Driving warning system and method

Abstract

The invention discloses a driving warning method implemented by a driving warning system, which comprises the following steps of: judging whether the distance between a first vehicle and a second vehicle is less than or equal to a distance threshold value or not according to film data under the condition that the second vehicle is judged to be positioned right behind the first vehicle, and outputting a first warning notice when the judgment result is yes. And under the condition that the second vehicle is determined to be positioned obliquely behind the first vehicle, superposing a dead angle distribution model related to the second vehicle to a two-dimensional plane top view which shows the relative position of the first vehicle and the second vehicle so as to determine whether to output a second warning notice or not.

Description

Technical Field

[0001] The present invention relates to a warning system, and more particularly to a driving warning system suitable for use in a vehicle. The present invention also relates to a driving warning method suitable for use in a vehicle. Background Art

[0002] Large vehicles (such as buses, trucks and articulated vehicles) have high driver seats and long bodies, so there are many blind spots in the driver's field of vision when driving. Moreover, according to statistics, when drivers of large vehicles or surrounding vehicles perceive danger during driving, they usually only have 0.6 to 1 second of effective response time. The above situation is the cause of many serious traffic accidents. In view of this, how to reduce the risk of traffic accidents caused by blind spots in the field of vision of large vehicles has become a topic worth discussing. Summary of the invention

[0003] Therefore, one of the objectives of the present invention is to provide a driving warning system that helps to reduce the incidence of traffic accidents.

[0004] The driving warning system of the present invention is suitable for application in a first vehicle, and the driving warning system comprises a storage unit, a shooting unit, an output unit, and a processing unit electrically connected to the storage unit, the shooting unit, and the output unit. The storage unit stores a front-to-back distance threshold, a left-to-right distance threshold different from the front-to-back distance threshold, a vehicle blind spot coverage threshold, and a vehicle data database, the data database comprising a vehicle blind spot distribution model group and a vehicle size reference data group. The shooting unit is suitable for being arranged in the first vehicle and used to shoot toward the rear of the first vehicle. The processing unit is used to: determine whether the video data captured by the shooting unit shows a second vehicle that is a large vehicle; if it is determined that the video data shows the second vehicle, enter a blind spot detection period to further determine whether the second vehicle is located directly behind or obliquely behind the first vehicle; after determining that the second vehicle is located directly behind the first vehicle, determine whether a first blind spot warning condition corresponding to the front-rear distance threshold is met based on the video data, and control the output unit to output a first warning notification when the judgment result is yes, wherein the first blind spot warning condition includes that the distance between the first vehicle and the second vehicle is less than equal to the distance between the first vehicle and the second vehicle. at the front-to-rear distance threshold; after determining that the second vehicle is located obliquely behind the first vehicle, the processing unit generates a two-dimensional plane top view of the distance and orientation of the first vehicle and the second vehicle according to the video data and the vehicle size reference data group, and at the same time, the processing unit extracts a blind spot distribution model corresponding to the second vehicle from the vehicle blind spot distribution model group, and superimposes the blind spot distribution model on the position of the second vehicle in the top view, and the processing unit determines whether a second blind spot warning condition corresponding to the left-right distance threshold and the vehicle blind spot coverage threshold are both met, and when the judgment result is yes, the output unit is controlled to output a second warning notification. The second blind spot warning condition includes that the distance between the first vehicle and the second vehicle is less than or equal to the left-right distance threshold, and the area covered by the vehicle blind spot distribution model of the second vehicle to the first vehicle is greater than or equal to the vehicle blind spot coverage threshold. It is worth mentioning that the blind spot distribution model group is pre-defined according to the types of large vehicles on the market, the characteristic appearance of the front of the vehicle, the position of the original rearview mirror and other conditions to correspond to the specific blind spot range around the body of each large vehicle. In addition, the size data of the first vehicle can be set in the driving warning system by default through the processing unit according to the size reference data group of the vehicle, and the size of the second vehicle can also be extracted through the data group.

[0005] In some embodiments of the driving warning system of the present invention, the processing unit determines whether the second vehicle is located directly behind or diagonally behind the first vehicle, including defining one or more lanes presented by the video data based on one or more road markings presented by the video data, and determining whether the second vehicle is located in the same lane as the first vehicle. If it is determined that the second vehicle and the first vehicle are located in the same lane, the second vehicle is determined to be directly behind the first vehicle; if it is determined that the second vehicle and the first vehicle are located in different lanes, the second vehicle is determined to be located diagonally behind the first vehicle.

[0006] In some embodiments of the driving warning system of the present invention, the duration of the blind spot detection period is a default fixed number of seconds, or is defined as the time from when the second vehicle presenting as a large vehicle is determined to appear in the video data to when it is determined to disappear.

[0007] In some embodiments of the driving warning system of the present invention, the storage unit stores two left and right distance thresholds and two vehicle blind spot coverage thresholds, the two left and right distance thresholds are respectively a first left and right distance threshold and a second left and right distance threshold greater than the first left and right distance threshold, and the vehicle blind spot coverage thresholds can be respectively a first vehicle blind spot coverage threshold and a second vehicle blind spot coverage threshold less than the first vehicle blind spot coverage threshold. In the case of determining that the second vehicle in the video data is a large vehicle, the processing unit further determines whether the video data meets a turning intention condition, wherein the turning intention condition represents that the image or sound of the video data indicates an expected turning side related to the subsequent driving mode of the second vehicle, and the expected turning side is one of the left and right sides of the second vehicle. After determining that the second vehicle is located obliquely behind the first vehicle, the processing unit operates in a manner that includes: after determining that the second vehicle is located obliquely behind the first vehicle, but the video data does not meet the turning intention condition, determining whether the second blind spot warning condition is met based on the video data, and controlling the output unit to output the second warning notification when the judgment result is yes, wherein the second blind spot warning condition includes that the distance between the first vehicle and the second vehicle is less than or equal to the first left-right distance threshold, and that the area covered by the vehicle blind spot distribution model of the second vehicle to the first vehicle area is greater than or equal to the first vehicle blind spot coverage threshold; after determining that the second vehicle is located obliquely behind the first vehicle, and the video data meets the turning intention condition, determining the first vehicle when whether the front position is biased toward the expected turning side of the second vehicle; if it is determined that the current position of the first vehicle is not biased toward the expected turning side of the second vehicle, whether the second blind spot warning condition is met according to the video data, and when the judgment result is yes, the output unit is controlled to output the second warning notification; if it is determined that the current position of the first vehicle is biased toward the expected turning side of the second vehicle, whether a third blind spot warning condition is met according to the video data, and when the judgment result is yes, the output unit is controlled to output a third warning notification, wherein the third blind spot warning condition includes that the distance between the first vehicle and the second vehicle is less than or equal to the second left-right distance threshold, and that the area covered by the vehicle blind spot distribution model of the second vehicle to the first vehicle area is greater than or equal to the second vehicle blind spot coverage threshold.

[0008] In some embodiments of the driving warning system of the present invention, the processing unit determines whether the video data meets the turning intention condition in a manner that includes: determining whether only one side of the second vehicle's turn signal lights are flashing based on the image of the video data; if the determination result is yes, the side of the left or right sides of the second vehicle on which the turn signal lights are flashing is used as the expected turning side indicated by the video data, and determining that the video data meets the turning intention condition.

[0009] In some implementations of the driving warning system of the present invention, the processing unit determines whether the video data meets the turning intention condition in a manner that includes: determining whether the sound of the video data contains a turning keyword indicating a turning side; if the determination result is yes, using the turning side indicated by the turning keyword as the expected turning side indicated by the video data, and determining that the video data meets the turning intention condition.

[0010] Another object of the present invention is to provide a driving warning method that helps to reduce the incidence of traffic accidents.

[0011] The driving warning method of the present invention is implemented by a driving warning system, which is suitable for use in a first vehicle and includes a storage unit, a shooting unit suitable for being arranged on the first vehicle and used for shooting toward the rear of the first vehicle, an output unit, and a processing unit electrically connected to the storage unit, the shooting unit and the output unit. The storage unit stores a front-to-back distance threshold, a left-to-right distance threshold different from the front-to-back distance threshold, a vehicle blind spot coverage threshold, and a vehicle data database, which includes a vehicle blind spot distribution model group and a vehicle size reference data group. The driving warning method includes: (A) the processing unit determines whether the video data captured by the shooting unit shows a second vehicle that is a large vehicle; (B) when it is determined that the video data shows the second vehicle, a blind spot detection period is entered to further determine whether the second vehicle is located directly behind or obliquely behind the first vehicle; (C) after determining that the second vehicle is located directly behind the first vehicle, the processing unit determines whether a first blind spot warning condition corresponding to the front-rear distance threshold is met based on the video data, and when the judgment result is yes, controls the output unit to output a first warning notification, wherein the first blind spot warning condition includes that the distance between the first vehicle and the second vehicle is less than or equal to the front-rear distance threshold; (D) after determining that the second vehicle is located obliquely behind the first vehicle, the processing unit determines according to step A whether the first blind spot warning condition corresponding to the front-rear distance threshold is met. The processing unit determines the large vehicle characteristics presented by the second vehicle, generates a two-dimensional plane top view of the distance and orientation of the first vehicle and the second vehicle according to the film data and the vehicle size reference data group, and at the same time, the processing unit extracts a blind spot distribution model corresponding to the second vehicle from the vehicle blind spot distribution model group, and overlays the blind spot distribution model on the position of the second vehicle in the top view, and the processing unit determines whether a second blind spot warning condition corresponding to the left and right distance threshold and the vehicle blind spot coverage threshold are both met, and when the judgment result is yes, controls the output unit to output a second warning notification, wherein the second blind spot warning condition includes that the distance between the first vehicle and the second vehicle is less than or equal to the left and right distance threshold, and that the area of ​​the first vehicle area covered by the blind spot of the second vehicle is greater than or equal to the vehicle blind spot coverage threshold.

[0012] In some embodiments of the driving warning method of the present invention, before step (A), the size data of the first vehicle can be set in the driving warning system in a default manner through the processing unit according to the size reference data set of the vehicle, and the size of the second vehicle can also be extracted through the data set.

[0013] In some embodiments of the driving warning method of the present invention, in step (B), the processing unit determines whether the second vehicle is located directly behind or diagonally behind the first vehicle, including defining one or more lanes presented by the video data based on one or more road markings presented by the video data, and determining whether the second vehicle is located in the same lane as the first vehicle. If it is determined that the second vehicle is located in the same lane as the first vehicle, the second vehicle is determined to be directly behind the first vehicle. If it is determined that the second vehicle and the first vehicle are located in different lanes, the second vehicle is determined to be located diagonally behind the first vehicle.

[0014] In some embodiments of the driving warning method of the present invention, after step (A), the duration of the blind spot detection period is a default fixed number of seconds, or is defined as the time from when the second vehicle presenting as a large vehicle is determined to appear in the video data to when it is determined to disappear.

[0015] In some embodiments of the driving warning method of the present invention, the storage unit stores two left and right distance thresholds and two vehicle blind spot coverage thresholds, the two left and right distance thresholds are respectively a first left and right distance threshold, and a second left and right distance threshold greater than the first left and right distance threshold. In addition, the two vehicle blind spot coverage thresholds may be respectively a first vehicle blind spot coverage threshold, and a second vehicle blind spot coverage threshold less than the first vehicle blind spot coverage threshold. In step (B): when determining that the video data presents the second vehicle, the processing unit further determines whether the video data meets a turning intention condition, wherein the turning intention condition represents that the image or sound of the video data indicates an expected turning side related to the subsequent driving mode of the second vehicle, and the expected turning side is one of the left and right sides of the second vehicle. Step (D) further includes: (d1) after determining that the second vehicle is located obliquely behind the first vehicle, but the video data does not meet the turning intention condition, the processing unit determines whether the second blind spot warning condition is met based on the video data, and controls the output unit to output the second warning notification when the judgment result is yes, wherein the second blind spot warning condition includes that the distance between the first vehicle and the second vehicle is less than or equal to the first left-right distance threshold, and the area covered by the blind spot of the second vehicle on the first vehicle is greater than or equal to the blind spot coverage threshold of the first vehicle; (d2) after determining that the second vehicle is located obliquely behind the first vehicle, and the video data meets the turning intention condition, the processing unit determines whether the current position of the first vehicle is biased toward the second vehicle expected turning side; (d3) if it is determined that the current position of the first vehicle is not biased towards the expected turning side of the second vehicle, the processing unit determines whether the second blind spot warning condition is met based on the video data, and controls the output unit to output the second warning notification when the judgment result is yes; (d4) if it is determined that the current position of the first vehicle is biased towards the expected turning side of the second vehicle, the processing unit determines whether a third blind spot warning condition is met based on the video data, and controls the output unit to output a third warning notification when the judgment result is yes, wherein the third blind spot warning condition includes that the distance between the first vehicle and the second vehicle is less than or equal to the second left-right distance threshold, and that the area covered by the blind spot of the second vehicle on the first vehicle is greater than or equal to the blind spot coverage threshold of the first vehicle.

[0016] In some embodiments of the driving warning method of the present invention, in step (B), the processing unit determines whether the video data meets the turning intention condition in a manner that includes: determining whether only the turn signal on one side of the second vehicle is flashing based on the image of the video data; if the determination result is yes, the side of the left or right side of the second vehicle on which the turn signal is flashing is used as the expected turning side indicated by the video data, and determining that the video data meets the turning intention condition.

[0017] In some implementations of the driving warning method of the present invention, in step (B), the processing unit determines whether the video data meets the turning intention condition in a manner that includes: determining whether the sound of the video data contains a turning keyword indicating a turning side; if the determination result is yes, using the turning side indicated by the turning keyword as the expected turning side indicated by the video data, and determining that the video data meets the turning intention condition.

[0018] Compared with the prior art, the driving warning system and method of the present invention can determine whether the second vehicle is directly behind or obliquely behind the first vehicle when the first vehicle is in the stationary state and there is a large vehicle (i.e., the second vehicle) behind it. If the second vehicle is directly behind, the driving warning system will subsequently determine whether to output the first warning notification based on whether the first blind spot warning condition corresponding to the front-to-back distance threshold is met, so as to prevent the driver of the first vehicle from passively entering the blind spot in front of the second vehicle without knowing it. If the second vehicle is located obliquely behind, the driving warning system will subsequently determine whether to output the second warning notification based on whether the second blind spot warning condition corresponding to the left-right distance threshold and the blind spot coverage threshold of the vehicle is met, so as to prevent the driver of the first vehicle from passively entering the blind spot around the second vehicle without knowing it. Thereby, the driving warning system can actively detect whether the first vehicle is located in the blind spot and inner wheel difference danger range of the second vehicle when the first vehicle is approached from the rear by the second vehicle, and output a warning notification when the risk increases. Therefore, the driving warning system helps to increase the driver's alertness to the surrounding large vehicles, thereby reducing the risk of traffic accidents caused by blind spots of large vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Other features and effects of the present invention will be clearly presented in the embodiments with reference to the accompanying drawings, in which:

[0020] Figure 1 is a block diagram, exemplarily showing an embodiment of the driving warning system of the present invention;

[0021] Figure 2 is a flow chart for exemplarily illustrating how the vehicle warning method is implemented in the embodiment; and

[0022] Figure 3 It is a flow chart used to exemplarily illustrate how the driving warning method is implemented in this embodiment. DETAILED DESCRIPTION

[0023] Before the present invention is described in detail, it should be noted that, if not specifically defined, the "electrical connection" described in this patent specification generally refers to the "wired electrical connection" achieved by connecting multiple electronic devices / devices / components to each other through conductive materials, and the "radio connection" for single / bidirectional wireless signal transmission through wireless communication technology. In addition, the "electrical connection" described in this patent specification also generally refers to the "direct electrical connection" formed by directly connecting multiple electronic devices / devices / components to each other, and the "indirect electrical connection" formed by indirectly connecting multiple electronic devices / devices / components to each other through other electronic devices / devices / components.

[0024] See also Figure 1 An embodiment of the driving warning system 1 of the present invention is suitable for being applied to a vehicle. In a preferred application example of the driving warning system 1, the vehicle may be a motorcycle or a bicycle. However, the driving warning system 1 may also be applied to a car, and is not limited to motorcycles and bicycles. In addition, for the convenience of subsequent description, the vehicle to which the driving warning system 1 of this embodiment is applied is referred to as a first vehicle.

[0025] In the present embodiment, the driving warning system 1 is implemented as a digital video recording device (Digital Video Recorder, DVR for short) with image recognition and driving warning functions, that is, a driving recording device. Moreover, the driving warning system 1 of the present embodiment can be independently manufactured and sold, and, for example, is installed on the first vehicle after the first vehicle leaves the factory. However, in other embodiments, the driving warning system 1 is not limited to being implemented as a digital image recording device, but can be, for example, built into the first vehicle before the first vehicle leaves the factory, and used as a driving safety assistance system installed on the first vehicle. Therefore, the specific implementation of the driving warning system 1 is not limited to the present embodiment.

[0026] The driving warning system 1 comprises a processing unit 11, a storage unit 12, a camera unit 13 and an output unit 14 which are electrically connected to the processing unit 11. The processing unit 11 is a central processing unit with data calculation and processing functions in this embodiment. The storage unit 12 is a data storage device (such as a memory card or a solid state hard disk) for storing digital data. The camera unit 13 is a video recording module including a lens, a photosensitive element and a microphone and having video recording and sound receiving functions. In addition, the camera unit 13 is suitable for being arranged on the first vehicle in a manner of facing backwards, and is used to record and shoot toward the rear of the first vehicle. In addition, in some embodiments, the camera unit 13 may not include a microphone (i.e., it does not have a sound receiving function). The output unit 14 is a vibrator in this embodiment, however, in other embodiments, the output unit 14 may further include a speaker. Further, in the case where the first vehicle is a motorcycle or a bicycle, the output unit 14 is suitable for being arranged on the handle of the first vehicle, or on the helmet worn by the user when riding the first vehicle. In the case where the first vehicle is a car, the vibrator of the output unit 14 can be, for example, disposed in the steering wheel of the first vehicle. It is additionally noted that, in a preferred embodiment, in order to ensure that the processing unit 11 can receive the photographing result of the photographing unit 13 as quickly as possible, the processing unit 11 and the storage unit 12 are suitable for being disposed together with the photographing unit 13 on the first vehicle, and the processing unit 11 is electrically connected to the photographing unit 13 by wire, but the present invention is not limited thereto.

[0027] In the present embodiment, the storage unit 12 stores an image recognition model M implemented by machine learning technology and capable of being loaded and run by the processing unit 11, a blind spot distribution model group K predefined to correspond to blind spots around different large vehicles, a front-to-back distance threshold P1, two left-to-right distance thresholds P2 different from the front-to-back distance threshold P1, and a vehicle size reference data group L corresponding to the first and second vehicles. Moreover, in the present embodiment, the two left-to-right distance thresholds P2 are respectively a first left-to-right distance threshold P2A, and a second left-to-right distance threshold P2B greater than the first left-to-right distance threshold P2A. In addition, the storage unit 12 also has two vehicle blind spot coverage thresholds corresponding to the blind spots around the vehicle, wherein the two vehicle blind spot coverage thresholds are respectively a first vehicle blind spot coverage threshold P3A, and a second vehicle blind spot threshold P3B less than the first vehicle blind spot coverage threshold.

[0028] In this embodiment, the image recognition model M is trained by performing deep learning using at least a plurality of large vehicle photos and a plurality of road photos showing road markings as training image data.

[0029] More specifically, the large vehicle photos are, for example, photos of various large vehicles taken from multiple different angles (e.g., front, side, and oblique front), and the large vehicles include, for example, but are not limited to large buses, large trucks, and articulated vehicles defined by government agencies. In addition, during the training process of the image recognition model M, the large vehicle photos are used to allow the image recognition model M to learn the appearance features of large vehicles, and the appearance features include, for example, the front appearance features and side appearance features of large vehicles. For example, the front appearance features may include, for example, but are not limited to, the shapes of the components such as the windshield, rear-view mirror, turn signal, headlight, speed indicator, etc. on the front of the large vehicle and the relative positional relationship between each other, while the side appearance features may include, for example, but are not limited to, the shapes of the components such as the doors, windows, tires, anti-roll devices, and truck beds on the sides of the large vehicle and the relative positional relationship between each other. Thus, after deep learning is performed based on the photos of large vehicles to learn the appearance features of large vehicles, when an image (such as a video) shows at least a portion of the front or body of a large vehicle, the image recognition model M can recognize the large vehicle presented in the image. It should be understood that the actual appearance features learned by the image recognition model M through deep learning are not the technical focus of this case. Therefore, the aforementioned appearance features are only used for illustrative purposes and are not used to limit the implementable aspects of the image recognition model M.

[0030] On the other hand, each road photo is, for example, a screenshot of a driving record showing one or more road markings for separating lanes, such as a directional restriction line, a no-overtaking line, a no-lane-change line, a driving direction dividing line, a lane line, etc., but is not limited thereto. Moreover, during the training process of the image recognition model M, the road photos are used to allow the image recognition model M to learn the appearance features of the road markings presented in the shooting angle of the driving recorder. Thereby, after deep learning is performed based on the road photos to learn the appearance features of various road markings (especially the appearance features of road markings belonging to lane dividing lines), when an image (such as a video) presents one or more road markings for separating lanes, the image recognition model M can recognize from the image one or more lanes defined by the road markings presented therein.

[0031] It should be noted that the use of deep learning to achieve computer vision recognition of specific targets (such as the aforementioned large vehicles and lanes) can be achieved using existing technologies, so this embodiment does not describe the training process of the image recognition model M in excessive detail.

[0032] In this embodiment, the front-to-rear distance threshold P1 is used to indicate a front-to-rear safety distance that the first vehicle should maintain when it stops in front of a large vehicle in order to avoid falling into the blind spot area in front of the large vehicle. The first left-right distance threshold P2A and the first vehicle blind spot coverage threshold P3A are used to measure whether the first vehicle passively falls into the blind spot of the second vehicle (e.g., the area around the vehicle that is easily ignored or cannot be directly observed outside the illumination angle of the rearview mirror, etc.) when the second vehicle is on the left or right side of the first vehicle, thereby reminding the first vehicle to pay attention to the driver's driving. The second left-right distance threshold P2B and the first vehicle blind spot coverage threshold P3B are used to further evaluate the possibility that the first vehicle may passively fall into the more dangerous inner wheel difference range of the turn from the blind spot and give a reminder.

[0033] In an actual implementation of this embodiment, the front-to-back distance threshold P1 is 1.6 meters, the first left-right distance threshold P2A is 2 meters, and the second left-right distance threshold P2B is 4 meters. The first vehicle blind spot coverage threshold P3A is 60% of the area of ​​the first vehicle falling into the vehicle blind spot model area, and the second vehicle blind spot coverage threshold P3B is 40%. However, it should be understood that the specific values ​​of the front-to-back distance threshold P1, the first left-right distance threshold P2A, and the second left-right distance threshold P2B can be freely set and adjusted according to different needs. In different embodiments, the first left-right distance threshold P2A can be implemented as long as it is different from the value of the front-to-back distance threshold P1, and the second left-right distance threshold P2B can be implemented as long as it is greater than the first left-right distance threshold P2A, and the second vehicle blind spot coverage threshold P3B can be implemented as long as it is less than the first vehicle blind spot coverage threshold P3A.

[0034] Related References Figure 2 and Figure 3 The following is an exemplary detailed description of how the driving warning system 1 of this embodiment implements a driving warning method during the driving of the first vehicle.

[0035] First, in step S1, the processing unit 11 determines whether the first vehicle is in a stationary state without moving. More specifically, in this embodiment, the processing unit 11, for example, first controls the shooting unit 13 to start recording in step S1, so as to obtain a video data shot by the shooting unit 13 (i.e., a real-time video recording of the shooting unit 13) from the shooting unit 13, and then determines whether the first vehicle is in the stationary state according to the video data. Since the shooting unit 13 shoots toward the rear of the first vehicle, the video data will present the environment behind the first vehicle. Moreover, since the shooting unit 13 of this embodiment has a sound receiving function, the video data has a sound part in addition to the image part in this embodiment, but it is not limited thereto. In addition, in different embodiments, the processing unit 11 can also determine whether the first vehicle is in the stationary state according to the activity sensing data generated by an activity sensor (e.g., an accelerometer) electrically connected to the processing unit 11 itself, so the way in which the processing unit 11 determines whether the first vehicle is in the stationary state is not limited to this embodiment.

[0036] If the processing unit 11 determines that the first vehicle is not in the stationary state (i.e., the first vehicle is moving), the process starts again from step S1, for example, when a delay period of a predetermined length of time ends. If the processing unit 11 determines that the first vehicle is indeed in the stationary state (e.g., it is not moving because it is waiting for a red light), the process proceeds to step S2.

[0037] In step S2 following step S1, when it is determined that the first vehicle is indeed in the stationary state, the processing unit 11 performs image recognition on the image portion of the video data by running the image recognition model M to determine whether the video data presents a second vehicle that is a large vehicle. If the processing unit 11 determines that the video data does not present any large vehicle, the process starts again from step S1, for example, at the end of a delay period with a predetermined time length. If the processing unit 11 determines that the video data does present a second vehicle that is a large vehicle, the process proceeds to step S3. It should be noted that in other embodiments, the processing unit 11 does not necessarily have to determine whether the video data presents a second vehicle that is a large vehicle when the first vehicle is in the stationary state. In other words, step S1 can be omitted in other embodiments.

[0038] In step S3 following step S2, when it is determined that the video data does present the second vehicle with image features belonging to a large vehicle, a blind spot detection period is entered, and the processing unit 11 determines whether the second vehicle is located directly behind or obliquely behind the first vehicle based on the image portion of the video data by running the image recognition model M. In addition, the size of the first vehicle can be set by the processing unit 11 in a default manner before the process is executed based on the size reference data set L of the vehicle.

[0039] In more detail, in the present embodiment, the processing unit 11 determines whether the second vehicle is currently located directly behind or diagonally behind the first vehicle by first running the image recognition model M to identify one or more road surface markings belonging to lane separators presented in the image portion of the video data, and defines one or more lanes presented by the video data based on the road surface markings. Next, the processing unit 11 determines whether the second vehicle is located in the same lane as the first vehicle based on the image portion of the video data. If it is determined that the second vehicle and the first vehicle are located in the same lane, the processing unit 11 will determine that the second vehicle is currently located directly behind the first vehicle, and if it is determined that the second vehicle and the first vehicle are located in different lanes, the processing unit 11 will determine that the second vehicle is currently located obliquely behind (i.e., behind the left or behind the right) the first vehicle. Then, the processing unit 11 generates a two-dimensional plane top view of the current distance and orientation of the first vehicle and the second vehicle based on the video data and the relevant vehicle data extracted from the vehicle data database L. Then, the processing unit 11 extracts a corresponding blind spot distribution model from the vehicle blind spot distribution model set K according to the characteristics of the second vehicle, and overlays the blind spot distribution model on the position of the second vehicle in the top view, and continuously calculates the distance between the second vehicle and the first vehicle in the top view and the area of ​​the overlapping area in which the first vehicle may fall in the blind spot distribution model. More specifically, the two-dimensional plane top view includes a first diagram corresponding to the first vehicle and a second diagram corresponding to the second vehicle. The size of the first diagram is determined by, for example, reducing the actual length and width of the first vehicle in proportion, and the size of the second diagram is obtained by the processing unit 11 from the vehicle data database L, and the size of the second diagram is determined by reducing the length and width of the vehicle type corresponding to the second vehicle in proportion. In addition, the relative positional relationship between the first diagram and the second diagram in the two-dimensional plane top view presents the current relative positional relationship (i.e., distance and orientation) between the first vehicle and the second vehicle in the real environment. On the other hand, the blind spot distribution model is, for example, superimposed by the processing unit 11 as a layer on the position corresponding to the second diagram in the two-dimensional plane top view, so that the two-dimensional plane top view and the blind spot distribution model can jointly present the blind spot range of the second vehicle and the current relative positional relationship between the blind spot ranges of the first vehicle and the second vehicle. In addition, the processing unit 11 continuously calculates the overlapping area between the first diagram and the blind spot distribution model to estimate how much of the first vehicle currently enters the blind spot range of the second vehicle in the real environment. Among them, simplifying the image data in space into a two-dimensional plane top view by projection, conversion of coordinate system and other techniques is a known technique in image processing. It is not the main feature of the invention in this case, so it will not be described here in detail.

[0040] However, in another embodiment, the processing unit 11 may also determine whether the second vehicle is located directly behind or obliquely behind the first vehicle based on the position of the second vehicle presented in the image portion of the video data. Specifically, the image portion of the video data, for example, has a central area that is predefined and arranged horizontally with each other, a left area arranged on the left side of the central area, and a right area arranged on the right side of the central area. Furthermore, if a center point of the front of the second vehicle is located within the central area of ​​the image portion of the video data, the processing unit 11 will determine that the second vehicle is currently located directly behind the first vehicle, and if the center point of the front of the second vehicle is located outside the central area of ​​the image portion of the video data (i.e., located within the left area or the right area of ​​the image portion of the video data), the processing unit 11 will determine that the second vehicle is currently located obliquely behind the first vehicle. Therefore, the manner in which the processing unit 11 determines whether the second vehicle is currently located directly behind or obliquely behind the first vehicle is not limited to this embodiment.

[0041] If the processing unit 11 determines that the second vehicle is located directly behind the first vehicle, the process proceeds to step S4. If the processing unit 11 determines that the second vehicle is located obliquely behind the first vehicle, the process proceeds to step S7.

[0042] In step S4 following step S3, after determining that the second vehicle is currently located directly behind the first vehicle, the processing unit 11 continuously determines whether a first blind spot warning condition corresponding to the front-rear distance threshold P1 is met based on the image portion of the video data within a blind spot detection period. The length of the blind spot detection period may be a default fixed number of seconds (e.g., 12 seconds, but not limited thereto), or may be defined as the time from when the second vehicle presenting as a large vehicle is determined to appear in the video data to when it is determined to disappear. On the other hand, the first blind spot warning condition in this embodiment represents that the first vehicle is still in the stationary state, and the distance between the first vehicle and the second vehicle is less than or equal to the front-rear distance threshold P1. It should be supplemented that, in some embodiments, the first blind spot warning condition does not necessarily involve whether the first vehicle is still in the stationary state.

[0043] If the processing unit 11 determines that the first blind spot warning condition is met before the blind spot detection period ends, the process proceeds to step S5. If the processing unit 11 determines that the first blind spot warning condition is not met before the blind spot detection period ends, the process proceeds to step S6.

[0044] In step S5 following step S4, once the processing unit 11 determines that the first blind spot warning condition is met, it means that the first vehicle is likely to be in the blind spot of the vehicle in front of the second vehicle. This is because when the first vehicle is located directly in front of the second vehicle, the blind spot of the second vehicle is simply in an area in front of the front of the vehicle. Therefore, it is possible to determine whether the first vehicle has fallen into the blind spot and calculate the area that may fall into the area based on the distance. In this case, the processing unit 11 controls the output unit 14 to output a first warning notification to warn the driver of the first vehicle. In this embodiment, the first warning notification is a vibration emitted by the vibrator, but it is not limited to this.

[0045] In step S6 following step S4, once the processing unit 11 determines that the first blind spot warning condition is not met until the end of the blind spot detection period, it means that the probability that the first vehicle is located in the blind spot of the second vehicle's driving vision is relatively low. In this case, the processing unit 11 determines whether the first vehicle is still in the stationary state and whether the video data still presents at least a portion of the second vehicle (i.e., whether the second vehicle is still partially present in the shooting range of the shooting unit 13). If the judgment result of the processing unit 11 is yes, the process, for example, starts again from step S3. If the judgment result of the processing unit 11 is no (i.e., the first vehicle has started to move, or the video data no longer presents the second vehicle), the process starts again from step S1.

[0046] In step S7 following step S3, once it is determined that the second vehicle is currently located obliquely behind the first vehicle, it means that the second vehicle may continue to move to the side of the first vehicle and be parallel to the first vehicle. In this case, the processing unit 11 determines whether the video data meets a turning intention condition related to the second vehicle based on the image part and the sound part of the video data, wherein the turning intention condition represents that the image part or the sound part of the video data indicates an expected turning side relative to the second vehicle and related to the subsequent driving mode of the second vehicle. The expected turning side is one of the left and right sides of the second vehicle, and is used to represent the direction in which the second vehicle is more likely to turn in the subsequent moving mode.

[0047] More specifically, since the image portion of the video data presents the environment behind the first vehicle, when the second vehicle enters the shooting range of the shooting unit 13, the video data has a chance to present the left and right turn signals located at the front of the second vehicle. Based on the above, in this embodiment, the processing unit will determine whether only one side of the turn signal of the second vehicle is flashing according to the image portion of the video data. If the result of the judgment of the processing unit 11 is yes, the processing unit 11 will take the side of the left and right sides of the second vehicle where the turn signal is flashing as the expected turning side indicated by the video data, and determine that the video data meets the turning intention condition accordingly. Specifically, if the processing unit 11 determines that the right turn signal of the second vehicle is flashing and the left turn signal is not flashing according to the image portion of the video data, the processing unit 11 will take the right side of the second vehicle as the expected turning side indicated by the video data, and determine that the video data meets the turning intention condition accordingly. On the contrary, if the processing unit 11 determines based on the image portion of the video data that the left turn signal of the second vehicle is flashing but the right turn signal is not flashing, the processing unit 11 will take the left side of the second vehicle as the intended turning side indicated by the video data, and determine accordingly that the video data meets the turning intention condition.

[0048] On the other hand, in view of the fact that some large vehicles currently play turning alarm voices through speakers (for example, "The vehicle is turning right, please pay attention", "Turn right, please be careful", etc.), therefore, in this embodiment, the processing unit 11 will also determine whether there is a turning keyword indicating a turning side (for example, "turn right" or "turn left") in the sound part of the video data. If the result of the judgment of the processing unit 11 is yes, the processing unit 11 will use the turning side indicated by the turning keyword as the expected turning side indicated by the video data, and determine that the video data meets the turning intention condition accordingly.

[0049] It is additionally noted that in other embodiments, the processing unit 11 may also determine whether the video data meets the turning intention condition based on only one of the image part and the sound part of the video data, and is not limited to this embodiment.

[0050] If the processing unit 11 determines that the video data does not meet the turning intention condition, the process proceeds to step S8. If the processing unit 11 determines that the video data does meet the turning intention condition, the process proceeds to step S11.

[0051] In step S8 following step S7, after determining that the second vehicle is currently located obliquely behind the first vehicle, but the video data does not meet the turning intention condition, the processing unit 11 continuously determines whether a second blind spot warning condition corresponding to the first left-right distance threshold P2A and the first vehicle blind spot coverage threshold P3A is met according to the video data within a blind spot detection period. The length of the blind spot detection period can be a default fixed number of seconds, or defined as the time from when the second vehicle presenting as a large vehicle is determined to appear in the video data to when it is determined to disappear. On the other hand, the second blind spot warning condition in this embodiment represents that the first vehicle is still in the stationary state, and the distance between the first vehicle and the second vehicle is less than or equal to the first left and right distance threshold P2A, and the first vehicle is in the blind spot distribution model of the second vehicle and the coverage area is greater than or equal to the blind spot coverage threshold P3A of the first vehicle, that is, through the distance and orientation of the two vehicles, it is comprehensively judged whether the first vehicle passively falls into the blind spot of the second vehicle and its falling area, and at the same time, the two vehicles are below a safe distance, reaching the need to trigger an alarm. It is supplemented that, in some embodiments, the second blind spot warning condition does not necessarily involve whether the first vehicle is still in the stationary state. If the processing unit 11 determines that the second blind spot warning condition is met before the end of the blind spot detection period, the process proceeds to step S9. If the processing unit 11 determines that the first blind spot warning condition is not met until the end of the blind spot detection period, the process proceeds to step S10.

[0052] In step S9 following step S8, once the processing unit 11 determines that the second blind spot warning condition is met, it means that the first vehicle is likely to be in the blind spot of the left or right side of the second vehicle. In this case, the processing unit 11 controls the output unit 14 to output a second warning notification to warn the driver of the first vehicle. In this embodiment, the second warning notification is, for example, the vibration emitted by the vibrator, but is not limited thereto.

[0053] In step S10 following step S8, once the processing unit 11 determines that the second blind spot warning condition is not met until the end of the blind spot detection period, it means that the first vehicle is relatively less likely to be located in the blind spot of the second vehicle's driving vision. In this case, the processing unit 11 determines whether the first vehicle is still in the stationary state and whether the video data still presents at least a portion of the second vehicle. If the determination result of the processing unit 11 is yes, the process, for example, starts again from step S3. If the determination result of the processing unit 11 is no, the process starts again from step S1.

[0054] In step S11 following step S7, after determining that the second vehicle is currently located obliquely behind the first vehicle and the video data meets the turning intention condition, the processing unit 11 determines whether the current position of the first vehicle is biased toward the expected turning side of the second vehicle based on the image portion of the video data. Specifically, if the expected turning side is the right side of the second vehicle, and the image portion of the video data shows that the second vehicle is located at the left rear or left side of the first vehicle (equivalent to the first vehicle being located at the right front or right side of the second vehicle), the processing unit 11 will determine that the current position of the first vehicle is indeed biased toward the expected turning side of the second vehicle. Similarly, if the expected turning side is the left side of the second vehicle, and the image portion of the video data shows that the second vehicle is located at the right rear or right side of the first vehicle (equivalent to the first vehicle being located at the left front or left side of the second vehicle), the processing unit 11 will also determine that the current position of the first vehicle is indeed biased toward the expected turning side of the second vehicle.

[0055] If the processing unit 11 determines that the current position of the first vehicle is not biased toward the expected turning side of the second vehicle, the process proceeds to step S12. If the processing unit 11 determines that the current position of the first vehicle is indeed biased toward the expected turning side of the second vehicle, the process proceeds to step S15.

[0056] In step S12 following step S11, once the processing unit 11 determines that the second vehicle is located obliquely behind the first vehicle and the video data meets the turning intention condition, but the first vehicle does not deviate to the expected turning side of the second vehicle, it means that the second vehicle may continue to move forward to the side of the first vehicle and has a relatively high chance of turning toward the side away from the first vehicle. In other words, if the second vehicle continues to move forward, it may cause the first vehicle to enter the blind spot of the second vehicle, but it will not cause the first vehicle to enter the inner wheel difference danger range when the second vehicle turns toward the expected turning side. In this case, the processing unit 11 continuously determines whether the second blind spot warning condition is met according to the video data within a blind spot detection period. The time length of the blind spot detection period can be a default fixed number of seconds, or defined as the second vehicle presenting a large vehicle is determined to appear in the video data until it is determined to disappear. If the processing unit 11 determines that the second blind spot warning condition is met before the end of the blind spot detection period, the process proceeds to step S13. If the processing unit 11 determines that the second blind spot warning condition is not met, the process proceeds to step S14.

[0057] In step S13 following step S12, once the processing unit 11 determines that the second blind spot warning condition is met, it means that the first vehicle is very likely to be in the blind spot of the vehicle to the left or right of the second vehicle. In this case, the processing unit 11 controls the output unit 14 to output the second warning notification to alert the driver of the first vehicle.

[0058] In step S14 following step S12, once the processing unit 11 determines that the second blind spot warning condition has not been met until the end of the blind spot detection period, it means that the first vehicle is relatively less likely to be located in the blind spot of the second vehicle. In this case, the processing unit 11 determines whether the first vehicle is still in the stationary state and whether the video data still presents at least a portion of the second vehicle. If the determination result of the processing unit 11 is yes, the process, for example, starts again from step S3. If the determination result of the processing unit 11 is no, the process starts again from step S1.

[0059] In step S15 following step S11, once the processing unit 11 determines that the second vehicle is located obliquely behind the first vehicle, the video data meets the turning intention condition, and the first vehicle is indeed biased toward the expected turning side of the second vehicle, it means that the second vehicle may continue to move forward to the side of the first vehicle, and there is a relatively high chance of turning toward the side close to the first vehicle. In other words, if the second vehicle continues to move forward, it may not only cause the first vehicle to enter the blind spot on the side of the second vehicle, but also cause the first vehicle to enter the inner wheel difference danger range when the second vehicle turns toward the expected turning side. In this case, the processing unit 11 determines whether a third blind spot warning condition is met based on the video data within a blind spot detection period. The time length of the blind spot detection period can be a default fixed number of seconds, or defined as the second vehicle presenting a large vehicle is determined to appear in the video data until it is determined to disappear. On the other hand, the third blind spot warning condition in this embodiment represents that the first vehicle is still in the stationary state, and the distance between the first vehicle and the second vehicle is less than or equal to the second left-right distance threshold P2B, and the first vehicle is in the blind spot distribution model of the second vehicle and the coverage area is greater than or equal to the blind spot coverage threshold P3B of the first vehicle, that is, when the second vehicle has the intention to turn toward one side of the first vehicle, a shorter threshold and a smaller blind spot coverage area are used to improve the sensitivity of the first vehicle to passively fall into the blind spot of the second vehicle, so that the system can trigger a warning more real-time. It is supplemented that in some embodiments, the third blind spot warning condition does not necessarily involve whether the first vehicle is still in the stationary state. If the processing unit 11 determines that the third blind spot warning condition is met before the end of the blind spot detection period, the process proceeds to step S16. If the processing unit 11 determines that the third blind spot warning condition is not met until the end of the blind spot detection period, the process proceeds to step S17.

[0060] In step S16 following step S15, once the processing unit 11 determines that the third blind spot warning condition is met, it means that the first vehicle is not only very likely to be stopped in the blind spot on the side of the second vehicle, but may also be located in the inner wheel difference danger range when the second vehicle turns toward the expected turning side. In this case, the processing unit 11 controls the output unit 14 to output a third warning notification to warn the driver of the first vehicle. In this embodiment, the third warning notification is, for example, also the vibration emitted by the vibrator, and, preferably, the vibration intensity and / or continuous vibration time of the third warning notification is, for example, greater than the second warning notification, thereby achieving a stronger warning effect than the second warning notification. It should be supplemented that, in the embodiment where the output unit 14 also includes a speaker, the third warning notification may also include a warning sound emitted by the speaker to further enhance the warning effect.

[0061] In step S17 following step S15, once the processing unit 11 determines that the third blind spot warning condition is not met until the end of the blind spot detection period, it means that the first vehicle is relatively less likely to be located in the blind spot of the second vehicle. In this case, the processing unit 11 determines whether the first vehicle is still in the stationary state and whether the video data still presents at least a portion of the second vehicle. If the determination result of the processing unit 11 is yes, the process, for example, starts again from step S3. If the determination result of the processing unit 11 is no, the process starts again from step S1.

[0062] The above is an example description of how the driving warning system 1 of this embodiment implements the driving warning method.

[0063] It should be understood that steps S1 to S17 of this embodiment and Figure 2 , Figure 3 The flowchart is only used to illustrate one possible implementation method of the driving warning method of the present invention. It should be understood that even if steps S1 to S17 are combined, split or adjusted in order, if the process after the combination, split or adjustment in order achieves substantially the same effect as the present embodiment in substantially the same manner, it still belongs to the implementable aspect of the driving warning method of the present invention. Therefore, steps S1 to S17 and Figure 2 , Figure 3 The flowchart is not intended to limit the applicable scope of the present invention.

[0064] It is to be noted that the image recognition model M of the present embodiment records the size information of various large vehicles, for example, and the size information may include but is not limited to the width and / or height of the front of various large vehicles, the width and / or height of the windshield, the width of a single headlight, the lateral spacing between two headlights, the length of the vehicle body, the diameter of the tire (which may be the inner diameter or the outer diameter), or the size of other appearance features themselves, and the relative position relationship between multiple appearance features, etc. In addition, the processing unit 11 calculates the distance between the first vehicle and the second vehicle, for example, by running the image recognition model M, according to the pixel size of each appearance feature of the second vehicle presented in the image part of the video data, the size information recorded in the image recognition model M, and the focal length and photosensitive element size of the shooting unit 13. It should be understood that the processing unit 11 can calculate the distance between the first vehicle and the second vehicle according to the video data by using existing technology, so this embodiment will not be described in detail. However, in other embodiments, the image recognition model M may also be a machine learning model that uses a depth camera to perform machine learning on the shooting results of various large vehicles in advance, so as to have the function of measuring the distance of large vehicles in two-dimensional images. Therefore, any technology that can measure the distance of the photographed objects based on two-dimensional images can be applied to the driving warning system 1 of the present invention.

[0065] In another embodiment of the driving warning system 1 of the present invention, the storage unit 12 stores only a single left-right distance threshold value P2 (for example, Figure 1 In addition, in the driving warning method implemented in the embodiment, if the processing unit 11 determines in step S3 that the second vehicle is located obliquely behind the first vehicle, the process directly proceeds to step S8. That is, the driving warning method implemented in the embodiment does not include Figure 2 Step S7 in the Figure 3 In other words, step S7 and steps S11 to S17 may be omitted in some embodiments.

[0066] In another embodiment of the driving warning system 1 of the present invention, the blind spot detection periods described in step S4, step S8, step S12 and step S15 may also continue until the processing unit 11 determines that the video data no longer presents the second vehicle. In other words, the length of the blind spot detection periods is not necessarily pre-set. Moreover, in this embodiment, the processing unit 11 only needs to determine whether the first vehicle is still in the stationary state in steps S6, step S10, step S14 and step S17 respectively following step S4, step S8, step S12 and step S15, without determining whether the video data still presents at least a portion of the second vehicle.

[0067] By implementing the driving warning method, the driving warning system 1 can determine whether the second vehicle is located directly behind or obliquely behind the first vehicle when the first vehicle is in the stationary state and there is a large vehicle (i.e., the second vehicle) behind it (step S3). If the second vehicle is located directly behind, the driving warning system 1 will subsequently determine whether to output the first warning notification based on whether the first blind spot warning condition corresponding to the front-to-back distance threshold P1 is met, so as to prevent the driver of the first vehicle from passively entering the blind spot in front of the second vehicle without knowing it. If the second vehicle is located obliquely behind, the driving warning system 1 will subsequently determine whether to output the second warning notification based on whether the second blind spot warning condition corresponding to the first left-right distance threshold P2A and the first blind spot coverage threshold P3A is met, so as to prevent the driver of the first vehicle from passively entering the blind spots on both sides of the second vehicle without knowing it. Furthermore, the driving warning system 1 can determine whether the second vehicle has a higher probability of turning toward the expected turning side in the future (step S7) based on the image part and the sound part of the video data, and determine whether the current position of the first vehicle is biased toward the expected turning side of the second vehicle (step S11). Thereby, the driving warning system 1 can evaluate whether the first vehicle is likely to enter the inner wheel difference danger range when the second vehicle turns toward the expected turning side, and when the judgment result is yes, the third blind spot warning condition is determined to be met based on the second left and right distance threshold P2B that is greater than the first left and right distance threshold and less than the second blind spot coverage threshold P3B, so as to decide whether to output the third warning notification. It should be noted that in the prior art, the blind spot range of a large vehicle can usually be pre-defined as a specific spatial range defined from a specific part of a large vehicle (such as a rearview mirror) toward a specific angle. Although the dash cam shooting results of a general small vehicle (such as the first vehicle in this embodiment) (such as the video data of this embodiment) can roughly present the relative position relationship between the large vehicle (such as the second vehicle in this embodiment) and the small vehicle itself, the accuracy of estimating the relative angle relationship between the two vehicles using only image recognition technology is limited, that is, it may not be sufficient to fully and accurately identify whether the small vehicle has entered the blind spot range and area of ​​the driving vision of the large vehicle.However, by implementing the driving warning method, this embodiment will determine which distance threshold to use based on the direction of the second vehicle relative to the first vehicle (directly behind or diagonally behind). If the two vehicles are in a relative position directly in front, the front-to-back distance threshold P1 is sufficient to determine whether the first vehicle passively falls into the blind spot in front of the second vehicle and whether there is a safe distance. If the two vehicles are in a relative position diagonally in front, the captured spatial image data is further combined with the data in the database in the storage unit to generate a two-dimensional plane top view of the distance, direction and blind spot of the first and second vehicles, and then the left and right distance threshold P2 is combined with the vehicle blind spot coverage threshold P3 to comprehensively determine whether the first vehicle passively falls into the vehicle blind spot of the second vehicle and whether there is a safe distance, as a basis for outputting a warning notification. Therefore, even though the output of the warning notification in this embodiment does not mean that the first vehicle is necessarily located in the driving blind spot of the second vehicle, this embodiment can indeed reflect the situation that the first vehicle has a relatively high probability of entering the driving blind spot of the second vehicle. Therefore, this embodiment provides a highly reliable driving safety warning mechanism for the driving blind spot of large vehicles by detecting the distance and direction between the two vehicles.

[0068] In summary, the driving warning system 1 can actively detect the risk of the first vehicle being located in the blind spot of the driving vision and the inner wheel difference danger range of the second vehicle when the first vehicle stops and the second vehicle approaches the first vehicle from the rear, and output a warning notification when the risk increases. Therefore, the driving warning system 1 helps to increase the driver's alertness to the surrounding large vehicles, thereby reducing the risk of traffic accidents caused by the blind spots of large vehicles, and thus can indeed achieve the purpose of the present invention.

[0069] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A driving warning system, suitable for use in a first vehicle; characterized in that: The driving warning system includes: a storage unit storing a front-to-back distance threshold, a left-to-right distance threshold different from the front-to-back distance threshold, a vehicle blind spot coverage threshold, a vehicle blind spot distribution model group, and a vehicle size reference data group; a photographing unit, adapted to be disposed on the first vehicle and used for photographing the rear of the first vehicle; an output unit; and a processing unit, electrically connected to the storage unit, the photographing unit and the output unit, and configured to: determining whether the video data captured by the capturing unit shows a second vehicle that is a large vehicle; If it is determined that the video data shows the second vehicle, determining whether the second vehicle is located directly behind or obliquely behind the first vehicle according to the video data; After determining that the second vehicle is located directly behind the first vehicle, determining whether a first blind spot warning condition corresponding to the front-rear distance threshold is met according to the video data, and controlling the output unit to output a first warning notification when the determination result is yes; and After determining that the second vehicle is located obliquely behind the first vehicle, the processing unit generates a two-dimensional plane top view of the distance and direction between the first vehicle and the second vehicle based on the film data and the vehicle size reference data group. At the same time, the processing unit extracts a blind spot distribution model corresponding to the second vehicle from the vehicle blind spot distribution model group, and overlays the blind spot distribution model on the position of the second vehicle in the two-dimensional plane top view. The processing unit determines whether a second blind spot warning condition corresponding to the left and right distance threshold and the vehicle blind spot coverage threshold is met, and controls the output unit to output a second warning notification when the judgment result is yes.

2. The driving warning system according to claim 1, characterized in that: The processing unit determines whether the second vehicle is located directly behind or diagonally behind the first vehicle, including defining one or more lanes presented by the video data based on one or more road markings presented by the video data, and determining whether the second vehicle is located in the same lane as the first vehicle. If it is determined that the second vehicle is located in the same lane as the first vehicle, the second vehicle is determined to be directly behind the first vehicle; if it is determined that the second vehicle and the first vehicle are located in different lanes, the second vehicle is determined to be located diagonally behind the first vehicle.

3. The driving warning system according to claim 1, characterized in that: When the processing unit determines that the video data presents the second vehicle, it enters a blind spot detection period. The duration of the blind spot detection period is a default fixed number of seconds, or is defined as the time from when the second vehicle presenting a large vehicle is determined to appear in the video data to when it is determined to disappear.

4. The driving warning system according to claim 1, characterized in that: The storage unit stores two left-right distance thresholds and two vehicle blind spot coverage thresholds, wherein the two left-right distance thresholds are respectively a first left-right distance threshold and a second left-right distance threshold greater than the first left-right distance threshold; the two vehicle blind spot coverage thresholds are respectively a first vehicle blind spot coverage threshold and a second vehicle blind spot coverage threshold less than the first vehicle blind spot coverage threshold; In the case where it is determined that the video data shows the second vehicle, the processing unit is further used to determine whether the video data meets a turning intention condition; and after determining that the second vehicle is located obliquely behind the first vehicle, the processing unit operates in a manner including: When the video data does not meet the turning intention condition, determining whether the second blind spot warning condition is met according to the video data, and controlling the output unit to output the second warning notification when the determination result is yes; When the video data meets the turning intention condition, determining whether the current position of the first vehicle is biased toward the expected turning side of the second vehicle; If it is determined that the current position of the first vehicle is not biased toward the expected turning side of the second vehicle, determining whether the second blind spot warning condition is met according to the video data, and controlling the output unit to output the second warning notification when the determination result is yes; If it is determined that the current position of the first vehicle is biased toward the expected turning side of the second vehicle, it is determined whether a third blind spot warning condition is met according to the video data, and when the determination result is yes, the output unit is controlled to output a third warning notification.

5. The driving warning system according to claim 4, characterized in that: The processing unit determines whether the video data meets the turning intention condition in the following manner: determining whether only the turn signal on one side of the second vehicle is flashing based on the image of the video data, or determining whether the sound of the video data contains a turning keyword indicating a turning side; if the determination result is yes, the side on the left or right side of the second vehicle where the turn signal is flashing is used as the expected turning side indicated by the video data, or the turning side indicated by the turning keyword is used as the expected turning side indicated by the video data, and the video data is determined to meet the turning intention condition.

6. A driving warning method, implemented by a driving warning system, the driving warning system is suitable for use in a first vehicle, and comprises a storage unit, a shooting unit suitable for being arranged in the first vehicle and used for shooting toward the rear of the first vehicle, an output unit, and a processing unit electrically connected to the storage unit, the shooting unit and the output unit, the storage unit storing a front-to-back distance threshold, a left-to-right distance threshold different from the front-to-back distance threshold, a vehicle blind spot coverage threshold, a vehicle blind spot distribution model group, and a vehicle size reference data group; characterized in that, The driving warning method includes: (A) the processing unit determines whether the video data captured by the capturing unit shows a second vehicle that is a large vehicle; (B) when it is determined that the video data shows the second vehicle, the processing unit determines, based on the video data, whether the second vehicle is located directly behind or obliquely behind the first vehicle; (C) after determining that the second vehicle is located directly behind the first vehicle, the processing unit determines whether a first blind spot warning condition corresponding to the front-rear distance threshold is met based on the video data, and controls the output unit to output a first warning notification when the determination result is yes; and (D) After determining that the second vehicle is located obliquely behind the first vehicle, the processing unit generates a two-dimensional plane top view of the distance and direction between the first vehicle and the second vehicle based on the film data and the vehicle size reference data group. At the same time, the processing unit determines that the second vehicle is a large vehicle according to the image features of step (A), extracts a blind spot distribution model corresponding to the second vehicle from the vehicle blind spot distribution model group, and overlays the blind spot distribution model on the position of the second vehicle in the top view. The processing unit determines whether a second blind spot warning condition corresponding to the left and right distance threshold and the blind spot coverage threshold of the vehicle is met, and controls the output unit to output a second warning notification when the judgment result is yes.

7. The driving warning method according to claim 6, characterized in that: In step (B), the processing unit determines whether the second vehicle is located directly behind or diagonally behind the first vehicle, including defining one or more lanes presented by the video data based on one or more road markings presented by the video data, and determining whether the second vehicle is located in the same lane as the first vehicle. If it is determined that the second vehicle is located in the same lane as the first vehicle, the second vehicle is determined to be directly behind the first vehicle; if it is determined that the second vehicle is located in different lanes than the first vehicle, the second vehicle is determined to be located diagonally behind the first vehicle.

8. The driving warning method according to claim 6, characterized in that: After step (A), the processing unit enters a blind spot detection period when it is determined that the video data presents the second vehicle. The duration of the blind spot detection period is a default fixed number of seconds, or is defined as the time from when the second vehicle presenting a large vehicle is determined to appear in the video data to when it is determined to disappear.

9. The driving warning method according to claim 6, characterized in that: The storage unit stores two left and right distance thresholds and two vehicle blind spot coverage thresholds, wherein the two left and right distance thresholds are respectively a first left and right distance threshold and a second left and right distance threshold greater than the first left and right distance threshold; the two vehicle blind spot coverage thresholds are respectively a first vehicle blind spot coverage threshold and a second vehicle blind spot coverage threshold less than the first vehicle blind spot coverage threshold, wherein: In step (B), when it is determined that the video data shows the second vehicle, the processing unit further determines whether the video data satisfies a turning intention condition, wherein the turning intention condition represents that the image or sound of the video data indicates an expected turning side related to the subsequent driving mode of the second vehicle, and the expected turning side is one of the left and right sides of the second vehicle; Step (D) further comprises: (d1) when the video data does not meet the turning intention condition, the processing unit determines whether the second blind spot warning condition is met according to the video data, and controls the output unit to output the second warning notification when the determination result is yes; (d2) after determining that the second vehicle is located obliquely behind the first vehicle and the video data meets the turning intention condition, the processing unit determines whether the current position of the first vehicle is biased toward the expected turning side of the second vehicle; (d3) if it is determined that the current position of the first vehicle is not biased toward the expected turning side of the second vehicle, the processing unit determines whether the second blind spot warning condition is met based on the video data, and controls the output unit to output the second warning notification when the determination result is yes; and (d4) If it is determined that the current position of the first vehicle is biased toward the expected turning side of the second vehicle, the processing unit determines whether a third blind spot warning condition is met based on the video data, and controls the output unit to output a third warning notification when the judgment result is yes.

10. The driving warning method according to claim 9, characterized in that: In step (B), the processing unit determines whether the video data meets the turning intention condition in the following manner: determining whether only the turn signal on one side of the second vehicle is flashing based on the image of the video data, or determining whether there is a turning keyword indicating a turning side in the sound of the video data; if the determination result is yes, the side of the left or right sides of the second vehicle on which the turn signal is flashing is used as the expected turning side indicated by the video data, or the turning side indicated by the turning keyword is used as the expected turning side indicated by the video data, and the video data is determined to meet the turning intention condition.

Images