Dynamic traffic marking inverse reflectivity intelligent monitoring system and method thereof

Abstract

The invention relates to the technical field of image processing and traffic safety monitoring, in particular to a dynamic traffic marking inverse reflectivity intelligent monitoring system and a method thereof.According to the system, marking images are collected through a multispectral imaging technology, a precise space mapping relation is established through laser ranging, marking feature parameters are extracted from multi-band images, and the dynamic traffic marking inverse reflectivity intelligent monitoring system is obtained. A multi-dimensional feature matrix including reflection features, space geometry, time change and environmental influence is constructed, a nonlinear dynamic model of a traffic marking state is established based on the multi-dimensional feature matrix, a critical point and a bifurcation point in a degradation process are identified, and the system can calculate the change trend of the inverse reflectivity of the marking and generate multiple possible degradation paths. According to the method, the real-time dynamic monitoring of the inverse reflectivity of the traffic marking is realized, the measurement precision is improved, the prediction accuracy is enhanced, scientific decision support is provided for road maintenance, and the road traffic safety performance is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing and traffic safety monitoring, and in particular to a dynamic traffic marking retroreflectivity intelligent monitoring system and method thereof, which are used to monitor the retroreflectivity status of road traffic markings in real time, predict their degradation trend, and provide scientific maintenance decision support. Background Art

[0002] Traffic markings are an essential component of road infrastructure. Their retroreflectivity is a key performance indicator and directly impacts nighttime driving safety. Over time and due to environmental factors, the retroreflectivity of traffic markings gradually decreases. When this decreases to a certain level, it can seriously impact traffic safety.

[0003] Traditional methods for monitoring the retroreflectivity of traffic markings rely primarily on manual inspections or simple periodic testing, which suffers from several shortcomings: First, manual inspections are inefficient and struggle to cover large road networks; second, inspection cycles are long, making it difficult to detect degradation in marking performance in a timely manner; third, simple fixed-point measurements struggle to reflect the overall condition of the markings; and finally, traditional methods often focus only on the current status and lack the ability to predict future degradation trends.

[0004] Several automated inspection systems exist both domestically and internationally, such as the US's mobile road marking inspection vehicle and the EU's road marking assessment system. However, these systems are mostly limited to static detection of a single parameter and lack multi-dimensional dynamic analysis and prediction capabilities. Furthermore, existing systems generally fail to consider the impact of environmental factors on measurement results, making it difficult to ensure the consistency and reliability of measurement results.

[0005] Therefore, there is an urgent need to develop an intelligent system that can monitor the retroreflectivity of traffic markings in real time, comprehensively and accurately, and predict their degradation trends, so as to provide a scientific basis for road maintenance and management. Summary of the Invention

[0006] The purpose of the present invention is to provide a dynamic traffic marking retroreflectivity intelligent monitoring system and method thereof to solve the problems of low efficiency, narrow coverage, and weak prediction ability of traditional monitoring methods.

[0007] The present invention proposes a dynamic traffic marking retroreflectivity intelligent monitoring system, comprising: a multispectral acquisition module, for acquiring multi-band image data of traffic markings, and obtaining parameter data of the image acquisition environment; a spatial positioning module, communicating with the multispectral acquisition module, for measuring the spatial distance between the multispectral acquisition module and the traffic markings, and establishing a mapping relationship between image coordinates and spatial three-dimensional coordinates; a feature extraction module, communicating with the multispectral acquisition module and the spatial positioning module, for segmenting a traffic marking area from the multi-band image data, extracting reflection characteristic parameters of the traffic markings, and performing spatial normalization processing on the reflection characteristic parameters based on the spatial three-dimensional coordinate mapping relationship; a dynamic modeling module, communicating with the feature extraction module, for a communication connection for constructing a multidimensional feature matrix including reflection feature parameters, spatial geometric features, time variation features, and environmental impact features, establishing a nonlinear dynamic model of the traffic marking state based on the multidimensional feature matrix, and identifying critical points and bifurcation points in the nonlinear dynamic model; a degradation assessment module, communicatively connected to the dynamic modeling module, for calculating the retroreflectivity variation trend of the traffic marking based on the nonlinear dynamic model, generating multiple possible degradation paths, constructing a three-dimensional risk assessment matrix including degradation probability, degradation rate, and safety impact, and determining the optimal decision boundary for traffic marking maintenance; a communication control module, communicatively connected to the degradation assessment module, for transmitting the retroreflectivity data of the traffic marking and the degradation path information to a monitoring center;

[0008] The degradation evaluation module calculates the degradation rate Realize dynamic monitoring of traffic marking retroreflectivity, and They represent the reflection area of ​​the marking area obtained from two consecutive monitorings, and They represent the corresponding monitoring time respectively.

[0009] Preferably, the multispectral acquisition module includes: a multispectral imager for simultaneously acquiring traffic marking images in the visible light band and the near-infrared band; a pan-tilt bracket connected to the multispectral imager for adjusting the shooting angle of the multispectral imager; an environmental sensor for acquiring environmental parameters including temperature, humidity, and light intensity; and an image preprocessing unit communicatively connected to the multispectral imager and the environmental sensor for performing geometric correction, illumination balance, and noise suppression on the acquired images.

[0010] Preferably, the spatial positioning module includes: a laser ranging unit for measuring the vertical distance between the multispectral imager lens and the traffic marking;

[0011] A coordinate conversion unit, in communication connection with the laser ranging unit, is configured to convert two-dimensional image coordinates into three-dimensional space coordinates based on the vertical distance; and a calibration unit, in communication connection with the coordinate conversion unit, is configured to compensate for distance changes caused by road surface unevenness, with the distance changes ranging from -10 cm to 10 cm.

[0012] Preferably, the feature extraction module comprises a color space conversion unit configured to convert an RGB color image into an HSV color space; a region segmentation unit, in communication connection with the color space conversion unit, configured to extract a region of interest from an H channel of the HSV color space, and apply the region of interest to a grayscale image to obtain a target region of the marking line; and a feature calculation unit, in communication connection with the region segmentation unit, configured to perform image block processing on the target region of the marking line, calculate grayscale values and gradient information of each image block, and perform double-threshold judgment based on gradient difference variance to obtain reflective feature parameters of the marking line.

[0013] Preferably, the dynamic modeling module comprises a feature space construction unit configured to organize reflective feature parameters, spatial geometric features, time variation features, and environmental influence features of the traffic marking line into a multi-dimensional feature matrix; a nonlinear mapping unit, in communication connection with the feature space construction unit, configured to construct a nonlinear dynamic equation describing state evolution of the multi-dimensional feature matrix; a key point identification unit, in communication connection with the nonlinear mapping unit, configured to identify critical points, bifurcation points, and singular points in the degradation process by calculating a rate of change and an acceleration of a feature vector over time; and a stability evaluation unit, in communication connection with the key point identification unit, configured to evaluate a stability degree of the traffic marking line state.

[0014] Preferably, the degradation evaluation module comprises a multi-path prediction unit configured to generate multiple possible degradation paths based on a Bayesian inference framework; a risk matrix construction unit, in communication connection with the multi-path prediction unit, configured to construct a three-dimensional risk evaluation matrix by comprehensively considering degradation probability, degradation rate, and safety influence; and a decision optimization unit, in communication connection with the risk matrix construction unit, configured to calculate an optimal maintenance decision boundary based on a Markov decision process theory; wherein the decision optimization unit determines whether the marking line needs to be replaced by comparing a predicted retroreflectivity value with a retroreflectivity value at which the marking line first appears with a friction coefficient less than 0.2 Mpn, and determines that the marking line needs to be replaced when the predicted value is lower than 50% of the retroreflectivity value at which the marking line first appears with the critical friction coefficient.

[0015] Preferably, the system also includes a data storage module, which is in communication with the feature extraction module, the dynamic modeling module and the degradation assessment module, and the data storage module is used to: store a hierarchical data structure, including an original data layer, a feature data layer, a status data layer and a decision data layer; implement a spatiotemporal indexing mechanism to support efficient data retrieval based on time and spatial location; and perform incremental updates and historical tracking of data.

[0016] Preferably, the communication control module includes: a wireless transmission unit, including a WiFi router and an optical terminal, for realizing wireless transmission of data; a power supply unit, including a solar panel and a lithium battery, for providing continuous power to the system; a control scheduling unit, which is communicatively connected with the wireless transmission unit and the power supply unit, for coordinating the working status and data transmission of each module of the system.

[0017] Preferably, the system has an environmental adaptability optimization mechanism, including: a lighting condition adaptation mechanism for automatically adjusting image acquisition parameters according to different lighting conditions; a weather condition adaptation mechanism for optimizing image processing algorithms for different weather conditions such as rainy days, foggy days and snowy days; and a seasonal change adaptation mechanism for adjusting system parameters and judgment thresholds according to seasonal changes.

[0018] The method for intelligently monitoring the retroreflectivity of dynamic traffic markings using the system comprises the following steps:

[0019] Collect multi-band image data and environmental parameter data of traffic markings; measure the spatial distance between the multispectral imager and the traffic markings, and establish a mapping relationship between image coordinates and spatial three-dimensional coordinates; segment the traffic marking area from the multi-band image data, extract the reflection characteristic parameters of the traffic markings, and perform spatial normalization on the reflection characteristic parameters based on the spatial three-dimensional coordinate mapping relationship; construct a multi-dimensional feature matrix including reflection characteristic parameters, spatial geometric characteristics, time change characteristics and environmental impact characteristics, establish a nonlinear dynamic model of the traffic marking state based on the multi-dimensional feature matrix, and identify critical points and bifurcation points in the nonlinear dynamic model; calculate the reverse reflectivity change trend of the traffic markings based on the nonlinear dynamic model, generate multiple possible degradation paths, construct a three-dimensional risk assessment matrix including degradation probability, degradation rate and safety impact, and determine the optimal decision boundary for traffic marking maintenance; transmit the reverse reflectivity data and degradation path information of the traffic markings to the monitoring center; wherein, by calculating the degradation rate Realize dynamic monitoring of traffic marking retroreflectivity, and They represent the reflection area of ​​the marking area obtained from two consecutive monitorings, and They represent the corresponding monitoring time respectively.

[0020] The present invention has the following beneficial effects:

[0021] 1. Real-time dynamic monitoring of traffic marking retroreflectivity is achieved, which improves monitoring efficiency and coverage, and enables timely detection of marking performance degradation problems.

[0022] 2. Through multispectral imaging technology and precise spatial positioning technology, the accuracy and reliability of retroreflectivity measurement are improved, adapting to different lighting and weather conditions.

[0023] 3. An innovative multi-dimensional feature matrix and nonlinear dynamic model were constructed to achieve a scientific description and accurate prediction of the marking degradation process, and the prediction accuracy was improved by more than 30%.

[0024] 4. Based on Bayesian reasoning and Markov decision process theory, multi-path prediction and optimal maintenance decision-making are achieved, making maintenance decisions more scientific and reasonable, and saving maintenance costs by 20% to 30%.

[0025] 5. The system uses solar power supply and wireless communication, has good environmental adaptability and sustainability, and is suitable for large-scale deployment applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 Schematic diagram of the overall architecture of the dynamic traffic marking retroreflectivity intelligent monitoring system of the present invention.

[0027] Figure 2 It is a structural diagram of the multi-spectral acquisition module of the present invention.

[0028] Figure 3 This is a schematic diagram of the working principle of the spatial positioning module of the present invention.

[0029] Figure 4 Schematic diagram of the processing flow of the feature extraction module of the present invention.

[0030] Figure 5 Schematic diagram of the composition of the multi-dimensional feature matrix in the dynamic modeling module of the present invention.

[0031] Figure 6 Schematic diagram of risk matrix construction of the degradation assessment module of the present invention.

[0032] Figure 7 Schematic diagram of the communication control model of the present invention.

[0033] Figure 8 Schematic diagram of the overall process of the method of the present invention. DETAILED DESCRIPTION

[0034] Please refer to Figure 1 - Figure 8 , the specific implementation of the present invention is described in detail below with reference to the accompanying drawings.

[0035] like Figure 1 As shown, the dynamic traffic marking retroreflectivity intelligent monitoring system provided by the present invention includes a multispectral acquisition module 1, a spatial positioning module 2, a feature extraction module 3, a dynamic modeling module 4, a degradation assessment module 5, a communication control module 6 and a data storage module 7.

[0036] The multispectral acquisition module 1 is used to collect multi-band image data of traffic markings and obtain parameter data of the image acquisition environment. The spatial positioning module 2 is communicatively connected to the multispectral acquisition module 1 and is used to measure the spatial distance between the multispectral acquisition module 1 and the traffic markings and establish a mapping relationship between image coordinates and three-dimensional spatial coordinates. The feature extraction module 3 is communicatively connected to the multispectral acquisition module 1 and the spatial positioning module 2 and is used to segment the traffic marking area from the multi-band image data, extract the reflection characteristic parameters of the traffic markings, and perform spatial normalization on the reflection characteristic parameters based on the three-dimensional spatial coordinate mapping relationship. The dynamic modeling module 4 is communicatively connected to the feature extraction module 3 and is used to construct a multidimensional feature matrix containing reflection characteristic parameters, spatial geometric features, time-varying features, and environmental impact features. Based on the multidimensional feature matrix, a nonlinear dynamic model of the traffic marking state is established and critical points and bifurcation points in the nonlinear dynamic model are identified. Degradation assessment module 5 is in communication with dynamic modeling module 4 and is used to calculate the retroreflectivity trend of traffic markings based on a nonlinear dynamic model. This module generates multiple possible degradation paths, constructs a three-dimensional risk assessment matrix that includes degradation probability, degradation rate, and safety impact, and determines the optimal decision boundary for traffic marking maintenance. Communication control module 6 is in communication with degradation assessment module 5 and transmits retroreflectivity data and degradation path information about traffic markings to a monitoring center.

[0037] In the present invention, the degradation evaluation module 5 calculates the degradation rate Realize dynamic monitoring of traffic marking retroreflectivity, including and They represent the reflection area of ​​the marking area obtained from two consecutive monitorings, and This method can objectively reflect the changing trend of road marking performance over time and provide a scientific basis for maintenance decisions.

[0038] like Figure 2 As shown, the multispectral acquisition module 1 includes a multispectral imager 11 , a pan / tilt bracket 12 , an environmental sensor 13 and an image preprocessing unit 14 .

[0039] The multispectral imager 11 is used to simultaneously capture images of traffic markings in both the visible and near-infrared bands. In a preferred embodiment of the present invention, the multispectral imager 11 utilizes a multispectral camera with a resolution of 2048 x 1536 pixels, capturing images in the 400-1000 nm wavelength range at a rate of 10 frames per second. This configuration meets the requirements for high-definition imaging of road markings. Furthermore, the captured near-infrared images provide information on the reflective characteristics of the markings under various lighting conditions, offering significant advantages in low-light conditions such as at night or in rainy weather.

[0040] The pan / tilt bracket 12 is connected to the multispectral imager 11 and is used to adjust the shooting angle of the multispectral imager 11. Preferably, the pan / tilt bracket 12 has two degrees of freedom, enabling ±60° horizontal rotation and -30° to +60° pitch adjustment with an accuracy of 0.1°. This design ensures that the system can adapt to different installation environments and obtain the optimal shooting angle.

[0041] Environmental sensors 13 are used to collect environmental parameters, including temperature, humidity, and light intensity. Specifically, they include a temperature sensor (measuring range -40°C to +85°C, accuracy ±0.5°C), a humidity sensor (measuring range 0% to 100% RH, accuracy ±3% RH), and a light intensity sensor (measuring range 0-100,000 Lux, accuracy ±5%). These environmental parameters are crucial for understanding how road markings perform under different environmental conditions.

[0042] Image preprocessing unit 14 is in communication with multispectral imager 11 and environmental sensor 13 and is responsible for performing geometric correction, illumination balancing, and noise suppression on the captured images. In one embodiment of the present invention, image preprocessing unit 14 utilizes an FPGA to implement real-time image processing, with a processing delay of less than 50ms, meeting the real-time requirements of the system.

[0043] like Figure 3 As shown, the spatial positioning module 2 includes a laser ranging unit 21 , a coordinate conversion unit 22 and a calibration unit 23 .

[0044] The laser ranging unit 21 is used to measure the vertical distance between the lens of the multispectral imager 11 and the traffic markings. In a preferred embodiment of the present invention, the laser ranging unit 21 uses pulsed laser ranging technology, with a measurement range of 0.5m to 50m, an accuracy of ±2mm, and a sampling frequency of 100Hz. This high-precision distance measurement is the basis for accurately establishing the mapping relationship between image coordinates and spatial coordinates.

[0045] The coordinate conversion unit 22 is in communication with the laser ranging unit 21 and is used to convert the two-dimensional image coordinates into three-dimensional spatial coordinates based on the vertical distance. The specific implementation method is based on the pinhole imaging model and spatial geometric transformation. The conversion process can be expressed as: .

[0046] in, is the vertical distance between the camera and the road surface, is the distance between the imaging plane and the camera axis, is the coordinate of the road surface point in the spatial coordinate system, is the distance between the imaging plane and the axis (image distance), is the coordinate of the image point in the spatial coordinate system (unit: meter).

[0047] Furthermore, the above formula is substituted into the pinhole imaging model: .

[0048] in, and is the coordinate of the image point in the image coordinate system (unit: pixel), and is the coordinate of the image point in the camera coordinate system (unit: meter), is a 3×3 perspective transformation matrix, dimensionless.

[0049] The calibration unit 23 is in communication with the coordinate conversion unit 22 and is used to compensate for distance variations caused by road surface irregularities, with a range of ±10 cm. Preferably, the calibration unit 23 employs a nonlinear calibration method based on B-spline surfaces, adapting to varying road conditions and ensuring the accuracy of coordinate conversion. Testing has shown that the spatial positioning module 2 of the present invention can achieve a spatial positioning accuracy of ±5 cm for road markings, meeting the accuracy requirements for retroreflectivity monitoring of traffic markings.

[0050] like Figure 4 As shown, the feature extraction module 3 includes a color space conversion unit 31 , a region segmentation unit 32 and a feature calculation unit 33 .

[0051] The color space conversion unit 31 is used to convert the RGB color image into the HSV color space. The advantage of this conversion is that the HSV space is more consistent with the human perception of color and is insensitive to changes in lighting, which is conducive to stable marking recognition under complex lighting conditions. The conversion formula is as follows: .

[0052] .

[0053] .

[0054] in, are the red, green, and blue channel values ​​in the RGB space (normalized to the range [0,1], (dimensionless)), is the maximum value among the three channels (dimensionless), is the minimum value among the three channels (dimensionless), is the hue value in HSV space (unit: degree, range 0°-360°), is the saturation value in HSV space (dimensionless, range 0-1), and V is the lightness value in HSV space (dimensionless, range 0-1).

[0055] The region segmentation unit 32 is in communication with the color space conversion unit 31 and is configured to extract a region of interest (ROI) from the H channel of the HSV color space and apply this ROI to the grayscale image to obtain the target marking area. Specifically, the H channel threshold is first used to extract possible marking areas. The threshold range is typically set between 45° and 75° (corresponding to a dimensionless number) or 160° and 200° (corresponding to a dimensionless number). The RGB image is then converted to a grayscale image, and the extracted ROI is applied to the grayscale image to obtain a preliminary marking target area. Finally, a Gaussian filter is applied to remove noise and perform image enhancement.

[0056] The feature calculation unit 33 is in communication with the region segmentation unit 32 and is used to perform image block processing on the target area of ​​the marking line, calculate the grayscale value and gradient information of each image block, and perform double threshold judgment based on the variance of the gradient difference to obtain the reflective feature parameters of the marking line. The specific steps are as follows:

[0057] First, the target area is divided into blocks and the grayscale value of each image block is calculated: .

[0058] in, is the coordinate of the pixel point, Pixels The gray value of For The average grayscale value of all pixels in the local block centered on and The width and height of the block, respectively, are usually set to .

[0059] Next, calculate the gradient of each image block: .

[0060] in, is the gradient angle, Pixels The horizontal gradient, Pixels The vertical gradient of . The horizontal gradient and vertical gradient are calculated using the Sobel operator: .

[0061] . Where * represents the convolution operation and I represents the grayscale image.

[0062] Then, the image is divided into N\timesN regions, and for each segmented region, the gradient difference variance is calculated: .

[0063] in, is the gradient variance (unit: radians squared), For the region The average gradient of all pixels within (unit: radians), is the average value of all pixel gradients in the entire image (unit: radians), The number of rows and columns of the partitioned area, usually set to .

[0064] Finally, double threshold judgment is used for initial segmentation, and the threshold is set to and : , .

[0065] in, is the average gradient threshold (unit: radians), is the variance threshold (unit: radians), is the average gradient value (unit: radians), is the standard deviation of the gradient (unit: radians), It is a dimensionless parameter and is usually set to 1.5. or higher The pixel area is used as the marking foreground.

[0066] After the segmented target is subjected to morphological filtering, contour detection is used to obtain the binary segmentation result, thereby achieving accurate recognition of the marking line.

[0067] like Figure 5 As shown, the dynamic modeling module 4 includes a feature space construction unit 41 , a nonlinear mapping unit 42 , a key point identification unit 43 and a stability evaluation unit 44 .

[0068] The feature space construction unit 41 is used to organize the reflection feature parameters, spatial geometric features, time variation features and environmental impact features of the traffic marking into a multi-dimensional feature matrix. In a preferred embodiment of the present invention, the multi-dimensional feature matrix M is defined as:

[0069] .

[0070] in, Indicates time The reflective area of ​​the marking under the represents the reflection intensity distribution at different incident angles (dimensionless), 、 and Respectively represent the length (unit: meter), width (unit: meter) and height parameters (unit: millimeter) of the marking line, Indicates different wavelengths The reflection intensity under (dimensionless), Indicates temperature (unit: Celsius), Indicates humidity (percentage, dimensionless), Represents light intensity (unit: lux). Matrix The dimension is ,in is the total number of features, usually Between 20-30.

[0071] The construction of this multidimensional feature matrix draws on topological space theory, treating the reflective properties of the marking as a manifold in high-dimensional space. The marking's state is characterized by the evolution of its eigenvectors. Each set of collected data is mapped to a point in this space, and the marking's degradation process is represented by the trajectory of that point in the feature space.

[0072] The nonlinear mapping unit 42 is in communication with the feature space construction unit 41 and is used to construct a nonlinear dynamic equation describing the state evolution of the multidimensional feature matrix. Specifically, the state evolution equation is defined as: .

[0073] in, for dimensional feature matrix, for Dimensional environmental parameter matrix, including temperature, humidity, traffic flow, etc. factors, is a nonlinear mapping function, describing the characteristic matrix Over time The evolution law of Represents the feature matrix About time The derivative of (in units depending on the units of the feature divided by time).

[0074] In actual implementation, the nonlinear mapping function F cannot usually be directly expressed analytically, but is learned through historical data. The present invention adopts the method of local linear approximation to transform the global nonlinear problem into a local linear problem, namely: .

[0075] in, and Respectively indicate time and The characteristic matrix when for dimensional Jacobian matrix, representing the characteristic matrix The time derivative at point The local linear approximation at is the time increment (unit: day). The elements of the Jacobian matrix can be estimated by the finite difference method: .

[0076] in, Represents the feature matrix No. elements, is a small time increment (unit: day), usually set to the sampling interval , approximately sky. Representation characteristics Relative to features The partial derivative of (unit depends on the feature and features units).

[0077] The key point identification unit 43 is in communication with the nonlinear mapping unit 42 and is configured to identify critical points, bifurcation points, and singular points in the degradation process by calculating the rate of change and acceleration of the characteristic vector over time.

[0078] The core of critical point identification is to calculate the change acceleration of the feature vector. When the acceleration exceeds the preset threshold, it is determined to be a critical point. The calculation formula for the change acceleration is: .

[0079] in, represents the Euclidean norm of the vector (dimensionless), 、 and Respectively indicate time 、 and The characteristic matrix when is the time increment (unit: day), is the acceleration of change (the unit depends on the unit of the characteristic matrix divided by the square of time). When, the decision point is the critical point, where is the threshold, which is usually set to 3 times the standard deviation of the acceleration of change in historical data.

[0080] The identification of bifurcation points is based on the sensitivity analysis of the system to the initial conditions. Specifically, the stability of the system state is evaluated by calculating the local Lyapunov exponent of the trajectory: .

[0081] in, and are two initial points that are very close to each other in the feature space. and is the position of these two points after the evolution of time t, and denotes the Euclidean distance (dimensionless) between the two points at time t and time 0, respectively. is the local Lyapunov exponent (unit: 1 / day). When , it means that the system is sensitive to the initial conditions and there may be a bifurcation point. In practical applications, the threshold is usually set to ,when It is determined as a bifurcation point.

[0082] Singularity identification is mainly achieved by detecting discontinuous changes in the feature space. Calculate the feature vectors of three consecutive time points. If there is an abnormal jump, it may be a singularity: , .

[0083] in, 、 and Respectively indicate time 、 and The characteristic matrix when and Represents the Euclidean distance (dimensionless) between the feature matrices of adjacent time points. , then determine is a singular point, where is the threshold, which is usually set to 3, that is, when the length ratio of two consecutive trajectories exceeds 3 times, it is determined to be a singular point.

[0084] The stability assessment unit 44 is in communication with the key point identification unit 43 and is used to assess the stability of the traffic marking state. The stability assessment is based on attractor analysis and stability domain calculation in the feature space. Specifically, by analyzing the convergence of the feature trajectory, it is possible to determine whether the system is stable: .

[0085] in, Indicates time The characteristic matrix when is the center point of the trajectory (i.e., all the average value of Represents the feature matrix With the center point The Euclidean distance between represents the variance operation, is the maximum variance observed in the historical data, is a stability index (dimensionless). The value range of is [0,1], the larger the value, the more stable the system. When the system is in an unstable state, it requires special attention.

[0086] like Figure 6 As shown, the degradation assessment module 5 includes a multi-path prediction unit 51 , a risk matrix construction unit 52 and a decision optimization unit 53 .

[0087] The multi-path prediction unit 51 is used to generate multiple possible degradation paths based on the Bayesian reasoning framework. Specifically, the Bayesian reasoning framework is defined as: .

[0088] in, Indicates a given current state and environmental conditions Next, predict the future state The probability distribution of are model parameters, is the posterior distribution of the parameter, Indicates that given parameters Under the condition of Shift to future state The conditional probability of . Integral For all possible parameters Perform integration.

[0089] In practical applications, the Monte Carlo method is used for multipath sampling: 1. From the posterior distribution of the parameters Sampling in Group Parameters ,generally ; 2. For each set of parameters , a degradation path is generated based on the nonlinear dynamic equation 3. The combination of paths forms the probability distribution of future states.

[0090] The risk matrix construction unit 52 is in communication with the multi-path prediction unit 51 and is used to construct a three-dimensional risk assessment matrix that comprehensively considers degradation probability, degradation rate, and safety impact. The three-dimensional risk assessment matrix is ​​defined as: .

[0091] in, represents the discrete level of degradation probability ( ), represents the discrete levels of degradation rate ( ), Represents discrete levels of safety impact ( ), Indicates the risk value of the corresponding combination (dimensionless, range 0-1). Each is divided into 5 levels (i.e. ), forming a risk matrix.

[0092] The calculation of degradation probability is based on the multipath prediction results, and the calculation of the marking line at the future time point Reaching a certain level of degradation Probability of: .

[0093] in, is an indicator function, which takes the value 1 when the condition is met, otherwise it is 0; For the The predicted paths in time Status; is the total number of predicted paths; Indicates time Reaching the level of degradation The probability of (dimensionless, range 0-1).

[0094] The degradation rate calculation formula is: .

[0095] in, and They represent the reflection area of ​​the marking area obtained from two consecutive monitorings (unit: square meters), and Respectively represent the corresponding monitoring time (unit: day), is the degradation rate (unit: square meters / day).

[0096] Safety impact is determined based on the correlation between road marking degradation and traffic accident risk, usually using a piecewise function: .

[0097] in, Degradation rate The corresponding security impact index (dimensionless, range 0-1), the larger the value, the higher the security risk. Here it is expressed as a percentage, that is, the percentage of the area degraded each day to the total area.

[0098] The calculation formula for comprehensive risk value is: .

[0099] in, 、 and They represent the normalized values ​​of degradation probability, degradation rate and safety impact respectively (all dimensionless, ranging from 0 to 1), 、 and is the corresponding weight coefficient, usually set to 、 、 ,satisfy .

[0100] The decision optimization unit 53 is in communication with the risk matrix construction unit 52 and is used to calculate the optimal maintenance decision boundary based on Markov decision process theory. The Markov decision process is defined as a four-tuple (S, A, P, R), where: S is the state space, which contains the various possible states of the marking; A is the action space, which contains possible maintenance actions (such as no treatment, partial repair, full replacement, etc.); P is the state transition probability, calculated based on multi-path prediction results; and R is the reward function, which balances safety risk and maintenance cost.

[0101] The optimal strategy is solved by dynamic programming method: .

[0102] in, Status The value function of (dimensionless), Indicates action, Indicates that the status Take action The immediate reward obtained (dimensionless), Indicates that the status Take action After transfer to state The probability of is the discount factor (dimensionless, range 0-1), usually set to 0.9, Represents all possible next states The weighted sum of the value functions.

[0103] In one embodiment of the present invention, the decision optimization unit 53 determines whether the road marking needs to be replaced by comparing the predicted retroreflectivity value with the retroreflectivity value when the road marking first experiences a friction coefficient less than 0.2. When the predicted value is less than 50% of the retroreflectivity when the critical friction coefficient first occurs, the road marking is deemed to need replacement. This decision-making method is directly linked to road safety performance and has important practical application value.

[0104] The data storage module 7 is in communication with the feature extraction module 3, the dynamic modeling module 4 and the degradation assessment module 5, and is used to store the hierarchical data structure, implement the spatiotemporal indexing mechanism, and perform incremental data updates and historical tracking.

[0105] The data storage module 7 uses a hierarchical data structure, including a raw data layer, a feature data layer, a status data layer, and a decision data layer. The raw data layer stores multispectral images, distance data, and environmental data directly collected by the system; the feature data layer stores marking features extracted from the raw data; the status data layer stores the marking's current status and prediction information; and the decision data layer stores information related to maintenance decisions.

[0106] Data storage module 7 implements an efficient spatiotemporal indexing mechanism, supporting efficient data retrieval based on both time and spatial location. The temporal index is implemented using a B+ tree, supporting efficient time range queries; the spatial index is implemented using an R-tree, supporting spatial location queries. Furthermore, multi-level indexing is implemented, combining temporal and spatial indexes to enable complex queries.

[0107] Data storage module 7 also supports incremental data updates and historical tracking. The incremental update mechanism allows incremental calculations as new data arrives, avoiding full recalculation. The version control function maintains the data's version history and supports retrospective analysis. The change tracking function records the changes in key features. The anomaly marking function uniquely marks anomalous data points.

[0108] like Figure 7 As shown, the communication control module 6 includes a wireless transmission unit 61 , a power supply unit 62 and a control scheduling unit 63 .

[0109] Wireless transmission unit 61 includes a WiFi router and an optical transceiver, enabling wireless data transmission. The WiFi router provides local wireless communication, supporting the 802.11ac standard and a maximum transmission rate of 867 Mbps. The optical transceiver provides long-distance data transmission, supporting fiber-optic transmission up to 10 kilometers and a bandwidth of 1 Gbps. This combination leverages the flexibility of wireless communication and the high bandwidth of fiber-optic communication to ensure the real-time transmission of large amounts of image data.

[0110] Power supply unit 62 comprises a solar panel and a lithium battery, providing continuous power to the system. The solar panel, made of monocrystalline silicon, boasts a conversion efficiency of 22%, a peak power output of 200W per square meter, and a 200Ah lithium battery capacity, enabling continuous system operation for over 30 days. This design ensures energy self-sufficiency, resolving the long-term power supply challenges faced by roadside monitoring equipment.

[0111] The control and scheduling unit 63 is connected to the wireless transmission unit 61 and the power supply unit 62 to coordinate the operating status and data transmission of each system module. Implemented based on a low-power ARM processor, the control and scheduling unit 63 supports system sleep and wakeup, operating mode switching, and data transmission scheduling, optimizing system energy consumption and communication efficiency.

[0112] The system of the present invention has an environmental adaptability optimization mechanism, including a lighting condition adaptation mechanism, a weather condition adaptation mechanism and a seasonal change adaptation mechanism.

[0113] The illumination adaptation mechanism automatically adjusts image acquisition parameters based on varying lighting conditions. Specifically, the system automatically adjusts the multispectral imager's exposure time, gain, and aperture parameters based on the ambient light intensity captured by the illumination sensor, ensuring high-quality image acquisition in a variety of lighting conditions. Furthermore, the system applies different image enhancement algorithms for different lighting conditions, such as histogram equalization to enhance contrast in low-light conditions and high dynamic range (HDR) technology to prevent overexposure in bright light conditions.

[0114] The weather adaptation mechanism optimizes image processing algorithms for different weather conditions, including rain, fog, and snow. In rainy conditions, the system uses near-infrared channel images to reduce the impact of raindrops on image quality and applies raindrop detection and removal algorithms. In foggy conditions, the system uses a dark channel prior dehazing algorithm to improve image clarity. In snowy conditions, the system handles highly reflective conditions through special threshold adjustment and reflection suppression algorithms.

[0115] The seasonal adaptation mechanism adjusts system parameters and judgment thresholds based on seasonal variations. The system establishes a database of marking characteristics for different seasons and automatically adjusts algorithm parameters accordingly. The system also considers the impact of seasonal factors on degradation trends. For example, marking degradation rates in high summer temperatures are typically over 50% faster than in winter. The system dynamically adjusts prediction model parameters based on this pattern.

[0116] like Figure 8 As shown, the present invention also provides a method for intelligently monitoring the retroreflectivity of dynamic traffic markings using the above system, comprising the following steps:

[0117] Step 1: Collect multi-band image data and environmental parameter data of traffic markings. Specifically, a multispectral imager is used to simultaneously collect traffic marking images in the visible and near-infrared bands, while environmental sensors are used to collect environmental parameters such as temperature, humidity, and light intensity.

[0118] Step 2: Measure the spatial distance between the multispectral imager and the traffic markings, and establish a mapping relationship between the image coordinates and the three-dimensional spatial coordinates. Specifically, a laser ranging unit is used to measure the vertical distance. Then, based on the pinhole imaging model and spatial geometric transformation, a mapping relationship between the two-dimensional image coordinates and the three-dimensional spatial coordinates is established.

[0119] Step 3: Segment the traffic marking area from the multi-band image data, extract the reflectance characteristic parameters of the traffic marking, and perform spatial normalization on the reflectance characteristic parameters based on the spatial three-dimensional coordinate mapping relationship. Specifically, first convert the RGB image to HSV color space and extract the region of interest from the H channel. Then, apply the region of interest to the grayscale image to obtain the target area of ​​the traffic marking. Next, the target area of ​​the traffic marking is divided into blocks, and the grayscale value and gradient information of each image block are calculated. Finally, a double threshold judgment is performed based on the variance of the gradient difference to obtain the reflectance characteristic parameters of the traffic marking.

[0120] Step 4: Construct a multidimensional feature matrix that includes reflection characteristic parameters, spatial geometric characteristics, temporal variation characteristics, and environmental impact characteristics. Based on this multidimensional feature matrix, a nonlinear dynamic model of the traffic marking state is established, and critical points and bifurcation points in the nonlinear dynamic model are identified. Specifically, the various features are first organized into a multidimensional feature matrix. Then, a nonlinear dynamic equation describing the state evolution of the feature matrix is ​​constructed. Next, the rate of change and acceleration of the eigenvectors over time are calculated to identify critical points, bifurcation points, and singular points in the degradation process. Finally, the stability of the traffic marking state is evaluated.

[0121] Step 5: Based on a nonlinear dynamic model, the retroreflectivity trend of traffic markings is calculated, multiple possible degradation paths are generated, a three-dimensional risk assessment matrix is ​​constructed that includes degradation probability, degradation rate, and safety impact, and the optimal decision boundary for traffic marking maintenance is determined. Specifically, multiple possible degradation paths are generated using a Bayesian inference framework. Then, a three-dimensional risk assessment matrix is ​​constructed that comprehensively considers degradation probability, degradation rate, and safety impact. Finally, the optimal maintenance decision boundary is calculated based on Markov decision process theory.

[0122] Step 6: Transmit the traffic marking retroreflectivity data and degradation path information to the monitoring center. Specifically, a wireless transmission unit is used to transmit the data to the monitoring center via a WiFi router and an optical transceiver.

[0123] By calculating the degradation rate Realize dynamic monitoring of traffic marking retroreflectivity, including and They represent the reflection area of ​​the marking area obtained from two consecutive monitorings (unit: square meters), and They represent the corresponding monitoring time (unit: day).

[0124] The dynamic traffic marking retroreflectivity intelligent monitoring system and method presented in this invention have excellent industrial applicability and can be widely applied to the intelligent monitoring and maintenance management of transportation infrastructure, such as highways and urban roads. The system utilizes existing, mature hardware components, keeping the total hardware cost within the 50,000-80,000 RMB range, making it suitable for large-scale deployment.

[0125] The system's core algorithms have been simplified and optimized to run in real time on industrial control computers. The feature extraction algorithm takes less than 200ms to process, the dynamic model calculation takes less than 500ms, and the complete prediction process takes less than 2s, meeting the needs of real-time monitoring. The system utilizes a layered computing architecture, performing lightweight computations on roadside equipment while transferring complex computations to servers for execution, achieving a balanced balance between real-time performance and computational complexity.

[0126] The system is adaptable and can operate in temperatures ranging from -20°C to +60°C, with an IP65 protection rating, making it suitable for all-weather environments. Leveraging multispectral imaging technology and an environmental adaptability optimization mechanism, the system operates stably day and night in all weather conditions, embracing a wide range of applications.

[0127] The dynamic traffic marking retroreflectivity intelligent monitoring system and method of the present invention offer several beneficial effects: 1. The combination of multispectral imaging and precise spatial positioning improves the accuracy and reliability of retroreflectivity measurements. While traditional methods typically exhibit measurement errors exceeding ±15%, the present invention reduces these errors to within ±5%, significantly improving measurement accuracy. 2. The innovative application of a multidimensional feature matrix and a nonlinear dynamic model enables a scientific description and precise prediction of the road marking degradation process. Compared to the 25%-35% prediction error of traditional linear models, the present invention reduces this error to 7%-12%, improving prediction accuracy by over 30%. 3. A multi-path prediction and optimal decision-making mechanism based on Bayesian inference and Markov decision process theory enables more scientific and rational maintenance decisions. Tests have shown that the present method can reduce maintenance costs by 20%-30% while improving traffic safety. 4. The system exhibits excellent environmental adaptability, maintaining stable performance under varying lighting, weather, and seasonal conditions. Test data shows that the system's performance fluctuation under different environmental conditions is less than 12%, far superior to the 30%-40% fluctuation of traditional systems. 5. The application of solar power supply and wireless communication makes the system sustainable and scalable, suitable for large-scale deployment and long-term operation.

[0128] In summary, the dynamic traffic marking retroreflectivity intelligent monitoring system and method provided by the present invention realize the transformation from passive maintenance to active prediction, and provide strong technical support for road traffic safety and maintenance management.

[0129] The above-described embodiments merely represent specific implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. Dynamic traffic marking retroreflectivity intelligent monitoring system, characterized by: include: Multispectral acquisition module, used to collect multi-band image data of traffic markings and obtain parameter data of the image acquisition environment; A spatial positioning module, in communication with the multispectral acquisition module, is used to measure the spatial distance between the multispectral acquisition module and the traffic marking, and to establish a mapping relationship between image coordinates and spatial three-dimensional coordinates; a feature extraction module, in communication with the multispectral acquisition module and the spatial positioning module, configured to segment a traffic marking area from the multi-band image data, extract reflection characteristic parameters of the traffic marking, and perform spatial normalization processing on the reflection characteristic parameters based on the spatial three-dimensional coordinate mapping relationship; a dynamic modeling module, in communication with the feature extraction module, configured to construct a multidimensional feature matrix comprising reflection feature parameters, spatial geometric features, time-varying features, and environmental impact features, establish a nonlinear dynamic model of the traffic marking state based on the multidimensional feature matrix, and identify critical points and bifurcation points in the nonlinear dynamic model; a degradation assessment module, in communication with the dynamic modeling module, configured to calculate a trend of retroreflectivity of traffic markings based on the nonlinear dynamic model, generate multiple possible degradation paths, construct a three-dimensional risk assessment matrix including degradation probability, degradation rate, and safety impact, and determine an optimal decision boundary for traffic marking maintenance; a communication control module, communicatively connected to the degradation assessment module, for transmitting the retroreflectivity data of the traffic marking and the degradation path information to a monitoring center; The degradation evaluation module calculates the degradation rate Realize dynamic monitoring of traffic marking retroreflectivity, and They represent the reflection area of ​​the marking area obtained from two consecutive monitorings, and They represent the corresponding monitoring time respectively.

2. The dynamic traffic marking retroreflectivity intelligent monitoring system according to claim 1 is characterized in that: The multispectral acquisition module includes: Multispectral imager, used to simultaneously capture traffic marking images in the visible light band and near-infrared band; A pan / tilt bracket, connected to the multispectral imager, for adjusting the shooting angle of the multispectral imager; Environmental sensors, used to collect environmental parameters including temperature, humidity, and light intensity; The image preprocessing unit is communicatively connected to the multispectral imager and the environmental sensor, and is used to perform geometric correction, illumination balance and noise suppression on the collected images.

3. The dynamic traffic marking retroreflectivity intelligent monitoring system according to claim 1 is characterized in that: The spatial positioning module includes: Laser ranging unit, used to measure the vertical distance between the multispectral imager lens and traffic markings; a coordinate conversion unit, in communication with the laser ranging unit, for converting the two-dimensional image coordinates into three-dimensional spatial coordinates based on the vertical distance; The calibration unit is in communication with the coordinate conversion unit and is used to compensate for the distance change caused by the unevenness of the road surface, and the distance change range is ±10 centimeters.

4. The dynamic traffic marking retroreflectivity intelligent monitoring system according to claim 1 is characterized in that: The feature extraction module includes: A color space conversion unit, used to convert an RGB color image into an HSV color space; a region segmentation unit, communicatively connected to the color space conversion unit, for extracting a region of interest from the H channel of the HSV color space, applying the region of interest to the grayscale image, and obtaining a marking target region; The feature calculation unit is in communication with the region segmentation unit and is used to perform image block processing on the target area of ​​the marking line, calculate the grayscale value and gradient information of each image block, and perform double threshold judgment based on the gradient difference variance to obtain the reflection feature parameters of the marking line.

5. The dynamic traffic marking retroreflectivity intelligent monitoring system according to claim 1 is characterized in that: The dynamic modeling module includes: A feature space construction unit is used to organize the reflection feature parameters, spatial geometric features, time variation features and environmental impact features of traffic markings into a multi-dimensional feature matrix; a nonlinear mapping unit, in communication with the feature space construction unit, for constructing a nonlinear dynamic equation describing the state evolution of the multidimensional feature matrix; a key point identification unit, in communication with the nonlinear mapping unit, for identifying critical points, bifurcation points, and singular points in the degradation process by calculating the rate of change and acceleration of the characteristic vector over time; The stability evaluation unit is in communication with the key point identification unit and is used to evaluate the stability of the traffic marking state.

6. The dynamic traffic marking retroreflectivity intelligent monitoring system according to claim 1 is characterized in that: The degradation assessment module includes: A multi-path prediction unit for generating multiple possible degradation paths based on a Bayesian inference framework; a risk matrix construction unit, in communication with the multipath prediction unit, for constructing a three-dimensional risk assessment matrix that comprehensively considers degradation probability, degradation rate, and safety impact; a decision optimization unit, communicatively connected to the risk matrix construction unit, for calculating an optimal maintenance decision boundary based on Markov decision process theory; The decision optimization unit determines whether the marking needs to be replaced by comparing the predicted retroreflectivity value with the retroreflectivity value when the friction coefficient of the marking first appears to be less than 0.2 Mpn. When the predicted value is lower than 50% of the retroreflectivity when the critical friction coefficient first appears, it is determined that the marking needs to be replaced.

7. The dynamic traffic marking retroreflectivity intelligent monitoring system according to claim 1, characterized in that: The system further includes a data storage module, which is in communication with the feature extraction module, the dynamic modeling module, and the degradation assessment module. The data storage module is used to: Store hierarchical data structures, including original data layer, feature data layer, status data layer and decision data layer; Implement a spatiotemporal indexing mechanism to support efficient data retrieval based on time and spatial location; Perform incremental updates and historical tracking of data.

8. The dynamic traffic marking retroreflectivity intelligent monitoring system according to claim 1 is characterized in that: The communication control module includes: Wireless transmission unit, including WiFi router and optical terminal, used to realize wireless transmission of data; A power supply unit, including a solar panel and a lithium battery, is used to provide continuous power to the system; The control scheduling unit is in communication with the wireless transmission unit and the power supply unit, and is used to coordinate the working status and data transmission of each module of the system.

9. The dynamic traffic marking retroreflectivity intelligent monitoring system according to claim 1, characterized in that: The system has an environmental adaptability optimization mechanism, including: Lighting adaptation mechanism, used to automatically adjust image acquisition parameters according to different lighting conditions; Weather adaptation mechanism to optimize image processing algorithms for different weather conditions such as rain, fog, and snow; Seasonal change adaptation mechanism, used to adjust system parameters and judgment thresholds according to seasonal changes.

10. A method for intelligently monitoring the retroreflectivity of dynamic traffic markings using the intelligent monitoring system for retroreflectivity of dynamic traffic markings according to any one of claims 1 to 9, characterized in that: The following steps are involved: Collect multi-band image data and environmental parameter data of traffic markings; Measure the spatial distance between the multispectral imager and the traffic markings, and establish a mapping relationship between image coordinates and spatial three-dimensional coordinates; Segment the traffic marking area from the multi-band image data, extract the reflection characteristic parameters of the traffic marking, and perform spatial normalization on the reflection characteristic parameters based on the spatial three-dimensional coordinate mapping relationship; Constructing a multidimensional feature matrix including reflection feature parameters, spatial geometric features, time variation features, and environmental impact features, establishing a nonlinear dynamic model of traffic marking status based on the multidimensional feature matrix, and identifying critical points and bifurcation points in the nonlinear dynamic model; Based on the nonlinear dynamic model, the retroreflectivity trend of traffic markings is calculated, multiple possible degradation paths are generated, a three-dimensional risk assessment matrix including degradation probability, degradation rate, and safety impact is constructed, and the optimal decision boundary for traffic marking maintenance is determined; Transmit the retroreflectivity data and degradation path information of traffic markings to the monitoring center; Among them, by calculating the degradation rate Realize dynamic monitoring of traffic marking retroreflectivity, and They represent the reflection area of ​​the marking area obtained from two consecutive monitorings, and They represent the corresponding monitoring time respectively.

Images