; acquire a standard three-dimensional head model corresponding to the multiple key points, and perform pose estimation on the standard three-dimensional head model based on the two-dimensional coordinates of the multiple key points to obtain an optimized three-dimensional head model; and extract the eye coordinates of the driver from the optimized three-dimensional head model.

[0073] In some embodiments, the acquisition module is further configured to acquire the driver's historical vehicle configuration information; construct a three-dimensional human body model corresponding to the driver based on the historical vehicle configuration information; and extract the driver's eye coordinates from the three-dimensional human body model.

[0074] In some embodiments, the determining module is further configured to determine a first direction vector of the light source based on the first coordinates of the light source; generate the driver's line-of-sight equation based on the eye coordinates and the first direction vector; obtain the plane equation where the light shield is located, and take the preset area region where the line-of-sight equation intersects the plane equation as the first light-shielding area of ​​the light shield.

[0075] In some embodiments, the determining module is further configured to normalize the first coordinates of the light source based on the camera's internal parameters to obtain the second coordinates of the light source; construct a second direction vector corresponding to the second coordinates; and perform coordinate transformation on the second direction vector to obtain the first direction vector of the light source.

[0076] In some embodiments, the prediction module is further configured to predict the trajectory of the light source at a second time based on the first coordinates of the light source at the first time and the type of the light source, to obtain the third coordinates of the light source, wherein the second time is after the first time; determine the second shading area of ​​the light shield based on the third coordinates of the light source and the eye coordinates of the driver at the second time; and perform shading processing on the second shading area of ​​the light shield based on the target shading strategy.

[0077] It should be noted that the description of the apparatus in this application embodiment is similar to the description of the method embodiment above, and has similar beneficial effects as the method embodiment, so it will not be repeated.

[0078] Figure 2 This is a schematic diagram of the structure of a vehicle provided in an embodiment of this application.

[0079] For example, such as Figure 2 As shown, the vehicle 200 includes a memory 201 and a processor 202. The memory 201 stores executable program code 2011, and the processor 202 is used to call and execute the executable program code 2011 to perform the data management method of the vehicle information processor.

[0080] This embodiment can divide the vehicle into functional modules according to the above method example. For example, each function can be assigned to a separate module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0081] When each functional module is divided according to its corresponding function, the vehicle may include: Acquire road images of vehicles at the first moment, and perform recognition processing on the road images to obtain the first coordinates and light source type of the light source; Obtain the driver's eye coordinates at the first moment; Based on the eye coordinates and the first coordinates of the light source, the first light-blocking area of ​​the light-blocking plate is determined; Based on a target shading strategy that matches the type of light source, the first shading area of ​​the shading plate is shading treated.

[0082] Some embodiments of this application provide corresponding to Figure 1 A computer-readable storage medium stores computer-executable instructions, which are configured to: acquire a road image of a vehicle at a first moment, and perform recognition processing on the road image to obtain the first coordinates and type of a light source; acquire the eye coordinates of the driver at the first moment; determine a first shading area of ​​a light shield based on the eye coordinates and the first coordinates of the light source; and perform shading processing on the first shading area of ​​the light shield based on a target shading strategy matching the type of the light source.

[0083] In this application, similar or identical parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the medium embodiments are basically similar to the method embodiments, so the description is relatively simple, and relevant parts can be referred to the description of the method embodiments.

[0084] The vehicles and media provided in this application are in one-to-one correspondence with the methods. Therefore, the vehicles and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the vehicles and media will not be repeated here.

[0085] Those skilled in the art will understand that embodiments of this application can be provided as methods, vehicles, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0086] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (vehicles), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0087] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0088] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0089] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0090] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0091] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0092] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0093] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for controlling a vehicle-mounted sunshade, characterized in that, The method includes: Acquire road images of vehicles at the first moment, and perform recognition processing on the road images to obtain the first coordinates and light source type of the light source; Obtain the driver's eye coordinates at the first moment; Based on the eye coordinates and the first coordinates of the light source, the first light-blocking area of ​​the light-blocking plate is determined; Based on a target shading strategy that matches the type of light source, the first shading area of ​​the shading plate is shading treated.

2. The vehicle-mounted sunshade control method according to claim 1, characterized in that, The process of identifying and processing the road image to obtain the first coordinates and type of the light source includes: Feature extraction is performed on the road image to obtain image features; Target detection is performed on the image features to obtain the first coordinates of the light source; Based on the image features corresponding to the first coordinate, the type of the light source is predicted to obtain the predicted probability of the light source corresponding to different types; The type corresponding to the highest predicted probability is taken as the light source type.

3. The vehicle-mounted sunshade control method according to claim 1, characterized in that, The step of obtaining the driver's eye coordinates at the first moment includes: Acquire the driver's facial image at the first moment, and perform key point detection on the facial image to obtain the two-dimensional coordinates of multiple key points; Obtain a standard 3D head model corresponding to multiple key points, and perform pose estimation on the standard 3D head model based on the 2D coordinates of multiple key points to obtain an optimized 3D head model. The driver's eye coordinates are extracted from the optimized 3D head model.

4. The vehicle-mounted sunshade control method according to claim 1, characterized in that, The step of obtaining the driver's eye coordinates at the first moment includes: Obtain the driver's historical vehicle configuration information; Based on the historical vehicle configuration information, a three-dimensional human body model corresponding to the driver is constructed. Extract the eye coordinates of the driver from the three-dimensional human body model.

5. The vehicle-mounted sunshade control method according to claim 1, characterized in that, Determining the first light-blocking area of ​​the light-blocking plate based on the eye coordinates and the first coordinates of the light source includes: Based on the first coordinates of the light source, determine the first direction vector of the light source; Based on the eye coordinates and the first direction vector, the driver's line of sight equation is generated; Obtain the plane equation where the light-shielding plate is located, and take the preset area where the line-of-sight equation intersects the plane equation as the first light-shielding area of ​​the light-shielding plate.

6. The vehicle-mounted sunshade control method according to claim 5, characterized in that, Determining the first direction vector of the light source based on its first coordinates includes: Based on the camera's internal parameters, the first coordinates of the light source are normalized to obtain the second coordinates of the light source. A second direction vector corresponding to the second coordinate is constructed, and the second direction vector is transformed to obtain the first direction vector of the light source.

7. The vehicle-mounted sunshade control method according to claim 1, characterized in that, The method further includes: Based on the first coordinates of the light source at the first moment and the type of the light source, the trajectory of the light source at the second moment is predicted to obtain the third coordinates of the light source. The second moment is after the first moment. Based on the third coordinate of the light source and the eye coordinate of the driver at the second moment, the second light-blocking area of ​​the light-blocking plate is determined; Based on the target shading strategy, the second shading area of ​​the shading plate is subjected to shading treatment.

8. A vehicle-mounted sunshade control device, characterized in that, The device includes: The recognition module is used to acquire the road image of the vehicle at the first moment, and to perform recognition processing on the road image to obtain the first coordinates of the light source and the type of the light source; The acquisition module is used to acquire the driver's eye coordinates at the first moment; The determining module is used to determine the first light-blocking area of ​​the light-blocking plate based on the eye coordinates and the first coordinates of the light source; A light-shielding module is used to perform light-shielding treatment on the first light-shielding area of ​​the light-shielding plate based on a target light-shielding strategy that matches the type of light source.

9. A vehicle, characterized in that, The vehicles include: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform a vehicle sunshade control method as described in any one of claims 1-7.

10. A computer storage medium storing computer-executable instructions, characterized in that, When the computer-executable instructions are executed, they implement a vehicle-mounted sunshade control method as described in any one of claims 1-7.

Images