Collapse discrimination method based on lane line extension

Abstract

The invention belongs to the technical field of intelligent driving assistance, relates to a collapse judgment method based on lane line extension, and aims to accurately predict road collapse. Comprising the steps of acquiring navigation road section data and lane line real-time morphological characteristics; the navigation road section data refers to a data set containing current positioning information of the vehicle, a road section line shape in a front preset range and lane parameters. The lane line real-time morphological features refer to a feature set of lane line positions, continuity and extension directions extracted by identifying real-time road surface images collected by a vehicle foresight camera. And generating a lane line pre-estimation form template based on the navigation road section data. The lane line pre-estimation form template refers to a front preset range lane line theoretical extension track and a feature set which are obtained through pre-judgment according to a front road section line shape and lane parameters. And comparing the real-time morphological characteristics of the lane line with the estimated morphological template of the lane line, and judging whether morphological difference exists or not. And if the morphological difference meets a preset collapse risk condition, triggering a collapse early warning prompt.

Description

Technical Field

[0001] This invention relates to the field of intelligent driving assistance technology, and in particular to a method for detecting lane line extension collapse. Background Technology

[0002] When driving at night, factors such as insufficient lighting and driver fatigue significantly reduce a driver's ability to recognize sudden hazards such as road collapses, increasing the risk of accidents. Current lane departure warning systems in vehicles can only identify lane lines based on real-time images to determine if the vehicle is deviating from its lane; they cannot predict or identify lane line interruptions caused by collapses. Traditional navigation software only provides route planning and directional guidance, without integrating with the vehicle's vision system, and therefore cannot provide warnings of road risks based on changes in lane line morphology.

[0003] The existing technology has the following drawbacks: First, it relies solely on real-time image recognition, which is prone to lane line recognition errors due to poor image quality in low-light environments at night, and it is difficult to distinguish between normal lane line breaks and abnormal interruptions caused by collapses; Second, it lacks the ability to predict the extension pattern of lane lines, and drivers can only react after visually noticing the lane line break when a collapse occurs, resulting in insufficient reaction time; Third, it is not adapted for nighttime scenarios, does not consider the impact of low-light environments on recognition accuracy, and has low reliability of warnings. Summary of the Invention

[0004] To address the problems mentioned in the background art, this invention provides a collapse detection method based on lane line extension, aiming to accurately predict road collapses.

[0005] This invention adopts the following technical solution: This invention provides a collapse discrimination method based on lane line extension, including: S1: Acquiring navigation road segment data and real-time lane line morphological features. Navigation road segment data refers to a dataset containing the vehicle's current location information and the alignment and lane parameters of a preset range of road segments ahead. Real-time lane line morphological features refer to a set of features extracted from real-time road surface images captured by the vehicle's forward-facing camera, including lane line position, continuity, and extension direction. S2: Generating a lane line prediction morphological template based on the navigation road segment data. The lane line prediction morphological template refers to the theoretical extension trajectory and feature set of lane lines within a preset range ahead, predicted based on the alignment and lane parameters of the road segment ahead. S3: Comparing the real-time lane line morphological features with the lane line prediction morphological template to determine if there are morphological differences. S4: If the morphological difference meets preset collapse risk conditions, a collapse warning is triggered.

[0006] Step S1, which involves acquiring navigation road segment data, includes: establishing real-time data interaction with the navigation software; acquiring navigation road segment data containing road alignment type, number of lanes, and lane width through data interaction; and associating the navigation road segment data with the vehicle's current location information to determine the boundaries of the road segment data within a preset range ahead.

[0007] The process of acquiring real-time road surface images in step S1 includes: detecting the current ambient light intensity; if the environment is low-light, controlling the vehicle's forward-facing camera to activate night vision mode; acquiring real-time road surface images covering the current lane and adjacent lanes using the forward-facing camera; performing occlusion detection on the real-time road surface images to ensure there is no lens obstruction interference before performing lane line recognition to obtain the real-time morphological features of the lane lines.

[0008] Step S2, generating the lane line prediction template, includes: extracting the alignment parameters of the road segment ahead and the current lane width parameters from the navigation segment data; converting the alignment parameters into theoretical parameters in the image coordinate system; and combining the theoretical parameters in the image coordinate system with the lane width parameters to draw the continuous extension trajectory of the lane line, thus forming the lane line prediction template.

[0009] The process of determining morphological differences in step S3 includes: matching the positional information in the real-time morphological features of the lane lines with the corresponding positional information in the lane line prediction morphological template at a preset time period; calculating the deviation value after the positional information matching; and determining whether there is an interruption in the continuity of the real-time morphological features of the lane lines. After excluding normal lane line breakpoint scenarios, if the deviation value exceeds a preset deviation threshold or there is an interruption in continuity, it is determined that there is a morphological difference.

[0010] Step S3, which determines whether the morphological difference meets the preset collapse risk conditions, includes: statistically analyzing the morphological difference determination results over multiple consecutive preset time periods. If all consecutive determination results indicate the presence of morphological differences, and interference factors such as abnormal navigation data or abnormal image acquisition are excluded, then the morphological difference is determined to meet the preset collapse risk conditions, triggering a collapse warning.

[0011] The process of triggering the collapse warning in step S4 includes: controlling the vehicle's central control screen to display a red warning icon; controlling the vehicle's steering wheel to vibrate and activating the voice broadcast device to provide a risk warning; and continuing to display the warning icon, vibrate the steering wheel, and broadcast the voice warning until a vehicle deceleration operation is detected or the morphological difference disappears.

[0012] The collapse detection method based on lane line extension also includes a false positive elimination step: real-time monitoring of the update status of navigation road segment data, the working status of the forward-facing camera, and the ambient light intensity. If a navigation data update delay, camera obstruction, or insufficient light intensity causing blurred imaging is detected, the morphological comparison operation in step S3 is paused, and an environment incompatibility warning message is output.

[0013] Step S1, which involves extracting real-time lane line morphology features, includes: processing the real-time road surface image using a deep learning algorithm; identifying solid white lines, solid yellow lines, dashed white lines, and dashed yellow lines in the image; and based on the identification results, extracting the position, continuity, and extension direction of each type of lane line to form real-time lane line morphology features.

[0014] The lane line extension-based collapse detection method also includes a warning recovery step: after triggering a collapse warning, the system continuously monitors changes in morphological differences and vehicle operation. If the morphological differences disappear, or if the driver slows down, the collapse warning is terminated.

[0015] Compared to existing technologies, the beneficial effects are as follows: 1. This application provides a collapse detection method based on lane line extension. Through a dual-linkage mechanism of navigation segment data acquisition and real-time visual recognition of lane lines, it overcomes the shortcomings of existing technologies that rely solely on real-time image recognition and are susceptible to interference in low-light environments, or that rely solely on navigation and cannot perceive changes in lane line morphology. Navigation segment data provides the theoretical basis for lane line extension, while real-time visual recognition captures the actual morphology. The combination of these two methods can accurately distinguish between normal lane line breaks and abnormal interruptions caused by collapses, significantly reducing detection errors in nighttime scenarios and improving the accuracy of collapse risk identification.

[0016] 2. Compared to existing technologies that can only passively identify lane line interruptions, this application can pre-determine the theoretical extension trajectory of lane lines within a preset range by generating lane line prediction templates. By comparing the real-time shape with the prediction template, it can proactively and quickly capture lane line shape differences caused by collapses, providing more reaction time for drivers who are fatigued at night and effectively reducing traffic accidents caused by delayed reactions. Through a dual misjudgment elimination design, on the one hand, normal scenarios such as stop lines at intersections and transitions between solid and dashed lines are excluded to avoid misjudging non-collapse factors as risks; on the other hand, the navigation data update status, camera working status, and light intensity are monitored in real time. If data delays, lens obstruction, or image blurring occur, the comparison is immediately paused and a warning is issued to avoid false alarms or missed alarms caused by environmental interference, ensuring the reliability of the warning signal. Attached Figure Description

[0017] Figure 1 This is a flowchart of a collapse detection method based on lane line extension provided in an embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] This application provides a collapse detection method based on lane line extension, for example, such as... Figure 1 As shown, it includes: S1: Acquire navigation road segment data and real-time lane line morphology features. Navigation road segment data refers to a dataset containing the vehicle's current location information and the alignment and lane parameters of a preset road segment ahead. Real-time lane line morphology features refer to a set of features extracted from real-time road surface images captured by the vehicle's forward-facing camera, including lane line position, continuity, and direction of extension.

[0020] Step S1, which involves acquiring navigation road segment data, includes: establishing real-time data interaction with the navigation software; acquiring navigation road segment data containing road alignment type, number of lanes, and lane width through data interaction; and associating the navigation road segment data with the vehicle's current location information to determine the boundaries of the road segment data within a preset range ahead.

[0021] For example, a real-time data connection is established with navigation software to obtain the vehicle's current latitude and longitude, driving speed and other positioning information, while simultaneously synchronizing detailed information about the road segment 50m-100m ahead, including the road segment alignment (straight line / circular curve / transition curve), curve curvature, number of lanes, lane width and road segment length.

[0022] As one possible implementation, the process of acquiring real-time road surface images in step S1 includes: detecting the current ambient light intensity; if in a low-light environment, controlling the vehicle's forward-facing camera to activate night vision mode; acquiring real-time road surface images covering the current lane and adjacent lanes using the forward-facing camera; performing occlusion detection on the real-time road surface images to ensure there is no lens obstruction interference before performing lane line recognition to obtain the real-time morphological features of the lane lines.

[0023] As one possible implementation, step S1, which involves extracting real-time lane line morphological features, includes: processing the real-time road surface image using a deep learning algorithm; identifying white solid lines, yellow solid lines, white dashed lines, and yellow dashed lines in the image; and based on the identification results, extracting the position, continuity, and extension direction of each type of lane line to form real-time lane line morphological features.

[0024] In terms of navigation segment data acquisition, step S1 establishes real-time data interaction with navigation software, associates the vehicle's current positioning information to determine the data boundaries of the preset road segment ahead, and simultaneously acquires core parameters such as road segment alignment type, number of lanes, and lane width. This solves the deficiency of traditional navigation systems, which can only provide route guidance but cannot provide data basis for lane alignment determination. The acquired road segment data is deeply bound to the vehicle's positioning, which can accurately lock the range of the road segment to be monitored ahead, providing a theoretical benchmark that fits the actual road conditions for subsequent lane alignment prediction, avoiding prediction deviations caused by data disconnection, and improving the basic data accuracy of the judgment system from the source.

[0025] In terms of real-time lane line morphology feature acquisition, step S1 optimizes the image acquisition logic for low-light nighttime scenarios: It automatically activates the camera's night vision mode through light intensity detection, combines occlusion detection to ensure interference-free images, and then uses a deep learning algorithm to identify white or yellow solid and dashed lines and extract features such as position and continuity. This effectively solves the problems of poor image quality, incomplete lane line recognition, or large errors in existing technologies under low-light conditions. Night vision mode and occlusion detection ensure image source quality, while the deep learning algorithm achieves comprehensive recognition of various lane line types, avoiding subsequent judgment failures due to missed or incorrect lane line identification. Simultaneously, the extracted multi-dimensional features can fully reflect the actual state of the lane lines, providing rich feature dimensions for comparison with predicted morphologies, further improving the accuracy of difference recognition. This provides crucial data support for rapid identification of lane collapses at night, indirectly extending driver reaction time and improving nighttime driving safety.

[0026] S2: Generate lane line prediction templates based on navigation road segment data. A lane line prediction template refers to the theoretical extension trajectory and feature set of lane lines within a preset range ahead, predicted based on the road segment alignment and lane parameters.

[0027] Step S2, generating the lane line prediction template, includes: extracting the alignment parameters of the road segment ahead and the current lane width parameters from the navigation segment data; converting the alignment parameters into theoretical parameters in the image coordinate system; and combining the theoretical parameters in the image coordinate system with the lane width parameters to draw the continuous extension trajectory of the lane line, thus forming the lane line prediction template.

[0028] For example, the lane line prediction template includes the lane line's extension direction, continuous length, and width range. For instance, in the prediction template for a straight section 100m ahead, the lane line extends horizontally continuously without interruption.

[0029] Step S2 extracts real road condition parameters from navigation segment data, such as road alignment and lane width, to ensure that the prediction template matches the actual road characteristics and avoids idealized predictions that deviate from road conditions. This solves the problems of existing technologies lacking theoretical basis and having low prediction accuracy. Step S2 converts the road alignment parameters into theoretical parameters of the image coordinate system, achieving coordinate unification between navigation data and vehicle forward-facing camera image data. This removes the obstacle of coordinate mismatch for the subsequent comparison of real-time lane line shape and prediction template in step S3, avoiding comparison errors caused by differences in data dimensions. In addition, by combining coordinate parameters and lane width, a continuous extension trajectory is drawn to form a template containing features such as extension direction and continuous length. This pre-determines the theoretical shape of the lane line ahead, breaking the limitation of existing technologies that passively wait for lane line interruptions for identification. It provides a clear reference for quickly capturing lane line anomalies caused by collapses, indirectly giving drivers more reaction time. At the same time, it provides a comprehensive and accurate comparison basis for subsequent difference judgment, improving the timeliness and accuracy of collapse detection at night.

[0030] S3: Compare the real-time morphological features of the lane lines with the estimated morphological template of the lane lines to determine whether there are any morphological differences.

[0031] The process of determining morphological differences in step S3 includes: matching the positional information in the real-time morphological features of the lane lines with the corresponding positional information in the lane line prediction morphological template at a preset time period; calculating the deviation value after the positional information matching; and determining whether there is an interruption in the continuity of the real-time morphological features of the lane lines. After excluding normal lane line breakpoint scenarios, if the deviation value exceeds a preset deviation threshold or there is an interruption in continuity, it is determined that there is a morphological difference.

[0032] For example, using a comparison period of 100ms, the lane line position coordinates identified in real time in step S2 are matched with the corresponding positions of the generated lane line prediction template. The deviation between the actual coordinates and the predicted coordinates is calculated. If the deviation exceeds a threshold, it is marked as a position difference. At the same time, the continuity of the actual lane line is checked. If there is an interruption in the predicted trajectory with a length exceeding a threshold, it is marked as a continuity difference. During the comparison process, normal breakpoint scenarios are automatically excluded, such as the identification of intersection stop lines and transition sections between solid and dashed lines. These scenarios are not marked as differences.

[0033] As one possible implementation, step S3, determining whether the morphological difference meets the preset collapse risk conditions, includes: statistically analyzing the morphological difference determination results over multiple consecutive preset time periods. If all consecutive determination results indicate the presence of morphological differences, and interference factors such as abnormal navigation data and abnormal image acquisition are excluded, then the morphological difference is determined to meet the preset collapse risk conditions, triggering a collapse warning.

[0034] For example, if positional differences and continuity differences are detected in three consecutive comparison cycles (300ms in total), it is determined that the lane line is abnormally interrupted, which meets the characteristics of collapse risk.

[0035] S4: If the shape difference meets the preset collapse risk conditions, a collapse warning will be triggered.

[0036] The process of triggering the collapse warning in step S4 includes: controlling the vehicle's central control screen to display a red warning icon; controlling the vehicle's steering wheel to vibrate and activating the voice broadcast device to provide a risk warning; and continuing to display the warning icon, vibrate the steering wheel, and broadcast the voice warning until a vehicle deceleration operation is detected or the morphological difference disappears.

[0037] The collapse detection method based on lane line extension also includes a false positive elimination step: real-time monitoring of the update status of navigation road segment data, the working status of the forward-facing camera, and the ambient light intensity. If a navigation data update delay, camera obstruction, or insufficient light intensity causing blurred imaging is detected, the morphological comparison operation in step S3 is paused, and an environment incompatibility warning message is output.

[0038] By monitoring the real-time update status of navigation road segment data, the working status of the forward-looking camera, and the ambient light intensity, interference is investigated from both the data source and hardware environment. This avoids deviations in real-time lane line feature recognition caused by navigation data delays leading to distortion of the lane line prediction template, camera obstruction, or low-light imaging blurring. In turn, it prevents misjudgments caused by erroneous data in step S3 form comparison, thus solving the problem of existing technologies failing to eliminate environmental and equipment interference and being prone to false alarms.

[0039] The lane line extension-based collapse detection method also includes a warning recovery step: after triggering a collapse warning, the system continuously monitors changes in morphological differences and vehicle operation. If the morphological differences disappear, or if the driver slows down, the collapse warning is terminated.

[0040] This application also provides a collapse detection system based on lane line extension, including a navigation data import module, an image acquisition and lane line recognition module, a lane line shape prediction module, a shape difference comparison module, and a warning response module.

[0041] For example, taking a collapse detection system based on lane line extension as an example, the entire detection process is explained as follows: The navigation data import module, as the system's data source, first establishes a stable real-time data interaction link with the Gaode navigation application through the vehicle's Bluetooth communication module. Based on the need for nighttime driving safety monitoring, the module presets the information acquisition range of the road segment ahead to 80m ahead of the vehicle's current position. At the same time, it obtains the vehicle's real-time positioning information through the Global Positioning System (GPS) module integrated into the navigation software, with a positioning accuracy of 8m and a data update cycle of 1s to ensure the timeliness and accuracy of road condition data. Finally, the navigation data import module synchronously acquires and parses the core information of the road segment ahead 80m (including the alignment of straight sections, the number of lanes, and the width of a single lane of 3.75m), providing realistic and practical theoretical data support for subsequent lane line shape prediction.

[0042] The image acquisition and lane line recognition module handles visual perception, establishing a hardware communication connection with the vehicle's built-in forward-facing camera to continuously acquire real-time images of the road ahead. Considering the impact of low-light conditions at night on image quality, the module defaults to enabling the camera's night vision enhancement mode to ensure the clarity of lane line features in the images. In the lane line recognition stage, the module uses the YOLO-Lane deep learning algorithm to perform pixel-level processing on the acquired images, focusing on identifying the solid white line of the current lane and outputting three core feature parameters: the precise position coordinates of the lane line in the image coordinate system, the continuity of the lane line (measured at 98%), and the actual physical width of the lane line (measured at 25cm). Parameter verification shows that the recognition result conforms to the characteristic standards of normal lane lines, providing a reliable basis for subsequent morphological comparison.

[0043] The lane line shape prediction module takes the "80m straight section and 3.75m lane width" data output by the navigation data import module as input. It uses a spatial geometric conversion algorithm to convert the road segment alignment parameters into theoretical parameters in the image coordinate system that match the forward-looking camera image. Based on these theoretical parameters, the module further draws the continuous extension trajectory of the lane lines, ultimately generating a lane line prediction shape template. In the template, the lane lines extend horizontally and continuously within the image coordinate system, with the x-axis coordinate ranging from 100 to 1800 pixels and the y-axis coordinate remaining stable within the range of 510 ± 10 pixels, without any interruptions. This template serves as the core theoretical comparison benchmark for subsequent judgment of whether lane lines are abnormal.

[0044] The morphological difference comparison module uses a 100ms monitoring cycle to continuously perform point-to-point matching and verification between the actual lane line features output by the image acquisition and lane line recognition module and the predicted morphological template generated by the lane line morphology prediction module. In the first monitoring cycle, the module detected an interruption when the actual lane line's x-axis extended to 1500 pixels (corresponding to a position of 65m on the actual road surface). Compared with the continuous extension trajectory of 1800 pixels on the x-axis in the predicted template (corresponding to a position of 80m on the actual road surface), there was a 25m interruption gap, and the deviation between the actual lane line position and the theoretical position of the template reached 15cm. In the second and third monitoring cycles, the module repeatedly detected the same interruption and deviation features, and used the built-in normal breakpoint recognition algorithm to exclude common lane line breakpoint scenarios such as stop lines and transition sections between solid and dashed lines, ultimately determining that the current lane line morphological difference exceeded the standard.

[0045] When the morphological difference comparison module determines that the morphological difference exceeds the standard for three consecutive monitoring cycles (cumulative 300ms), the warning response module immediately triggers a tiered warning mechanism: First, it controls the vehicle's central control display screen to show a red warning icon for an abnormal lane ahead, providing a direct visual reminder; second, it drives the steering wheel to vibrate at a frequency of 8Hz, stimulating the driver's attention through tactile feedback; third, it activates the in-vehicle voice broadcast system, repeatedly playing the prompt "Abnormal lane ahead, please slow down and observe." Once the driver receives the warning signal and performs a speed reduction operation (reducing the vehicle speed from 60km / h to 40km / h), the warning prompt automatically terminates after 5 seconds to avoid excessive interference with driving.

[0046] The system also has an anomaly detection and error correction mechanism. When the image acquisition and lane line recognition module detects that the forward-facing camera is obscured by rain and the image is blurred due to the image clarity detection algorithm, the protection logic is immediately triggered: on the one hand, the subsequent comparison operation of the morphological difference comparison module is suspended to avoid misjudgment based on the blurred image; on the other hand, the driver is prompted with the message "The current environment does not meet the judgment conditions, please drive with caution" through the central control screen, guiding the driver to actively increase driving vigilance, forming a dual safety guarantee of system suspension and manual prevention, further reducing the risk of false alarms when driving at night.

[0047] The above description, in conjunction with preferred technical solutions, provides a further detailed explanation of the present invention and should not be construed as limiting the specific implementation of the invention to these descriptions. For those skilled in the art, simple deductions and substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A collapse detection method based on lane line extension, characterized in that, include: S1: Acquire navigation road segment data and real-time lane line morphology features; wherein, the navigation road segment data refers to a dataset containing the vehicle's current location information and the alignment and lane parameters of a preset range of road segments ahead; the real-time lane line morphology features refer to a set of features extracted from real-time road surface images captured by the vehicle's forward-facing camera, including lane line position, continuity, and extension direction; S2: Generate a lane line prediction morphology template based on the navigation road segment data; the lane line prediction morphology template refers to a set of features and parameters of the theoretical extension trajectory of lane lines within a preset range ahead, predicted based on the alignment and lane parameters of the road segments ahead; S3: Compare the real-time lane line morphology features with the lane line prediction morphology template to determine if there are any morphological differences; S4: If the morphological differences meet preset collapse risk conditions, trigger a collapse warning.

2. The collapse detection method based on lane line extension according to claim 1, characterized in that, The process of obtaining navigation road segment data in step S1 includes: establishing real-time data interaction with navigation software; obtaining navigation road segment data containing road segment alignment type, number of lanes, and lane width through the data interaction; associating the navigation road segment data with the vehicle's current positioning information to determine the road segment data boundary of the preset range ahead.

3. The collapse detection method based on lane line extension according to claim 1, characterized in that, The process of acquiring real-time road surface images in step S1 includes: detecting the current ambient light intensity; if it is in a low-light environment, controlling the vehicle's front-view camera to turn on night vision mode; acquiring real-time road surface images covering the current lane and adjacent lanes through the front-view camera; performing occlusion detection on the real-time road surface images to ensure that the images are not interfered with by lens occlusion, and then performing lane line recognition to obtain the real-time morphological features of the lane lines.

4. The collapse detection method based on lane line extension according to claim 1, characterized in that, The process of generating the lane line prediction template in step S2 includes: extracting the alignment parameters of the road segment ahead and the current lane width parameters from the navigation road segment data; converting the alignment parameters of the road segment into theoretical parameters in the image coordinate system; and combining the theoretical parameters in the image coordinate system with the lane width parameters to draw the continuous extension trajectory of the lane line and form the lane line prediction template.

5. The collapse detection method based on lane line extension according to claim 1, characterized in that, The process of determining the morphological difference in step S3 includes: matching the position information in the real-time morphological features of the lane line with the corresponding position information of the lane line estimated morphological template at a preset time period; calculating the deviation value after the position information is matched, and determining whether there is an interruption in the continuity of the real-time morphological features of the lane line; after excluding normal lane line breakpoint scenarios, if the deviation value exceeds the preset deviation threshold or there is an interruption in continuity, it is determined that there is a morphological difference.

6. The collapse detection method based on lane line extension according to claim 5, characterized in that, The process of determining whether the morphological difference meets the preset collapse risk conditions in step S3 includes: statistically analyzing the morphological difference determination results within multiple consecutive preset time periods; if the consecutive determination results all show that there is a morphological difference, and interference factors such as abnormal navigation data and abnormal image acquisition are excluded, then the morphological difference is determined to meet the preset collapse risk conditions, triggering a collapse warning prompt.

7. The collapse detection method based on lane line extension according to claim 1, characterized in that, The process of triggering the collapse warning in step S4 includes: controlling the vehicle's central control screen to display a red warning icon; controlling the vehicle's steering wheel to vibrate and activating the voice broadcast device to provide risk warnings; continuing the display of the warning icon, the vibration of the steering wheel, and the voice broadcast until a vehicle deceleration operation is detected or the shape difference disappears.

8. The collapse detection method based on lane line extension according to claim 1, characterized in that, It also includes a misjudgment elimination step: real-time monitoring of the update status of the navigation road segment data, the working status of the forward-looking camera, and the ambient light intensity; if a navigation data update delay, camera obstruction, or insufficient light intensity causing blurry imaging is detected, the morphological comparison operation in step S3 is paused, and an environment incompatibility prompt message is output.

9. The collapse detection method based on lane line extension according to claim 3, characterized in that, The process of extracting real-time lane line morphology features in step S1 includes: processing the real-time road surface image using a deep learning algorithm; identifying white solid lines, yellow solid lines, white dashed lines, and yellow dashed lines in the image; and based on the identification results, extracting the position, continuity, and extension direction of each type of lane line to form real-time lane line morphology features.

10. A collapse detection method based on lane line extension according to claim 6, characterized in that, It also includes a warning recovery step: after triggering a collapse warning, continuously monitor the changes in the shape difference and the vehicle operation status; if the shape difference disappears or the driver performs a vehicle speed reduction operation, then terminate the collapse warning.

Images