Method for providing video-based vehicle accident prediction and notification and system implementing the same

Abstract

A method for predicting and providing notifications of vehicle-related accidents based on video and a system for implementing the same are provided. A method for predicting and providing notifications of vehicle-related accidents based on video according to an embodiment of the present invention may include: receiving a vehicle-related image captured by at least one camera mounted on a vehicle by a vehicle-related accident prediction and notification server; analyzing a plurality of image images included in the vehicle-related image to recognize one of a pre-stored driving situation; creating a virtual area that matches the recognized driving situation and overlaying it on the vehicle-related image; monitoring whether an object of interest corresponding to a pre-set detection condition is detected within the overlaid virtual area in the vehicle-related image; determining whether there is a risk factor for a vehicle-related accident based on image changes in the vehicle-related image when the object of interest is detected; and providing a driving guide notification to at least one of a guidance device mounted on the vehicle and a driver terminal based on the determination result.

Description

Technology Field

[0001] The present invention relates to a method for predicting and notifying vehicle-related accidents based on images and a system for implementing the same. Background Technology

[0002] Due to geographical characteristics, domestic transportation largely relies on road transport; consequently, the inconvenience caused by traffic congestion and the risk of traffic accidents between vehicles are consistently cited as persistent issues. While various driving guidance systems designed to prevent accidents are installed in vehicles and provided to drivers, technology capable of predicting accident occurrences by comprehensively considering both the external environment and the driver's internal situation is lacking.

[0003] In particular, drivers of passenger transportation vehicles such as city / intercity buses, village buses, and taxis spend long hours on the road and are exposed to a higher risk of accidents than general drivers of private vehicles, and the scale of damage in the event of an accident is significant. Nevertheless, systems to assist these drivers in safe driving by providing immediate danger alerts based on the condition of surrounding roads and vehicles have not yet been properly established.

[0004] Meanwhile, since the ratio of fault is determined according to the Road Traffic Act and other relevant laws in the event of an accident, which in turn affects the level of punishment and compensation, information regarding traffic regulations concerning liability for such accidents is a matter of significant interest among drivers. In the case of public transportation as well, disciplinary and employment issues regarding the driver may be linked depending on whether the vehicle is the at-fault or victim vehicle in the event of an accident, making this a matter that must be determined and recorded with importance.

[0005] However, conventional methods for determining fault have been limited to taking photos of the accident scene or directly reviewing dashcam footage after an accident occurs; therefore, there is a need to implement technology that predicts and warns drivers of the potential to cause an accident before it happens. Furthermore, due to the low utilization of video footage continuously recorded from the exterior and / or interior of the vehicle, there is a need to enhance overall traffic safety by providing various driver analysis and accident prevention functions. The problem to be solved

[0006] The present invention aims to solve the above-mentioned problems by providing a video-based method for predicting and notifying vehicle accidents and a system for implementing the same, which enables a driver to prepare for accident risks in advance by predicting and notifying the occurrence of various types of accidents based on images captured by a vehicle camera.

[0007] In addition, the present invention aims to provide a video-based method for predicting and notifying vehicle accidents and a system for implementing the same, which measures the speed and acceleration of a vehicle through changes in the captured image and utilizes this to predict the occurrence of accidents, thereby preventing collisions with objects such as other vehicles and inducing the driver to brake the vehicle.

[0008] In addition, the present invention aims to provide a video-based method for predicting and notifying vehicle-related accidents and a system for implementing the same, which enables driving habits of individual drivers to be reflected in a prediction model and improves the accuracy of accident prediction by continuously learning from data analyzed within a server and accident-related data collected from the outside.

[0009] The problem to be solved by the present invention is not limited thereto, and may also include objectives or effects that can be identified from the means of solving the problem or embodiments described below. means of solving the problem

[0010] A method for providing a driving guide notification based on an image of a vehicle accident according to an embodiment of the present invention may include: receiving a vehicle surrounding image captured from at least one camera mounted on a vehicle by a server providing a vehicle surrounding image; analyzing a plurality of image images included in the vehicle surrounding image to recognize one of a pre-stored driving situation; creating a virtual area that matches the recognized driving situation and overlaying it on the vehicle surrounding image; monitoring whether an object of interest corresponding to a pre-set detection condition is detected within the overlaid virtual area in the vehicle surrounding image; determining whether there is a risk factor for a driving accident based on an image change of the vehicle surrounding image when the object of interest is detected; and providing a driving guide notification to at least one of a guidance device mounted on the vehicle and a driver terminal based on the result of the determination.

[0011] Additionally, when the object of interest is detected, the step of determining whether there is a risk factor for a culpable accident based on image changes in the vehicle surrounding image may include: a step of analyzing at least one position and range change data among the images of the object of interest and external objects in the vehicle surrounding image; a step of calculating the relative speed between the object of interest and the vehicle and the acceleration of the vehicle based on the position and range change data; and a step of determining whether there is a risk factor for a culpable accident by comparing the relative speed and the acceleration with a pre-stored culpable accident prediction range.

[0012] In addition, the external object is characterized as being a stationary object including at least one of a road, lane, road marking, crosswalk, traffic light, and traffic sign, and the acceleration is calculated by a change in the range ratio of the external object image among a plurality of image images of the same time interval included in the vehicle surrounding image, and the relative speed can be calculated by a change in the range ratio of the object of interest image among a plurality of image images of the same time interval.

[0013] In addition, the predicted range of the accident causing the accident regarding the first driving situation is characterized by the relative speed having a negative (-) value, and if it is determined that there is a risk factor for the accident causing the accident regarding the first driving situation, a driving guide notification including a stop-waiting instruction for the vehicle can be generated.

[0014] In addition, the predicted range of the accident causing the accident regarding the second driving situation is characterized by the acceleration having a positive (+) value, and if it is determined that there is a risk factor for the accident causing the accident regarding the second driving situation, a driving guide notification including deceleration and braking instructions for the vehicle can be generated.

[0015] Additionally, the step of analyzing a plurality of image files included in the vehicle surrounding video to recognize one of the pre-stored driving situations may include: a step of detecting at least one situation recognition object included in the plurality of image files; a step of calculating a matching rate for each of the pre-stored driving situations by comparing the type, shape, color, location, size, angle, range, and change state of the detected at least one situation recognition object with standard data for each driving situation; and a step of recognizing the driving situation to which the matching rate belongs as the driving situation of the vehicle when any of the matching rates exceeds a recognition threshold value.

[0016] In addition, the above situation recognition object includes at least one of a road, lane, road marking, crosswalk, traffic light, traffic sign, vehicle door, surrounding vehicle, structure, and pedestrian, and the above pre-stored driving situation may include at least one of a road stopping situation, a U-turn situation, a side road driving situation, an intersection entry situation, a signal stop situation, and an off-road entry situation.

[0017] Additionally, if it is determined that there is a risk factor for the above-mentioned accident, the method further includes the step of generating the driving guide notification, which includes at least one of a warning signal, information on the accident, information on driver instructions, and a virtual guidance graphic, wherein the information on driver instructions may include instructions for stopping, shifting gears, braking, and steering of the vehicle.

[0018] Additionally, the method may further include the step of extracting a portion of video of a predetermined length from the vehicle surrounding video that includes a point in time determined to have the risk factor for the accident; the step of cumulatively collecting the extracted portion of video; the step of generating driving data of the driver by analyzing the collected plurality of portions of video; and the step of changing at least one condition data used to determine the presence or absence of the risk factor for the accident based on the driving data.

[0019] A video-based vehicle accident prediction and notification provision system according to another embodiment of the present invention comprises: a vehicle including at least one camera for capturing a vehicle surrounding video and a guidance device for outputting at least one of visual content and auditory content; a video-based vehicle accident prediction and notification provision server including a processor that receives and analyzes the captured vehicle surrounding video and determines a risk situation for an accident based on the analysis result; and a driver terminal that receives a driving guide notification according to the risk situation for an accident from the video-based vehicle accident prediction and notification provision server. The processor analyzes a plurality of video images included in the vehicle surrounding video to recognize one of the driving situations stored in advance, creates a virtual area matching the recognized driving situation and overlaps it with the vehicle surrounding video, monitors whether an object of interest corresponding to a preset detection condition is detected within the overlapped virtual area of ​​the vehicle surrounding video, and if the object of interest is detected, determines the presence or absence of the accident risk factor based on the image change of the vehicle surrounding video, and can generate the driving guide notification based on the determination result. Effects of the invention

[0020] According to the image-based vehicle accident prediction and notification provision method and the system implementing the same, based on an embodiment of the present invention, the occurrence of various types of accidents is predicted and a risk notification is provided based on images captured by a vehicle camera, thereby enabling the driver to prepare for accident risks in advance.

[0021] In addition, according to the image-based vehicle accident prediction and notification provision method and the system implementing the same according to one embodiment of the present invention, the speed and acceleration of a vehicle are measured through changes in the image of a captured video and utilized to predict the occurrence of an accident, thereby preventing collisions with objects such as other vehicles and inducing the driver to brake the vehicle.

[0022] In addition, according to the image-based vehicle accident prediction and notification provision method and the system implementing the same according to one embodiment of the present invention, the driving habits of each driver are reflected in the prediction model and the accuracy of accident prediction can be improved by continuously learning from data analyzed within the server and accident-related data collected from the outside. Brief explanation of the drawing

[0023] A brief description of each drawing is provided to help to better understand the drawings cited in the detailed description of the invention. FIG. 1 is a conceptual diagram schematically illustrating a video-based vehicle accident prediction and notification provision system according to one embodiment of the present invention. FIG. 2 is a conceptual diagram schematically illustrating an embodiment in which images of the vehicle's surroundings are provided from a plurality of cameras mounted on the vehicle. Figure 3 is a block diagram schematically showing the detailed configuration of the image-based vehicle accident prediction and notification provision server illustrated in Figure 1. FIG. 4 is a flowchart illustrating a method for predicting and notifying vehicle accidents based on images according to an embodiment of the present invention. FIG. 5 is a flowchart showing a specific implementation example of step S450 illustrated in FIG. 4. FIG. 6 is a flowchart showing a specific implementation example of step S420 illustrated in FIG. 4. FIG. 7 is a flowchart illustrating a method for improving a driver-specific accident prediction model according to an additional embodiment of the present invention. Specific details for implementing the invention

[0024] The present invention is capable of various modifications and may have various embodiments, and specific embodiments are illustrated and described in the drawings. However, this is not intended to limit the invention to specific embodiments, and it should be understood that the invention includes all modifications, equivalents, and substitutions that fall within the spirit and scope of the invention.

[0025] Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but said components are not limited by said terms. Such terms are used solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the second component may be named the first component, and similarly, the first component may be named the second component. The term "and / or" includes a combination of a plurality of related described items or any of a plurality of related described items.

[0026] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. On the other hand, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between.

[0027] The terms used in this application are used merely to describe specific embodiments and are not intended to limit the invention. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, terms such as "comprising" or "having" are intended to specify the presence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0028] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the present invention pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0029] Hereinafter, embodiments will be described in detail with reference to the attached drawings, provided that identical or corresponding components are given the same reference number regardless of the drawing symbols, and redundant descriptions thereof will be omitted.

[0031] FIG. 1 is a conceptual diagram schematically illustrating a video-based vehicle accident prediction and notification provision system according to one embodiment of the present invention.

[0032] As illustrated in FIG. 1, an image-based vehicle accident prediction and notification provision system according to one embodiment of the present invention may include an image-based vehicle accident prediction and notification provision server (100), a vehicle (200), and a driver terminal (300), and the vehicle (200) may include at least one camera (210) and a guidance device (220). In addition, an image-based vehicle accident prediction and notification provision system according to one embodiment of the present invention may further include an external server (400).

[0033] A video-based vehicle accident prediction and notification provision server (100) according to one embodiment of the present invention may be implemented as a computer device or a plurality of computer devices that provide commands, code files, content, services, etc. The video-based vehicle accident prediction and notification provision server (100) may be implemented in the form of software for electronic devices, a webpage, or an application (APP) as needed. In this case, data exchange between the video-based vehicle accident prediction and notification provision server (100) and the driver terminal (300) and / or the guidance device (220) may be performed by the driver accessing the webpage or application through the driver terminal (300) or by executing software for electronic devices installed on the guidance device (220) mounted in the vehicle.

[0034] The video-based vehicle accident prediction and notification provision server (100) may refer to a server that predicts the possibility of a vehicle accident occurring based on real-time footage of the vehicle's surroundings and provides a danger notification to the driver. Additionally, the video-based vehicle accident prediction and notification provision server (100) may refer to a server that analyzes stored footage of the vehicle's surroundings to diagnose the driver's driving habits and builds an accident prediction model for each driver based on this.

[0035] The vehicle (200) may be equipped with at least one camera (210) that captures images of the vehicle's surroundings. The at least one camera (210) captures images of the vehicle's surroundings and can provide them to a video-based vehicle accident prediction and notification server (100).

[0036] For example, as illustrated in FIG. 2, a video-based vehicle accident prediction and notification providing server (100) and a plurality of cameras (210_1, 210_2, …, 210_n) can communicate. FIG. 2 is a conceptual diagram schematically illustrating an embodiment in which vehicle surrounding images are provided from a plurality of cameras mounted on a vehicle (200).

[0037] Here, multiple cameras (210_1, 210_2, …, 210_n) are each mounted at different locations on the vehicle (200) so that images of the vehicle's surroundings in different directions can be provided to the image-based vehicle accident prediction and notification provision server (100). As an example, the first camera (210_1) is mounted on the front of the vehicle (200) and can capture images facing the front of the vehicle (200) in the driving direction. The second camera (210_2) is mounted on the left side of the vehicle (200) and can capture images facing the front left direction of the vehicle (200).

[0038] The vehicle (200) may be further equipped with a guidance device (220) that outputs at least one of visual content and auditory content. For example, the guidance device (200) may include a voice guidance device of the vehicle and, accordingly, output auditory content that warns the driver of the risk of a collision. As another example, the guidance device (200) may include a navigation device of the vehicle and, accordingly, output visual content based on information regarding a collision and driver instructions to the driver, or output virtual guidance graphics.

[0039] However, in the image-based vehicle accident prediction and notification provision system according to the embodiment of the present invention, the guidance device (220) is not necessarily required to be included, and it is also possible for the driver terminal (300) to perform the function of the aforementioned guidance device (200).

[0040] The driver terminal (300) may refer to a terminal owned by a driver operating a vehicle (200). The driver terminal (300) may be connected to a video-based vehicle accident prediction and notification provision server (100) to enable communication.

[0041] The driver can provide necessary information, data, and signals to the image-based vehicle accident prediction and notification provision server (100) through the driver terminal (300). For example, the driver can input personal information and vehicle information through the driver terminal (300) so that it can be utilized to predict an accident according to an embodiment of the present invention. In addition, the driver can receive necessary information, data, and signals from the image-based vehicle accident prediction and notification provision server (100) through the driver terminal (300). For example, the driver can receive driving guide notifications according to the accident risk situation through the driver terminal (300) to prepare for the occurrence of an accident.

[0042] For reference, the driver terminal (300) may be implemented as a terminal capable of connecting to a remote server or terminal via a network. For example, the driver terminal (300) may be a wireless communication device that ensures portability and mobility, and may include all kinds of handheld-based wireless communication devices such as navigation, PCS (Personal Communication System), GSM (Global System for Mobile communication), PDC (Personal Digital Cellular), PHS (Personal Handphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), Wibro (Wireless Broadband Internet) terminal, smartphone, smartpad, tablet PC, etc.

[0043] The external server (400) may refer to a server that possesses video footage that can be utilized to improve the accuracy of the accident prediction of the video-based vehicle accident prediction and notification provision server (100) according to an embodiment of the present invention. For example, the external server (400) may correspond to a server of a platform where various black box videos of traffic accidents are shared, and the video-based vehicle accident prediction and notification provision server (100) may receive and learn from the black box videos provided by the external server (400). Accordingly, the accuracy of the prediction module of the video-based vehicle accident prediction and notification provision server (100) can be continuously improved.

[0045] FIG. 3 is a block diagram schematically showing the detailed configuration of the image-based vehicle accident prediction and notification provision server (100) illustrated in FIG. 1.

[0046] As illustrated in FIG. 3, the image-based vehicle accident prediction and notification providing server (100) may include a communication module (110), a processor (130), and a database (150).

[0047] For reference, the components (110, 130, 150) of the image-based vehicle accident prediction and notification provision server (100) illustrated in FIG. 3 are merely exemplary components for explaining the image-based vehicle accident prediction and notification provision method according to an embodiment of the present invention. That is, it is evident that the image-based vehicle accident prediction and notification provision server (100) according to an embodiment of the present invention may additionally include other components other than those illustrated.

[0048] The communication module (110) can establish a communication channel with the video-based vehicle accident prediction and notification provision server (100) and external devices. The communication module (110) can communicate with external devices by accessing a network via wireless or wired communication. Specifically, the communication module (110) can transmit and receive necessary signals and data with at least one camera (210) and a guidance device (220). Additionally, the communication module (110) can transmit and receive necessary signals and data with a driver terminal (300). Additionally, the communication module (110) can transmit and receive necessary signals and data with an external server (400).

[0049] For example, the communication module (110) can receive images of the vehicle surroundings captured from at least one camera (210). As another example, the communication module (110) can transmit a driving guide notification to at least one of the guidance device (220) and the driver terminal (300). As yet another example, the communication module (110) can receive images necessary for the learning of the accident prediction module from an external server (400).

[0050] The processor (130) may include at least one of a central processing unit (CPU), an application processor (AP), or a communication processor (CP). The processor (130) is electrically connected to the communication module (110) and the database (150) and can execute operations or data processing regarding the control and / or communication of other components according to instructions, programs, or software stored in memory during operation. Accordingly, the execution of the instructions, application programs, or software can be understood as the operation of the processor (130).

[0051] For example, the processor (130) receives and analyzes captured vehicle surrounding video and can determine a risk situation for causing an accident based on the analysis results. Specifically, the processor (130) can recognize one of the pre-stored driving situations by analyzing multiple video images included in the vehicle surrounding video. Additionally, the processor (130) can create a virtual area and overlap it with the vehicle surrounding video. Additionally, the processor (130) can monitor whether an object of interest matching a pre-set detection condition is detected within the virtual area. Additionally, the processor (130) can determine whether there is a risk factor for causing an accident based on changes in the image of the vehicle surrounding video. Additionally, the processor (130) can generate a driving guide notification.

[0052] The database (150) may be implemented in memory within the video-based vehicle accident prediction and notification provision server (100) or on a separate storage medium. The database (150) may store all contents and history of data transmitted and received with the driver terminal (300). Additionally, the database (150) may store vehicle surrounding images provided by at least one camera (210). Furthermore, the database (150) may store result data analyzed through the processor (130) and driving data for each driver.

[0053] Data stored in the database (150) can be updated regularly according to a predetermined period, and can be updated frequently when new data is entered through the driver terminal (300) or external server (400).

[0054] Meanwhile, according to various embodiments, since the data stored in the database (150) is sensitive customer information, it may be distributed and stored in a blockchain network to enhance security regarding the use of said information. When the database (150) is distributed and stored in a blockchain network, the history of transmission, modification, deletion, addition, etc. of the information included in the database (150) can be managed more securely in said blockchain network.

[0056] Hereinafter, with reference to FIGS. 4 to 7, the operation steps of an image-based vehicle accident prediction and notification provision server (100) according to an embodiment of the present invention will be described.

[0057] FIG. 4 is a flowchart illustrating a method for predicting and notifying vehicle accidents based on images according to an embodiment of the present invention.

[0058] Referring to FIG. 4, an image-based vehicle accident prediction and notification provision server (100) according to one embodiment of the present invention can receive a vehicle surrounding image from at least one camera (210) mounted on a vehicle (200) (S410).

[0059] The vehicle surrounding images may include images taken from various directions, such as the front front image, front left image, front right image, and rear image of the vehicle (200). The vehicle surrounding images may include objects such as structures, other vehicles, pedestrians, traffic lights, and traffic signs located in close proximity to the vehicle (200), as well as objects such as roads, lanes, and road markings that allow the driving position of the vehicle (200) to be identified. At this time, the vehicle surrounding images may include specific color information of objects, including the respective colors corresponding to the flashing of traffic lights and the colors corresponding to the flashing of taillights of other vehicles. Additionally, the vehicle surrounding images may include vehicle door objects that allow the opening or closing of the vehicle (200)'s door to be identified.

[0060] A video-based vehicle accident prediction and notification provision server (100) can recognize one of the previously stored driving situations by analyzing multiple video images included in the vehicle surrounding video (S420).

[0061] That is, the vehicle surrounding image may include multiple image images corresponding to each stationary state included in the image, and the processor (130) may analyze each of the multiple image images.

[0062] The video-based vehicle accident prediction and notification provision server (100) may store data regarding at least one driving situation. Here, a driving situation may refer to a situation occurring during operation based on traffic information such as the current location of the vehicle (200) and signals. By first identifying the driving situation of the vehicle (200), the possibility of an accident occurring in that driving situation can be specifically predicted. For example, the driving situation stored in advance in the video-based vehicle accident prediction and notification provision server (100) may include at least one of a road stopping situation, a U-turn situation, a side road driving situation, an intersection entry situation, a signal stop situation, and an off-road entry / exit situation.

[0063] A video-based vehicle accident prediction and notification provision server (100) can recognize that an object identified in a video image corresponds to one of the previously stored driving situations, and can execute an accident prediction model corresponding to the recognized driving situation. At this time, a specific example of recognizing one of the previously stored driving situations will be described in FIG. 6, which will be described later.

[0064] Referring further to FIG. 4, the image-based vehicle accident prediction and notification provision server (100) can create a virtual area that matches the recognized driving situation and overlap it with the vehicle surrounding image (S430).

[0065] Here, the virtual area is a designated area for detecting objects at risk of collision or lanes that can be encroached upon in order to predict a high-risk accident based on the current driving situation of the vehicle (200). In the image-based vehicle accident prediction and notification provision server (100), rules for generating virtual areas for each driving situation may be designated and stored in advance, and as the driving situation of the vehicle (200) is recognized, a virtual area matching the driving situation may be applied to the vehicle surrounding image and overlapped.

[0066] For example, the driving situation of the vehicle (200) may be recognized as a 'road stopping situation,' and the road stopping situation may refer to a situation where the vehicle (200) is temporarily stopped at the right edge of the road for purposes such as disembarking or unloading goods. The virtual area matching the road stopping situation may correspond to a certain area located on the rear left side of the vehicle (200), and the virtual area corresponding thereto may be overlaid on the vehicle surrounding video captured in real time. The driver may check the virtual area overlaid on the vehicle surrounding video through the guidance device (220) and / or driver terminal (300) as needed.

[0067] Next, the video-based vehicle accident prediction and notification provision server (100) can monitor whether an object of interest matching a preset detection condition is detected within an overlapping virtual area (S440).

[0068] Here, an object of interest may refer to an object that poses a threat when entering the virtual area, in predicting culpable accidents with a high risk of occurrence depending on each driving situation. For example, objects of interest may include other vehicles and stop lines.

[0069] For example, if the driving situation of a vehicle (200) is recognized as a 'road stop situation' and a corresponding virtual area is created and overlapped, at least some of the other vehicles entering the virtual area may be detected as objects of interest. In this case, a detection condition for detecting objects of interest may be set and stored in advance as a 'silhouette of a vehicle shape'. Then, the image-based vehicle accident prediction and notification provision server (100) can detect only other vehicles as objects of interest among the various objects included in the virtual area through an object analysis and extraction method. Meanwhile, various methods by a person of ordinary skill in the art may be used for the object analysis and extraction method, and may include, for example, an extraction method based on color threshold analysis.

[0070] Next, the video-based vehicle accident prediction and notification provision server (100) can determine whether there is a risk factor for an accident based on changes in the image of the vehicle surrounding video when an object of interest is detected (S450).

[0071] That is, since the mere existence of a detected object of interest does not cause a risk of a culpable accident, there may be conditions where the risk of a culpable accident is high depending on the current driving situation and the type of object of interest. The image-based vehicle culpable accident prediction and notification provision server (100) may set and store conditions in advance to determine whether there is a risk factor for a culpable accident, and may determine that there is a risk factor for a culpable accident if the detected object of interest and / or external object causes an image change that corresponds to the above conditions.

[0072] For example, if the driving situation of the vehicle (200) is recognized as a 'road stop situation' and an 'other vehicle' that has entered the virtual area is detected as an object of interest, the image-based vehicle accident prediction and notification provision server (100) may determine that there is a risk factor for an accident if the speed at which the object of interest approaches the vehicle (200) corresponds to '20 km / h' or higher. That is, if there is another vehicle rapidly approaching from the left rear lane of the vehicle (200) while the vehicle (200) is about to start after stopping on the road, there is a risk of collision with the other vehicle when the vehicle (200) starts, so this can be determined as a risk factor for an accident.

[0073] Then, the video-based vehicle accident prediction and notification provision server (100) can provide a driving guide notification to at least one of the guidance device (220) and driver terminal (300) installed on the vehicle (200) based on the judgment result.

[0074] Here, a driving guide notification may refer to a notification containing information to be provided to the driver in the event of a risk factor for a causative accident. For example, a driving guide notification may include at least one of a warning signal, information on a causative accident, information on driver instructions, and virtual guidance graphics. The information on driver instructions may include instructions for stopping, shifting gears, braking, and steering. The virtual guidance graphics, as auxiliary graphics necessary to prevent the risk of a causative accident for the driver, may include virtual area graphics, virtual centerline graphics, and safety braking distance guidance graphics.

[0075] For example, if the driving situation of the vehicle (200) is recognized as a 'stopping on the road' and it is determined that there is a risk of an accident caused by a 'other vehicle' detected as an object of interest, the driving guide notification may include a 'stopping to start' warning signal, information that 'an accident caused by driver responsibility will occur if there is a collision with the vehicle behind,' and driver instruction information to 'stop and wait.' By receiving such driving guide notifications, the driver can start the vehicle (200) after stopping and waiting for a certain period of time while paying attention to other vehicles.

[0077] FIG. 5 is a flowchart illustrating a specific implementation example of step S450 shown in FIG. 4. That is, it specifically illustrates a method for determining the presence or absence of a risk factor for a culpable accident based on changes in the image of the vehicle's surroundings when an object of interest is detected.

[0078] Referring to FIG. 5, when an object of interest is detected within an overlapping virtual area (S440), the image-based vehicle accident prediction and notification provision server (100) can analyze at least one location and range change data among the images of the object of interest and external objects in the vehicle surrounding images (S510). Here, the external object may be characterized as being a stationary object including at least one of a road, lane, road marking, crosswalk, traffic light, and traffic sign.

[0079] Then, the video-based vehicle accident prediction and notification provision server (100) can calculate the relative speed between the object of interest and the vehicle and the acceleration of the vehicle based on location and range change data (S520).

[0080] At this time, acceleration can be calculated by the change in the range ratio of an external object image among multiple image images of equal time intervals included in the vehicle surrounding image. For example, multiple image images may commonly include a 'traffic light' located in front among the external objects, and the image-based vehicle accident prediction and notification provision server (100) can calculate the acceleration of the vehicle (200) by calculating the degree of change in the range ratio of the 'traffic light' image extracted from each image image to the entire image image. As such, since the external object is an object in a stationary state, the acceleration due to the driving of the vehicle (200) itself can be calculated according to the change in the range ratio of the external object image.

[0081] Additionally, relative speed can be calculated by the change in the range ratio of the object of interest image among multiple video images of the same time interval. For example, multiple video images may commonly include an image of an 'other vehicle' which is the object of interest, and the video-based vehicle accident prediction and notification provision server (100) can calculate the relative speed between the vehicle (200) and the object of interest by calculating the degree of change in the ratio of the range occupied by the 'other vehicle' image extracted from each video image within the entire video image. As such, since the object of interest, such as the 'other vehicle', and the vehicle (200) are in a stationary or moving state, the relative speed of approaching or moving away from the vehicle (200) in relation to the object of interest can be calculated according to the change in the range ratio of the object of interest image.

[0082] Next, the video-based vehicle accident prediction and notification provision server (100) can determine whether there is a risk factor for an accident by comparing the relative speed and acceleration with a pre-stored accident prediction range (S530).

[0083] Here, the predicted range of the accident may refer to a predicted range that considers the movement of the vehicle (200) and the object of interest so as to determine the risk of an accident occurring in an accident that may occur due to the object of interest detected in the current operating situation.

[0084] That is, for example, if the relative speed between the vehicle (200) and the object of interest has a negative (-) value, it means that the vehicle (200) and the object of interest are moving closer to each other. Therefore, when the driving situation is recognized as a 'U-turn situation,' it may be determined that there is a risk of an accident caused by the approach of an 'other vehicle' turning right from the opposite left lane. In this case, the vehicle (200) needs to wait for the 'other vehicle' turning right to proceed, and then drive after it disappears within the virtual area.

[0085] The range of accident prediction regarding the first driving situation may be characterized by the relative speed having a negative (-) value, and as an example, the first driving situation may correspond to a 'U-turn situation'. In this case, if the image-based vehicle accident prediction and notification provision server (100) determines that there is a risk factor for an accident regarding the first driving situation, it may generate a driving guide notification including a vehicle stop waiting instruction.

[0086] Alternatively, for example, if the acceleration of the vehicle (200) has a positive (+) value, it means that the vehicle (200) is accelerating without braking, so it may be determined that there is a risk factor for causing an accident when the driving situation is perceived as a 'signal stop situation'. In this case, the vehicle (200) needs to brake and decelerate from the point where a safe braking distance to the stop line is secured.

[0087] The range of accident prediction regarding the second driving situation may be characterized by acceleration having a positive (+) value, and as an example, the second driving situation may correspond to a 'signal stop situation'. In this case, if the image-based vehicle accident prediction and notification provision server (100) determines that there is a risk factor for an accident regarding the second driving situation, it may generate a driving guide notification including instructions for deceleration and braking of the vehicle.

[0088] Meanwhile, as mentioned above, relative speed and acceleration may each be included in separate accident prediction ranges, but for the same driving situation, relative speed and acceleration may also be simultaneously included in the accident prediction range and utilized.

[0090] FIG. 6 is a flowchart illustrating a specific implementation example of step S420 shown in FIG. 4. That is, it specifically illustrates a method for recognizing a driving situation by receiving a vehicle surrounding image.

[0091] Referring to FIG. 6, when a vehicle surrounding image is received from at least one camera (210) (S410), the image-based vehicle accident prediction and notification provision server (100) can detect at least one situation recognition object included in a plurality of image images (S610).

[0092] Here, a context-aware object may refer to an object that provides hints to recognize the current driving situation of a vehicle from multiple image frames included in the vehicle surrounding video. For example, a context-aware object may include at least one of a road, lane, road marking, crosswalk, traffic light, traffic sign, vehicle door, surrounding vehicle, structure, and pedestrian.

[0093] Then, the video-based vehicle accident prediction and notification provision server (100) can calculate the respective match rate for each driving situation stored in advance by comparing the type, shape, color, location, size, angle, range, and change state of at least one detected situation recognition object with standard data for each driving situation (S620).

[0094] For example, the video-based vehicle accident prediction and notification provision server (100) may have standard data of an 'intersection entry situation' stored in advance during driving. As an example, the standard data of an 'intersection entry situation' may include situation recognition objects such as 'lane', 'crosswalk', and 'traffic light', and specifically, among the situation recognition objects, the 'lane' is detected as being widely broken, a 'crosswalk' is detected at the point where the lane is broken, and a 'traffic light' containing four or more indicator lights is detected, and this information may be included that it is recognized as an 'intersection entry situation'.

[0095] In this case, the shape, color, location, and change state of each of the situation recognition objects of 'lane', 'crosswalk', and 'traffic light' detected in the vehicle surrounding image provided by at least one camera (210) can be compared with the information of the aforementioned standard data to calculate the matching rate.

[0096] As another example, the video-based vehicle accident prediction and notification provision server (100) may have standard data of a ‘side road driving situation’ stored in advance during driving, and the standard data of the ‘side road driving situation’ may include a situation recognition object of a ‘road’, and specifically, among the situation recognition objects, if there is no center lane on the ‘road’ and the road width falls within the range of 4m to 7m, it may include information that it is recognized as a side road driving situation.

[0097] In this case, the shape, color, location, change state, etc. of each 'road' situation recognition object detected in the vehicle surrounding image provided by at least one camera (210) can be compared with the information of the aforementioned standard data to calculate the matching rate.

[0098] Next, the video-based vehicle accident prediction and notification provision server (100) can recognize the driving situation to which one of the matching rates belongs as the driving situation of the vehicle (200) when any one of the matching rates exceeds a recognition threshold value (S630).

[0099] For example, the recognition threshold value can be pre-set to 80%, and the image-based vehicle accident prediction and notification provision server (100) can recognize that the vehicle (200) is currently in the corresponding driving situation when the match rate between the characteristics of the situation recognition object detected in the vehicle surrounding image and the standard data for each driving situation exceeds 80%.

[0101] FIG. 7 is a flowchart illustrating a method for improving a driver-specific accident prediction model according to an additional embodiment of the present invention.

[0102] Referring to FIG. 7, the image-based vehicle accident prediction and notification provision server (100) can extract an image portion of a predetermined length that includes a point in time determined to have an accident risk factor among the images around the vehicle (S710). Then, the extracted image portion can be collected cumulatively (S720).

[0103] Next, the video-based vehicle accident prediction and notification provision server (100) can generate driver driving data by analyzing multiple collected video portions (S730). Here, the driving data may include data regarding whether the driver drove the vehicle (200) in accordance with the driving guide notification when a notification was provided because it was determined that there was a risk factor for an accident. Additionally, the driving data may include data regarding the reaction time taken by the driver to respond to the danger after a notification regarding the risk factor for an accident occurred.

[0104] The video-based vehicle accident prediction and notification provision server (100) can change at least one condition data used to determine whether there is a risk factor for an accident based on driving data (S740). Accordingly, differences based on driving habits of each driver can be individually applied when determining whether there is a risk factor for an accident, and the accuracy of accident prediction and preparedness for each driver can be further improved.

[0106] Hereinafter, specific application embodiments of the present invention will be described.

[0107] First application example

[0108] This describes an embodiment in which a “road stop situation” is recognized during operation, and another vehicle approaches from the rear, providing the driver with a notification to stop starting.

[0109] According to the present embodiment, standard data regarding a “road stopping situation” may include a “lane” and a “vehicle door” as situation recognition objects. For example, the location and range of the “lane” may be set to be concentrated on the left side of the vehicle (200) to indicate that the vehicle (200) is currently stopped at the right edge of the road, or the “vehicle door” image may be set to be displayed on a camera (210) that has captured the side or rear of the vehicle (200) to indicate that a person is getting on or off the vehicle or unloading goods on the vehicle (200). The image-based vehicle accident prediction and notification provision server (100) can recognize the vehicle’s driving situation as a “road stopping situation” by calculating the match rate between this standard data and the situation recognition objects detected in the actual vehicle surrounding image and checking whether it exceeds a recognition threshold value.

[0110] Then, the 'area where other vehicles can approach from the rear' of the vehicle (200) can be designated and created as a virtual area and can be overlapped with the vehicle surrounding image. Then, the 'other vehicle' can act as an object of interest and the detection of a shape similar to the vehicle can be set as a detection condition, and accordingly, it can be monitored whether an object of interest corresponding to the 'other vehicle' is detected within the virtual area.

[0111] Additionally, regarding the driving situation of a “road stopping situation,” the accident prediction range for determining whether there is a risk factor for an accident can be set such that the relative speed value between the vehicle (200) and the object of interest has a negative (-) value greater than a predetermined value. Accordingly, when the vehicle (200) and another vehicle approaching from the rear get closer to each other at a speed greater than a predetermined speed during the road stopping situation of the vehicle (200), it can be determined that there is a risk factor for an accident involving the vehicle (200).

[0112] If it is determined that there is a risk factor for a cause-and-run accident, the video-based vehicle cause-and-run accident prediction and notification provision server (100) can generate and provide a driving guide notification to the driver that includes a notification to stop the driver from starting. Here, the notification to stop the driver may include a “stop and wait instruction” as a driver instruction.

[0114] 2nd Application Example

[0115] An embodiment is described in which a “U-turn situation” is recognized during driving, and a vehicle turning right or proceeding straight in the opposite lane ahead is provided to the driver with a U-turn restriction notification.

[0116] According to the present embodiment, standard data regarding a “U-turn situation” may include a “lane,” “road marking,” “traffic light,” and “traffic sign” as situation recognition objects. For example, the shape of the “lane” and the “road marking” may be set to be displayed in accordance with the U-turn lane to indicate that the current vehicle (200) is a vehicle waiting to make a U-turn, or the “traffic light” and “traffic sign” may be set so that a sign indicating a left turn signal for a U-turn is not recognized, indicating that the current vehicle (200) intends to make a U-turn on a road without a U-turn signal. The image-based vehicle accident prediction and notification provision server (100) can recognize the vehicle's driving situation as a “U-turn situation” by calculating the match rate between the standard data and the situation recognition objects detected in the actual vehicle surrounding image and checking whether it exceeds a recognition threshold value.

[0117] Then, a ‘part of the front left right-turn lane’ and / or a ‘part of the front opposite straight lane’ of the vehicle (200) can be designated and created as a virtual area and can be overlapped with the vehicle surrounding image. Additionally, a ‘other vehicle’ can act as an object of interest, and the detection condition can be the detection of a shape similar to the vehicle, and accordingly, it can be monitored whether an object of interest corresponding to the ‘other vehicle’ is detected within the virtual area.

[0118] Additionally, regarding the driving situation of a “U-turn situation,” the accident prediction range for determining whether there is a risk factor for an accident can be set such that the relative speed value between the vehicle (200) and the object of interest has a negative (-) value greater than a predetermined value. Accordingly, when another vehicle intending to turn right or go straight from the opposite lane approaches the vehicle (200) at a speed greater than a predetermined speed while the vehicle (200) is waiting to make a U-turn, it can be determined that there is a risk factor for an accident involving the vehicle (200).

[0119] If it is determined that there is a risk factor for a cause-and-run accident, the video-based vehicle cause-and-run accident prediction and notification provision server (100) can generate and provide a driving guide notification to the driver, including a notification to stop the driver from making a U-turn. Here, the U-turn stop-and-run notification may include a “stop and wait instruction” as a driver instruction.

[0121] Third Application Example

[0122] An embodiment is described in which a “side road driving situation” is recognized during driving, and a virtual center line violation alert is provided to the driver due to a vehicle approaching from the opposite direction ahead.

[0123] According to the present embodiment, standard data regarding “side road driving situation” may include a “road” as a situation recognition object, and, for example, may be set so that the center lane is not recognized in the shape of the “road” and the road width falls within a predetermined range, thereby indicating that the current vehicle (200) is driving on a side road. The image-based vehicle accident prediction and notification provision server (100) can recognize the vehicle’s driving situation as a “side road driving situation” by calculating the match rate between this standard data and the situation recognition object detected in the actual vehicle surrounding image and checking whether it exceeds a recognition threshold value. At this time, a center point within the side road may be calculated to generate a virtual center line.

[0124] Then, the 'area where other vehicles can approach from the opposite direction of the front of the vehicle (200)' can be designated and created as a virtual area and can be overlapped with the vehicle surrounding image. In addition, the 'other vehicle' can act as an object of interest, and the detection condition can be that a shape similar to the vehicle is detected as an image of a predetermined size or larger, and accordingly, it can be monitored whether an object of interest corresponding to the 'other vehicle' is detected within the virtual area.

[0125] Additionally, regarding the driving situation of the “side road driving situation,” the accident prediction range for determining whether there is a risk factor for an accident can be set such that the relative speed value between the vehicle (200) and the object of interest has a negative (-) value, and the acceleration value of the vehicle (200) has a positive (+) value. Accordingly, when the vehicle (200) is driving on a side road, if there is another vehicle approaching from the opposite direction and the vehicle (200) is accelerating without braking, it can be determined that there is a risk factor for an accident involving the vehicle (200).

[0126] If it is determined that there is a risk factor for a culpable accident, the video-based vehicle culpable accident prediction and notification provision server (100) can determine whether the vehicle (200) is crossing the virtual center line by comparing the position of the vehicle (200) with a pre-generated virtual center line. Then, it can generate and provide a driving guide notification to the driver that includes a virtual center line crossing notification. Here, the virtual center line crossing notification may include "right steering instruction" and "acceleration stop instruction" as driver instructions.

[0128] 4th Application Example

[0129] This describes an embodiment in which a “situation of entering an intersection” is recognized during driving, and a situation in which a vehicle entering the intersection ahead is congested, so a stop notification within the stop line is provided to the driver.

[0130] According to the present embodiment, standard data regarding “intersection entry situation” may include “lanes,” “crosswalks,” and “traffic lights” as situation recognition objects, for example, the location and range of the “lanes” may be set to be cut off in front, the location of the “crosswalks” may be set to where the “lanes” are cut off, and the shape of the “traffic lights” may be set to include four or more, thereby indicating that there is an intersection in front of the vehicle (200). The image-based vehicle accident prediction and notification provision server (100) can recognize the vehicle’s driving situation as an “intersection entry situation” by calculating the match rate between these standard data and the situation recognition objects detected in the actual vehicle surrounding image and checking whether it exceeds a recognition threshold value.

[0131] Then, the 'front vehicle congestion possible area' of the vehicle (200) can be designated and created as a virtual area and can be overlapped with the vehicle surrounding image. And, the 'red taillight' can act as an object of interest, and the detection condition can be set as the detection of a pair of red taillight objects, and accordingly, it can be monitored whether an object of interest corresponding to the 'red taillight' is detected within the virtual area.

[0132] Additionally, regarding the driving situation of “entering an intersection,” the accident prediction range for determining whether there is a risk factor for an accident can be set such that the relative speed value between the vehicle (200) and the object of interest has a negative (-) value. Accordingly, when the vehicle (200) is scheduled to enter an intersection, if a vehicle located in front of the vehicle (200) is stopped due to congestion but the vehicle (200) does not stop driving and approaches it, it can be determined that there is a risk factor for an accident involving the vehicle (200).

[0133] If it is determined that there is a risk factor for a cause-and-run accident, the image-based vehicle cause-and-run accident prediction and notification provision server (100) can recognize a stop line ahead and generate and provide a driving guide notification to the driver that includes a stop notification within the stop line. Here, the stop notification within the stop line may include a “stop waiting instruction” as a driver instruction.

[0135] Fifth Application Example

[0136] An embodiment is described in which a “signal stop situation” is recognized during operation, and the vehicle (200) enters within the safe braking distance and a deceleration notification is provided to the driver.

[0137] According to the present embodiment, standard data regarding a “signal stop situation” may include a “stop line” and a “traffic light” as situation recognition objects, and for example, the “stop line” may be displayed at a certain position in front and the color of the “traffic light” may be set to be red to indicate a situation where the vehicle (200) needs to stop at a signal. The image-based vehicle accident prediction and notification provision server (100) can recognize the vehicle's driving situation as a “signal stop situation” by calculating the match rate between this standard data and the situation recognition objects detected in the actual vehicle surrounding image and checking whether it exceeds a recognition threshold value. At this time, a safe braking distance according to the speed of the vehicle (200) can be calculated so that a starting point for braking is displayed.

[0138] Then, the 'safety braking distance securing area' of the vehicle (200) can be designated and created as a virtual area and can be overlapped with the vehicle surrounding image. And, the 'stop line' can act as an object of interest, and the detection condition can be that a white horizontal line is detected in front, and accordingly, it can be monitored whether an object of interest corresponding to the 'stop line' is detected within the virtual area.

[0139] Additionally, regarding the driving situation of a “signal stop situation,” the accident prediction range for determining whether there is a risk factor for an accident can be set such that the relative speed value between the vehicle (200) and the object of interest has a negative (-) value, and the acceleration of the vehicle (200) has a value of “” or a positive (+). Accordingly, if the vehicle (200) does not brake (decelerate) even though it is approaching the stop line within the safe braking distance, it can be determined that there is a risk factor for an accident involving the vehicle (200).

[0140] If it is determined that there is a risk factor for a cause-and-run accident, the video-based vehicle cause-and-run accident prediction and notification provision server (100) can generate and provide a driving guide notification to the driver that includes a deceleration notification within the safe braking distance. Here, the deceleration notification may include a “braking instruction” as a driver instruction.

[0142] 6th Application Example

[0143] An embodiment is described in which a “road exit situation” is recognized during operation, and a vehicle is approaching the lane scheduled for entry, so a stop notification is provided to the driver before entering the road.

[0144] According to the present embodiment, standard data regarding “off-road entry / exit situation” may include “road surface” and “lane” as situation recognition objects, and, for example, the position and angle of the “road surface” and “lane” may be set to be located close to the front of the vehicle (200) while forming a predetermined angle skewed to the left, thereby indicating a situation in which the vehicle (200) is currently scheduled to enter the road from outside the road surface. The image-based vehicle accident prediction and notification provision server (100) can recognize the vehicle’s driving situation as an “off-road entry / exit situation” by calculating the match rate between this standard data and the situation recognition objects detected in the actual vehicle surrounding image and checking whether it exceeds a recognition threshold value.

[0145] Then, the 'area where a straight-moving vehicle can approach the far right lane' of the vehicle (200) can be designated and created as a virtual area and can be overlapped with the vehicle surrounding image. In addition, the detection condition can be set so that 'another vehicle' acts as an object of interest and detects a shape similar to the vehicle, and accordingly, it can be monitored whether an object of interest corresponding to 'another vehicle' is detected within the virtual area.

[0146] Additionally, regarding the operation situation of “ex-road entry / exit situation,” the accident prediction range for determining whether there is a risk factor for an accident can be set such that the relative speed value between the vehicle (200) and the object of interest has a negative (-) value. Accordingly, when the vehicle (200) attempts to enter the road, if the vehicle (200) and a moving straight vehicle approach each other at a speed greater than a predetermined speed, it can be determined that there is a risk factor for an accident involving the vehicle (200).

[0147] If it is determined that there is a risk factor for a cause-and-run accident, the video-based vehicle cause-and-run accident prediction and notification provision server (100) can generate and provide a driving guide notification to the driver, including a stop notification before the driver enters the road. Here, the stop notification before entering the road may include a “stop and wait instruction” as a driver instruction.

[0149] As described above, according to the image-based vehicle accident prediction and notification provision method and the system implementing the same according to one embodiment of the present invention, the occurrence of various types of accidents is predicted and a risk notification is provided based on images captured through a vehicle camera, thereby enabling the driver to prepare for accident risks in advance.

[0150] In addition, according to the image-based vehicle accident prediction and notification provision method and the system implementing the same according to one embodiment of the present invention, the speed and acceleration of a vehicle are measured through changes in the image of a captured video and utilized to predict the occurrence of an accident, thereby preventing collisions with objects such as other vehicles and inducing the driver to brake the vehicle.

[0151] In addition, according to the image-based vehicle accident prediction and notification provision method and the system implementing the same according to one embodiment of the present invention, the driving habits of each driver are reflected in the prediction model and the accuracy of accident prediction can be improved by continuously learning from data analyzed within the server and accident-related data collected from the outside.

[0153] The term "part" as used in this embodiment refers to a software or hardware component, such as a field-programmable gate array (FPGA) or an ASIC, and the "part" performs certain roles. However, the meaning of "part" is not limited to software or hardware. The "part" may be configured to reside in an addressable storage medium or configured to run one or more processors. Thus, as an example, the "part" includes components such as software components, object-oriented software components, class components, and task components, as well as processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functions provided within the components and "parts" may be combined into a smaller number of components and "parts" or further separated into additional components and "parts." In addition, the components and '~parts' may be implemented to play one or more CPUs within the device or secure multimedia card.

[0154] Although the invention has been described above with reference to embodiments, this is merely illustrative and does not limit the invention. Those skilled in the art will understand that various modifications and applications not exemplified above are possible within the scope of the essential characteristics of the embodiments. For example, each component specifically shown in the embodiments may be modified and implemented. Furthermore, differences related to such modifications and applications should be interpreted as being included within the scope of the invention as defined in the appended claims.

Claims

Claim 1 A method comprising: receiving a vehicle surrounding image captured by at least one camera mounted on a vehicle via a video-based vehicle accident prediction and notification provision server; analyzing a plurality of image files included in the vehicle surrounding image to recognize one of a pre-stored driving situation; creating a virtual area matching the recognized driving situation and overlapping it with the vehicle surrounding image; monitoring whether an object of interest corresponding to a pre-set detection condition is detected within the overlapping virtual area in the vehicle surrounding image; if the object of interest is detected, determining the presence or absence of a risk factor for an accident based on image changes in the vehicle surrounding image; providing a driving guide notification to at least one of a guidance device mounted on the vehicle and a driver terminal based on the determination result; extracting a portion of image of a predetermined length from the vehicle surrounding image that includes a point in time where the risk factor for an accident is determined to exist; cumulatively collecting the extracted portion of image; and analyzing the collected plurality of portions of image to generate driving data of the driver. The method further includes a step of changing at least one condition data used to determine the presence or absence of a risk factor for the accident based on the driving data, wherein the driving data includes guidance compliance data regarding whether the driver drove in accordance with the driving guide notification when the driving guide notification is provided, and reaction time data regarding the reaction time taken to respond to the danger after the driving guide notification occurs, and the step of determining the presence or absence of a risk factor for the accident based on image changes in the vehicle surrounding video when the object of interest is detected comprises: a step of analyzing at least one position / range change data among the images of the object of interest and external objects in the vehicle surrounding video; and a step of calculating the relative speed between the object of interest and the vehicle and the acceleration of the vehicle based on the position / range change data.A method for providing image-based vehicle accident prediction and notification, comprising the step of determining whether there is a risk factor for an accident by comparing the relative speed and the acceleration with a pre-stored accident prediction range, wherein the external object is a stationary object including at least one of a road, lane, road surface marking, crosswalk, traffic light, and traffic sign, wherein the acceleration is calculated by the change in the range ratio of the external object image among a plurality of image images of the same time interval included in the vehicle surrounding image, and the relative speed is calculated by the change in the range ratio of the object of interest image among a plurality of image images of the same time interval. Claim 2 delete Claim 3 delete Claim 4 A video-based vehicle accident prediction and notification provision method according to claim 1, wherein the accident prediction range regarding the first driving situation is characterized in that the relative speed has a negative (-) value, and if it is determined that there is an accident risk factor regarding the first driving situation, a driving guide notification including a vehicle stop waiting instruction is generated. Claim 5 A video-based vehicle accident prediction and notification provision method according to claim 1, wherein the accident prediction range regarding the second driving situation is characterized in that the acceleration has a positive (+) value, and if it is determined that there is an accident risk factor regarding the second driving situation, a driving guide notification including deceleration and braking instructions for the vehicle is generated. Claim 6 In claim 1, the step of analyzing a plurality of image images included in the vehicle surrounding image to recognize one of the pre-stored driving situations comprises: a step of detecting at least one situation recognition object included in the plurality of image images; a step of calculating a matching rate for each of the pre-stored driving situations by comparing the type, shape, color, location, size, angle, range, and change state of the at least one detected situation recognition object with standard data for each driving situation; and a step of recognizing the driving situation to which the matching rate belongs as the driving situation of the vehicle when any of the matching rates exceeds a recognition threshold value. Claim 7 In claim 6, the situation recognition object includes at least one of a road, lane, road surface marking, crosswalk, traffic light, traffic sign, vehicle door, surrounding vehicle, structure, and pedestrian, and the pre-stored driving situation includes at least one of a road stopping situation, a U-turn situation, a side road driving situation, an intersection entry situation, a signal stop situation, and an off-road entry / exit situation, a video-based method for predicting and notifying vehicle-caused accidents. Claim 8 A method for providing image-based vehicle accident prediction and notification, wherein, in claim 1, if it is determined that there is a risk factor for the accident, the method further comprises the step of generating the driving guide notification including at least one of a warning signal, accident information, driver instruction information, and virtual guidance graphics, wherein the driver instruction information includes instructions for stopping, shifting gears, braking, and steering of the vehicle. Claim 9 delete Claim 10 A vehicle comprising at least one camera for capturing images of the vehicle's surroundings and a guidance device for outputting at least one of visual content and auditory content; and an image-based vehicle accident prediction and notification provision server comprising a processor that receives and analyzes the captured images of the vehicle's surroundings and determines a risk situation for an accident based on the results of the analysis; The system includes a driver terminal that receives a driving guide notification according to the risk situation of the accident based on the image-based vehicle accident prediction and notification provision server, and the processor analyzes a plurality of image files included in the vehicle surrounding image to recognize one of the pre-stored driving situations, creates a virtual area matching the recognized driving situation and overlaps it with the vehicle surrounding image, monitors whether an object of interest corresponding to a pre-set detection condition is detected within the overlapped virtual area in the vehicle surrounding image, and if the object of interest is detected, determines the presence or absence of the accident risk factor based on the image change of the vehicle surrounding image, generates the driving guide notification based on the determination result, extracts an image portion of a predetermined length including a point in time where the accident risk factor is determined to exist in the vehicle surrounding image, cumulatively collects the extracted image portion, analyzes the collected plurality of image portions to generate the driver's driving data, and modifies at least one condition data used to determine the presence or absence of the accident risk factor based on the driving data. The driving data includes guidance compliance data regarding whether the driver drove in accordance with the driving guide notification when the driving guide notification is provided, and reaction time data regarding the reaction time taken to respond to the danger after the driving guide notification occurs, and when determining the presence or absence of the risk factor for the accident, the processor analyzes at least one position and range change data among the images of the object of interest and external objects in the vehicle surrounding images, andA video-based vehicle accident prediction and notification provision system that calculates the relative speed between the object of interest and the vehicle and the acceleration of the vehicle based on the above location and range change data, determines the presence or absence of the accident risk factor by comparing the relative speed and the acceleration with a pre-stored accident prediction range, and is characterized in that the external object is a stationary object including at least one of a road, lane, road marking, crosswalk, traffic light, and traffic sign, wherein the acceleration is calculated by the change in the range ratio of the external object image among a plurality of image images of the same time interval included in the vehicle surrounding image, and the relative speed is calculated by the change in the range ratio of the object of interest image among a plurality of image images of the same time interval.

Images