Traffic event detection and signal timing linkage control method based on unmanned aerial vehicle video stream

Abstract

The invention provides a traffic incident detection and signal timing linkage control method based on an unmanned aerial vehicle video stream. The method comprises the following steps: constructing a continuous space function of the position of a construction fence in a road space along with time change based on the unmanned aerial vehicle video stream; analyzing the movement trend of the fence, generating a short-time space predictor for a next signal control period, performing space overlapping calculation on a predicted road occupation range and a lane level road network model, reconstructing and outputting the effective passable length and reduced traffic capacity of each lane in real time, and generating a total effective traffic capacity constraint in each entrance direction; according to the method, the signal timing scheme is continuously generated and updated, rolling updating is triggered on the basis of the latest unmanned aerial vehicle video stream before each signal period is started, so that the timing scheme matched with the dynamic evolution of the enclosure is output, the dynamic change of the construction enclosure can be sensed in real time, and the signal timing is continuously adjusted according to the dynamic change of the construction enclosure. And self-adaptive accurate regulation and control of the traffic signal of the construction road section are realized.

Description

A Traffic Incident Detection and Signal Timing-Based Coordinated Control Method Based on UAV Video Stream Technical Field

[0001] This disclosure relates to the interdisciplinary fields of intelligent traffic control and computer vision, and in particular to a method for traffic event detection and signal timing linkage control based on UAV video streams. Background Technology

[0002] In urban transportation systems, there is a strict matching relationship between road capacity and traffic signal timing. Traditional signal control theory relies on stable road geometry and traffic flow parameters. However, road construction activities dynamically change the number of lanes, effective width, and alignment of road segments, creating time-varying traffic bottlenecks and directly disrupting the steady-state supply conditions upon which signal timing depends. With the rapid development of smart cities and intelligent transportation systems, using drones for traffic monitoring and event detection has become an important means of improving road management efficiency. Drones, with their high-altitude field of view and maneuverability, can quickly capture and record emergencies such as road construction and traffic accidents. Existing technologies have attempted to analyze video streams transmitted from drones to identify the positions of static obstacles such as construction barriers, and input these as fixed event boundaries into downstream signal control systems to generate corresponding timing schemes. These methods can achieve certain results in scenarios where the construction area remains stable.

[0003] However, in actual road construction sites, construction barriers are not static but continuously move forward, shrink, or undergo partial reconstruction as the project progresses, resulting in dynamic changes in the passable road space. The core flaw of existing signal timing methods based on drone detection lies in treating construction barriers as stable, unchanging event boundaries. Once the barriers move, the system still uses the physical boundaries identified in the previous cycle or even earlier to calculate traffic light timings, causing a severe disconnect between the generated timing scheme and the current actual passable space. This control lag, dependent on historical boundaries, leads to a rigidity in the signal timing scheme between adjacent control cycles, making it impossible to flexibly adjust with the construction progress. This results in a mismatch between the traffic light release strategy and the actual passable vehicle paths, wasting traffic capacity and potentially causing unnecessary congestion or even safety hazards at the forefront of the construction area.

[0004] Therefore, there is an urgent need for a linkage control method that can sense the dynamic changes of construction site enclosures in real time and continuously adjust the signal timing accordingly, in order to solve the problems of signal control lag and spatial mismatch caused by relying on static event boundaries, and achieve adaptive and precise control of traffic signals in construction sections. Summary of the Invention

[0005] In view of this, in order to solve the problems caused by the existing technology, this application provides a traffic event detection and signal timing linkage control method based on UAV video stream.

[0006] In a first aspect, this disclosure provides a method for traffic event detection and signal timing-based linkage control based on UAV video streams, the method comprising:

[0007] S1. Construct a continuous spatial function of the position of construction site barriers in road space as a function of time based on UAV video stream;

[0008] S2. Analyze the movement trend of the enclosure based on the continuous spatial function, and generate a short-time spatial prediction for the next signal control cycle. The short-time spatial prediction includes the road occupancy range predicted based on the movement trend of the enclosure.

[0009] S3. Perform spatial overlap calculations between the predicted lane occupancy range and the lane-level road network model, reconstruct and output the effective passable length and reduced capacity of each lane in real time, and generate the total effective capacity constraints for each approach direction.

[0010] S4. Based on the total effective passage capacity constraint of each inlet direction, continuously generate and update the signal timing scheme, and re-execute step S1 to this step before the start of each signal cycle based on the latest UAV video stream to output a timing scheme that matches the dynamic evolution of the enclosure.

[0011] Optionally, S1 includes:

[0012] Process the drone video stream to extract the pixel-level outline of the construction site fence;

[0013] The pixel-level contours are mapped from the image coordinate system to the road coordinate system to obtain a discrete set of location points of the enclosure in the road space.

[0014] Based on the discrete location point set, a continuous spatial function is constructed to characterize the continuous change of the fence boundary position over time.

[0015] Optionally, S2 includes:

[0016] Discretize the continuous spatial function according to the signal period to obtain the position sequence of the fence at multiple consecutive moments, which is aligned with the signal period;

[0017] Based on the location sequence, calculate the average rate of change of the enclosure in the longitudinal and transverse directions of the road;

[0018] Based on the average rate of change, the spatial displacement and boundary position of the enclosure in the next signal cycle are predicted, and the predicted boundary position is extended into a rectangular predicted road occupancy range.

[0019] Optionally, calculating the average rate of change of the construction site fence in the longitudinal and transverse directions of the road includes:

[0020] Based on the aforementioned position sequence, a sliding window sample containing position coordinates at multiple consecutive time points is constructed;

[0021] Calculate the mean position of the sliding window sample in both the vertical and horizontal directions;

[0022] Based on the average position of the current sliding window and the average position of the previous sliding window, the average rate of change in the vertical and horizontal directions is calculated.

[0023] Optionally, S3 includes:

[0024] Instantiate the target road area into multiple lanes with geometric parameters;

[0025] Calculate the overlap length and lane occupancy ratio between the predicted lane occupancy area and each lane;

[0026] The effective passable length of each lane is calculated based on the overlap length, and the reduced capacity of each lane is calculated using a nonlinear reduction model based on the lane occupancy ratio.

[0027] Based on the reduced capacity of each lane, the total effective capacity constraint for each approach direction is generated.

[0028] Optionally, the calculation of the road occupancy ratio includes:

[0029] Calculate the overlap width between the predicted lane occupancy range and the lane in the lateral direction;

[0030] If the lateral overlap width is greater than zero, then the overlap length between the two in the longitudinal direction of the road is further calculated;

[0031] The lane occupancy ratio is determined based on the ratio of the longitudinal overlap length to the original lane length.

[0032] Optionally, the continuous generation and updating signal timing scheme includes:

[0033] Based on the total effective throughput constraints of each import direction, the standardized release weight of each direction is calculated.

[0034] Based on the release weight, an initial value for the green light duration is assigned to each signal phase serving different directions;

[0035] The initial value of the green light duration is constrained by the minimum and maximum green light times, and the period duration is normalized to obtain a phase green light time scheme that can be directly executed.

[0036] Optionally, the step of imposing constraints on the initial value of the green light duration regarding the minimum and maximum green light times, and performing periodic duration normalization, includes:

[0037] The initial value of the green light duration for each phase is limited to between the preset minimum green light time and the maximum green light time to obtain the truncated green light duration for each phase.

[0038] Calculate the sum of the green light durations after truncation for all phases;

[0039] The truncated green light duration of each phase is scaled proportionally so that the sum of the scaled green light durations of each phase equals the total effective green light duration within one signal cycle.

[0040] In a second aspect, this disclosure provides an electronic device including a memory and at least one processor, the memory storing a computer program, and the processor executing the computer program to implement the method of the first aspect described above.

[0041] Thirdly, this disclosure provides a computer storage medium storing a computer program that, when executed, implements the method described in the first aspect.

[0042] The beneficial effects of this disclosure are that, compared with the prior art, this disclosure has the following advantages:

[0043] 1) To address the problem that existing technologies incorrectly treat actual, dynamically moving construction barriers as stable and unchanging event boundaries, this paper describes the barrier as a continuous dynamic spatial constraint object by extracting the barrier outline in real time from continuous UAV video streams and constructing its continuous evolution spatial function over time. This fundamentally changes the way event boundaries are modeled, enabling it to accurately depict the continuous changes such as the barrier's forward movement, contraction, or reconstruction. This provides an accurate data foundation for solving the spatial mismatch problem caused by rigid boundary descriptions.

[0044] 2) Addressing the signal control lag issue arising from existing technologies' reliance on historical event boundaries from the previous cycle or even earlier, this approach analyzes the recent movement trends of the continuous spatial function of road closures to proactively predict their encroachment range in the next signal control cycle. This prediction serves as the direct basis for generating timing schemes. By updating the signal control decision-making basis from outdated historical observations to advanced trend predictions, this approach achieves a shift from lagging control to feedforward control, effectively eliminating the disconnect between the control scheme and current traffic conditions caused by information update delays.

[0045] 3) To address the issue of signal timing schemes becoming fixed between adjacent control cycles due to control lag and unable to be flexibly adjusted to keep pace with construction progress, a rolling update mechanism based on signal cycles is established. Before the start of each cycle, the entire process calculation from event detection and trend prediction to timing generation is re-executed based on the latest video stream data. This ensures that the timing scheme for each signal cycle is tailored to the latest enclosure status and its evolution trend, thereby breaking the fixed nature of the scheme between cycles and realizing continuous, adaptive following and coordinated control of signal timing for the dynamic progress of construction. Attached Figure Description

[0046] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0047] Figure 1 shows a flowchart of the traffic event detection and signal timing linkage control method based on UAV video stream provided in an embodiment of this disclosure;

[0048] Figure 2 shows a flowchart of the road accessibility space parameter reconstruction provided in an embodiment of this disclosure.

[0049] The accompanying drawings have illustrated specific embodiments of this disclosure, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this disclosure to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0050] The present disclosure will be further described below with reference to the accompanying drawings. The following embodiments are only used to illustrate the technical solutions of the present disclosure more clearly, and should not be used to limit the scope of protection of the present disclosure.

[0051] The components of the embodiments of the invention described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0052] In the following, the terms “comprising,” “having,” and their cognates, which may be used in various embodiments of the invention, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as excluding, firstly, the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more features, numbers, steps, operations, elements, components, or combinations thereof.

[0053] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the invention pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the invention.

[0054] In the field of traffic engineering, the core of signal timing optimization lies in achieving a dynamic balance between traffic demand and road capacity. When road sections experience dynamic lane occupancy due to construction, the essence is that the road cross-sectional structure and spatial resource supply change over time. Traditional timing methods based on historical data or fixed parameters are difficult to adapt to such rapid changes in supply. This method addresses this problem by using UAV video streams to perceive the dynamic changes in the construction site in real time, and then adjusting the signal timing scheme accordingly, establishing a complete technical chain from perceiving spatial physical changes to generating traffic control parameters and executing signal commands. The following examples detail how to transform the dynamic geometric changes of the construction area perceived by the UAV into key control parameters in traffic engineering, and ultimately generate an adaptive signal timing scheme.

[0055] Figure 1 is a flowchart of the traffic event detection and signal timing linkage control method based on UAV video stream provided in this embodiment of the present disclosure. As shown in Figure 1, the method may include the following steps:

[0056] S1: Construct a continuous spatial function of the position of construction site barriers in road space as a function of time, based on drone video stream.

[0057] The purpose of this step is to establish accurate dynamic spatial constraint inputs for downstream traffic control. In road and traffic engineering, the movement of construction barriers directly alters the effective width, alignment, and available length of lanes, which are fundamental parameters for calculating road segment capacity and level of service. Continuous video capture is performed on the area containing the construction section and its upstream signal-controlled intersection using cameras mounted on a drone, obtaining the raw drone video stream. To accurately analyze the continuous changes of the construction barriers in the road space, the raw video sequence needs to be constructed into a spatial position function of the barriers that changes continuously over time. This is achieved through the following sub-steps.

[0058] S1.1: Construct a unified time reference and standardize the frame sequence of the raw video stream from the drone.

[0059] Due to factors such as airflow and equipment buffering during drone flight data collection, the timing of video frame acquisition varies. The non-uniform time intervals introduce noise into the subsequent analysis of the continuous spatial changes of the enclosure. Therefore, it is necessary to construct a unified timeline and perform time-aligned resampling on the original video frames to obtain a frame sequence under a standard time reference.

[0060] Specifically, a unified reference time for the start of the video should be set. Typically, the capture time is taken from the first frame of the video stream. Then, a uniform time step is defined. The step size is selected based on the requirements of subsequent processing accuracy and computational efficiency, and its value typically ranges from 0.1 seconds to 1.0 seconds. Based on this, a series of standard, equally spaced time points can be constructed. Its calculation formula is , where i is an integer index that increments from 0, representing the i-th standard time point.

[0061] For each standard time point It is necessary to start from the original uneven video frames. The corresponding standard frame is reconstructed from the middle. This process employs a time-based linear weighted interpolation algorithm. The algorithm first finds the corresponding timestamps in the original timestamp sequence. Two adjacent raw frame time points and ,satisfy Then, using the image data from these two original frames... and The standard frame is calculated using the following weighting formula. :

[0062]

[0063] This formula ensures a smooth transition of image content along the timeline, effectively avoiding visually non-physical jumps in the fence's position that might occur due to directly using adjacent frames. This lays a reliable temporal foundation for the subsequent accurate extraction of the fence's continuous motion. This step yields a video frame sequence with a unified time base and strictly equal frame intervals. .

[0064] S1.2: Extract the pixel-level visible outline of the construction site fence from standardized video frames.

[0065] Obtaining video frame sequences under a standard time reference Next, each frame of the image needs to be processed to accurately identify and extract the visible outline of the construction site fence. In images, construction site fences are usually represented as continuous edge areas with specific colors and textures, which are significantly different from the background such as roads and vehicles.

[0066] For each frame image This method employs edge detection algorithms from computer vision, such as the Canny operator, or deep learning-based semantic segmentation models, to perform pixel-level recognition of the visible edges of the construction site fence. The process outputs a set of numerous pixels that collectively depict the contour shape of the fence in the current image. This set of contour points is denoted as... Its mathematical expression is .in, Let represent the two-dimensional coordinates of the j-th enclosure contour pixel in the image at time i, with the top-left corner of the image as the origin. This is the total number of contour points successfully detected at that moment, which is usually between 50 and 500 points, enough to depict the approximate shape of the fence.

[0067] To enhance the stability and anti-interference capabilities of the contour, and to avoid the influence of a few outliers caused by local image occlusion, reflection, or recognition errors on the overall contour position, the weighted center of the entire contour is further calculated as a reference benchmark. For each contour point... Assign a trusted weight This weight ranges from 0 to 1 and can be determined based on the response strength of edge detection or the confidence level of the segmentation model output; points with low confidence will be assigned smaller weights. Subsequently, a stable reference center for the contour is established. , It is obtained by calculating the weighted average of all contour points, as shown in the formula:

[0068]

[0069] This reference center , This provides a stable, less noise-affected geometric reference point for subsequent steps, thus abstracting the construction site hoarding from the complex video background into a contour object with clear geometric meaning.

[0070] S1.3: Map the fence outline from the pixel coordinate system to the road coordinate system.

[0071] The fence outline obtained in step S1.2 Its reference center remains in the pixel coordinate system of the image, and its coordinates These pixel coordinates only represent positions on a two-dimensional image plane and do not possess true physical scale meaning. To analyze the impact of construction barriers on actual traffic space, these pixel coordinates must be converted to a real-world road coordinate system.

[0072] This conversion requires an imaging geometry model based on the camera. Using the UAV's positioning and attitude determination system, such as GPS and IMU, and camera calibration parameters, a mapping relationship between pixel coordinates and road plane coordinates can be established. To simplify calculations and ensure practicality, this embodiment uses a linear correction model based on scale factor and offset for mapping. This model assumes that the road area is approximately a plane in the image. This assumption aligns with the geometric characteristics of most urban roads and highways, ensuring the engineering practicality of mapping from two-dimensional images to three-dimensional road space, and forming the basis for subsequent accurate traffic space analysis and control.

[0073] The specific mapping formula is as follows: ; ;in, It is a pre-defined reference origin in the image coordinate system. These are the coordinates of a known reference point in the road coordinate system corresponding to the origin. and These are two scaling factors, representing the actual physical length of one pixel in the horizontal and vertical directions of the image, respectively. The unit is meters per pixel, and their values ​​are usually between 0.02 meters per pixel and 0.2 meters per pixel. The specific values ​​are obtained through on-site calibration.

[0074] The above formula can be used to obtain the contour set. All pixels Convert to spatial points in the road coordinate system This allows us to obtain the set of location points of the enclosure in the actual road space. The essence of this step is to transform the movement of the fence at the image level into a change in its position in physical space, so that its subsequent movement analysis and prediction have practical engineering application value.

[0075] S1.4: Construct a continuous spatial function for construction site enclosures based on spatial point sets at discrete time points.

[0076] At different moments The spatial point set of the lower enclosure in the road coordinate system Then, these snapshots of discrete moments need to be integrated into a mathematical model that can continuously describe the change of the fence position over time, that is, to construct a continuous spatial function of the construction fence.

[0077] First, for the point set at each time step The process involves processing to characterize the overall position of the construction site fence. The average coordinates of all contour points at that moment are calculated along the longitudinal direction of the road (typically the driving direction, i.e., the x-axis) and the transverse direction (i.e., the y-axis), and this average is used as the equivalent boundary position of the fence at that moment. The specific calculation formula is as follows: ; Where M is the current time. The number of valid contour points involved in the calculation. It is a point set The coordinates of the j-th point in the given discrete sequence. }and{ } are constructed as two continuous functions of the continuous time variable t using interpolation or curve fitting methods, denoted as . and These two functions characterize the continuous evolution of the boundary positions of the fencing along the longitudinal and transverse directions of the road over time.

[0078] Define the continuous space function of the construction site enclosure as follows: It consists of two continuous functions and Common representation. The process of the continuous evolution of the boundary positions of the fencing along the road's longitudinal and transverse directions over time is depicted, respectively. B(t) is a comprehensive expression of data transfer between steps, and its physical meaning consists of two components. and This is a common manifestation. In subsequent analysis, we will primarily use... and These two continuous functions are analyzed for direction and rate. B(t) comprehensively reflects the spatial position of the enclosure. The continuous evolution process over time t can effectively characterize the dynamic changes of the construction site, such as forward movement, retraction, or partial reconstruction.

[0079] In the technical solution of this disclosure, the time reference is unified and the frame sequence is standardized for the UAV video stream, and the pixel-level outline of the construction site fence is extracted and mapped to the road coordinate system. This ultimately constructs a spatial position function of the fence that can continuously change over time. This step transforms the originally discrete event observation, which relies on single-frame images, into a constrained object model that continuously evolves in physical space. This lays a precise and reliable spatiotemporal data foundation for subsequent accurate analysis of the fence's dynamic movement and for achieving dynamic linkage of signal timing.

[0080] S2: Analyze the movement trend of the enclosure based on the continuous spatial function, and generate a short-time spatial prediction for the next signal control cycle. The short-time spatial prediction includes the road occupancy range predicted based on the movement trend of the enclosure.

[0081] After constructing a continuous spatial function for construction site fencing, the dynamic changes of this function are quantitatively analyzed, and predictions of fencing space occupancy in the near future are generated. Abandoning the traditional approach of relying on fixed event boundaries from the previous cycle, this method proactively predicts the location and encroachment area that the fencing may reach in the next signal control cycle by analyzing its trajectory and speed in the most recent timeframe, thus providing a forward-looking decision-making basis for signal timing. This is achieved through the following sub-steps.

[0082] S2.1: Discretize the continuous spatial function according to the signal period and construct a period-aligned sliding window sample.

[0083] The continuous spatial function B(t) output in step S1 describes the continuous change of the fence position over time. However, the decision-making and execution of the signal control system are carried out at a fixed period. In order to synchronize the continuous physical changes with the discrete control decisions in time, it is first necessary to discretize the continuous spatial function B(t) according to the time scale of the signal control to construct a discrete position sequence that is strictly aligned with the signal period.

[0084] Specifically, let the fixed period of the signal control system be... Its value typically ranges from 30 to 180 seconds, depending on the specific traffic flow at the intersection. The start time of the control cycle requiring prediction and decision-making is defined as... Next, a sampling step size matching the signal period or its frequency division is determined. This step size is typically between 0.2 and 2.0 seconds to ensure a sufficient number of sampling points within a single period to characterize the dynamics of the enclosure. Based on this, a series of discrete sampling time points synchronized with the signal control clock can be generated. The calculation formula is: , where i is an integer index that increments from 0, representing the i-th aligned sampling time.

[0085] At each sampling time From continuous functions and The equivalent boundary positions of the enclosure on the longitudinal and transverse sides of the road at that moment are read and denoted as follows: and This yields a discrete, periodically aligned sequence of fence locations. , its origin and Together they constitute the whole. In actual calculations, they are used separately. and Further analysis will be conducted.

[0086] To analyze the movement trend of the construction site fence, a sliding observation window needs to be constructed over its time series. A window length is defined. It represents the number of consecutive sampling points contained within the window, typically ranging from 3 to 15, corresponding to a time length of approximately [missing information]. Covering recent history from 1 second to 20 seconds. For the current moment... The corresponding sliding window sample set Includes going back from the current moment to the previous moment. The coordinates of all fence positions at each time point, i.e. This window of samples provides the data foundation for the next step of calculating the average movement trend of the construction site fence in the near future.

[0087] S2.2: Calculate the average rate of change of the fence in the longitudinal and transverse directions based on the sliding window sample.

[0088] Based on the constructed sliding window sample set This study quantitatively extracts the instantaneous velocity and direction of the construction site's longitudinal and transverse movements along the road. Since the construction site's advancement is a continuous process, directly calculating the rate using the positional difference between adjacent frames is susceptible to noise from single-point measurements. Therefore, a difference method based on the average value of a time window is employed to obtain a more stable rate of change that better reflects the true trend. This method borrows data smoothing techniques commonly used in short-term traffic flow parameter prediction, effectively filtering out single-point detection noise and making the extracted motion trend more closely resemble the macroscopic process of continuous construction of the construction site.

[0089] First, regarding the current moment corresponding window Calculate all vertical positions within the window respectively. and horizontal position The arithmetic mean, denoted as and The calculation formula is: ; ;

[0090] These two mean points It can be regarded as the average position of the fence within the recent window time, effectively smoothing out the jitter that may be caused by instantaneous measurement.

[0091] The average rate of change of the fencing in both directions is obtained by calculating the difference between the mean of the current window and the mean of the previous window. (Longitudinal rate of change) and lateral rate of change The calculation formula is as follows: ; ;in, and It is the previous sampling time. The average value of the corresponding sliding window position. Rate and The unit is meters per second. A positive value indicates that the fence is advancing forward or expanding outward in the corresponding direction, while a negative value indicates that it is retreating or shrinking.

[0092] To clearly express the direction of propulsion regardless of the speed, further define direction markers. and They are essentially normalized forms of the sign function of the rate value, calculated using the following formula: ; ;in, It is a very small positive number, for example, between 0.001 and 0.05, used to prevent the denominator from being zero. and The value ranges approximately from -1 to 1, its sign clearly indicates the direction of movement, and its absolute value remains stable around 1, facilitating subsequent processing. This sub-step yields the current movement state of the fence, i.e., the direction it is moving in and its speed.

[0093] S2.3: Predict the spatial displacement and boundary position of the enclosure in the next signal cycle based on the current rate of change.

[0094] Having obtained the current movement speed of the enclosure , After the direction markings, predict the location of the construction site in the next complete signal control cycle. This predicts the spatial displacement that will occur within the system. This prediction allows signal timing schemes to be generated based on potential future spatial occupancy, thus achieving true feedforward control and avoiding lag.

[0095] The basic assumption of the prediction is that the movement trend of the enclosure is persistent in the short term. Therefore, the rate of change calculated at the current moment is regarded as the rate of change in the next cycle. The average velocity within the range is used to calculate the predicted displacement. Longitudinal predicted displacement. and lateral predicted displacement The calculation formula is as follows: ; ; This indicates the distance the construction site fence is expected to move longitudinally along the road in the next cycle, starting from the current moment. This indicates the expected expansion or contraction distance in the horizontal direction.

[0096] To directly obtain the predicted position of the enclosure boundary at the end of the next signal cycle, the predicted displacement is superimposed on the observed boundary position at the current moment. Define the predicted boundary position. and for: ; ;in, and It is the current fence boundary position obtained in sub-step S2.1. and Together, they described the expected coordinates of the equivalent boundary of the enclosure in the road coordinate system at the end of the next control cycle. Finally, all the prediction information was summarized into a short-term spatial prediction of the enclosure. It includes longitudinal predicted displacement. Lateral displacement prediction Longitudinal prediction of boundary location and the location of the lateral predicted boundary This short-term spatial prediction represents the spatial occupancy range that the enclosure will create in the next signal cycle, predicted based on the current movement trend. This prediction facilitates data transfer between steps; in actual use, each sub-prediction value will be retrieved separately.

[0097] S2.4: Encapsulate the predicted boundary location into a rectangular predicted road occupancy range and output it as the dynamic event boundary.

[0098] Sub-step S2.3 predicts the theoretical location line of the construction site enclosure boundary. However, in actual traffic engineering, the impact of construction enclosures on traffic is not a line, but an area with physical width. Therefore, it is necessary to extend the predicted boundary location line into a two-dimensional road occupancy area to more realistically reflect the occupation of road space by the enclosure and to provide direct geometric input for subsequent calculation of the available traffic length of each lane.

[0099] This sub-step will predict the boundary location. and As the regional center, the longitudinal influence of the fencing is introduced to half the width. And the horizontal influence half width This is used to define a rectangular predicted road occupancy area. Wherein, The value is usually between 0.5 meters and 5 meters. Values ​​are typically between 0.3 meters and 3 meters, determined based on the typical physical dimensions of the enclosure and the construction safety buffer zone. Therefore, the boundary interval of the occupied area in the road coordinate system is defined as follows: , , , .in, and The starting and ending points of the road occupation area in the longitudinal direction of the road are defined. and The start and end ranges in the horizontal direction are defined.

[0100] Ultimately, this rectangular road occupancy area is encapsulated into a unified description. It consists of four boundary coordinates , , , Definition. Description of the area occupied by the road. It replaces the static, historical event boundaries of traditional methods. It represents the dynamic spatial constraints that will form in the next signal cycle, predicted based on current movement trends. This fundamentally solves the signal timing lag and fixation problems caused by relying on outdated boundaries, providing accurate and forward-looking input for step S3's real-time reconstruction of road traffic space. From a traffic control perspective, this prediction essentially defines the location and geometry of dynamic physical bottlenecks in the next control cycle, serving as a key input for refined lane-level capacity analysis and signal resource pre-allocation.

[0101] In the technical solution of this disclosure embodiment, based on a continuous spatial function, the movement trajectory of the enclosure is discretized and windowed to quantitatively extract its real-time change rate and direction in the longitudinal and lateral directions. Based on this rate, its displacement and road occupation range in the next signal cycle are predicted. This step abandons the traditional approach of relying on fixed event boundaries of the previous cycle. By actively generating forward-looking spatial evolution predictions, the predicted boundary position lines are encapsulated into specific rectangular road occupation area descriptions, thereby providing accurate and future-oriented dynamic input for subsequent reconstruction of the passage space and avoiding the solidification of control schemes due to information lag from the source.

[0102] S3: Spatial overlap calculation is performed between the predicted lane occupancy range and the lane-level road network model, and the effective passable length and reduced capacity of each lane are reconstructed in real time to generate the total effective capacity constraint for each approach direction.

[0103] This step is the core converter connecting spatial physical changes with traffic control parameters. Its theoretical basis is the fundamental principle of traffic engineering, namely that lane capacity is a direct function of lane geometry, width, length, and alignment. Construction encroachment leading to a reduction in effective lane width or lane disappearance inevitably causes a nonlinear reduction in lane capacity and may trigger lane changes and redistribution of traffic flow. After obtaining the predicted lane occupancy range of construction barriers within future signal cycles in step S2, this abstract geometric range is transformed into engineering parameters that directly guide traffic control, enabling signal timing to directly respond to changes in spatial resources rather than rigid event labels. Figure 2 shows a flowchart of the road passability space parameter reconstruction provided in this embodiment of the disclosure, as shown in Figure 2, specifically implemented through the following sub-steps.

[0104] S3.1: Perform lane-level digital instantiation of the target road, construct a lane-level road network model, and define the geometric parameters of each lane.

[0105] To conduct a refined spatial impact assessment, the target road area—including the construction impact zone and adjacent signalized intersections—needs to be digitally analyzed according to the actual traffic organization methods to create a lane-level road network model. This includes determining the number of approach directions at intersections and the number and geometric attributes of lanes in each direction.

[0106] Assume the target intersection has D entrance directions, such as east, west, south, and north. For each entrance direction d, where... This direction includes Each lane needs to be uniquely identified and parameterized. Therefore, a unified numbering rule is used. Where d represents the direction of entry, and k represents the lane number within that direction. The set of all lanes is denoted as . .

[0107] Each lane unit In the road coordinate system, it is approximated by a rectangular region. Its geometry is defined by several key parameters: the longitudinal coordinates of the lane's starting point. Longitudinal coordinates of the termination point lateral coordinates of the lane centerline and lane width Lane width This is a design parameter, typically between 2.75 meters and 3.75 meters. Based on these parameters, the left and right lateral boundaries of the lane rectangle can be calculated: ; ;in, and Representing lanes The coordinates of the left and right boundaries. Finally, the complete geometric boundary parameters for each lane are encapsulated as... It essentially represents a set. This step discretizes the continuous road space into a series of independently computable lane units, providing a precise geometric basis for subsequent quantification of the specific impact of the construction barriers on each lane.

[0108] S3.2: Calculate the predicted lane occupation area and the spatial overlap length and lane occupation ratio of each lane unit.

[0109] After lane instantiation is completed, the predicted area occupied by the construction site is quantitatively calculated. With each lane unit The spatial intersection between them. The area occupied by the road is determined by... and The defined rectangular area is described. The calculation is divided into two stages: first, lateral overlap is determined to see if the enclosure affects the lane; if lateral overlap exists, the occupied length in the longitudinal direction is further calculated.

[0110] Calculate the horizontal overlap width It indicates the area of ​​enclosure and the driveway. Width of the overlapping portion in the horizontal direction: ;if This indicates that the construction fence does not intersect with the lane laterally, and the lane is unaffected. If This indicates the presence of lateral influence, requiring the calculation of the longitudinal road occupancy length. That is, the length of the enclosure area that encroaches on the lane along the direction of vehicle travel: ; The value ranges from 0 to the total length of the lane. between.

[0111] To more generally express the severity of lane occupancy, the lane occupancy ratio is further calculated. That is, the ratio of the occupied length to the original lane length: ;in, It is a very small positive number, for example, between 0.01 and 0.1, used to avoid the denominator being zero. It is a dimensionless ratio between 0 and 1, intuitively reflecting the proportion of lane space occupied by construction barriers. This step transforms the abstract occupied area into a quantifiable space loss indicator for each lane: lane occupation length. Lane occupancy ratio This method of transforming spatial geometric intersection into an impact index is a common technique in traffic impact assessment, providing a direct basis for quantitatively evaluating the reduction effect of road construction on road capacity.

[0112] S3.3: Calculate the effective passable length and reduced capacity of each lane based on the length of the occupied lane.

[0113] Once the occupied road length is obtained, the actual remaining effective passable length of each lane under the impact of construction can be calculated. (Lane) Effective passable length Defined as the original available length of the lane minus the occupied length: ;

[0114] Its unit is meters. When When approaching the full length of the lane, A value close to 0 means that the lane is essentially unusable for traffic.

[0115] However, for signal timing, a more critical parameter is traffic capacity. Space occupancy not only reduces available length but also leads to a non-linear decrease in traffic efficiency due to lane narrowing and distorted vehicle trajectories. Therefore, it is necessary to optimize the space occupancy ratio. This is mapped to a reduction in traffic capacity. Let the lanes be... The baseline traffic capacity under conditions of no construction interference is The unit is vehicles per hour, and the specific value is usually between 600 and 2000, depending on the lane type, speed limit and traffic organization method.

[0116] A nonlinear reduction model is used to calculate the effective capacity. The calculation formula is: ;in, It is a reduction index used to characterize the non-linear reduction effect of road occupancy on traffic efficiency, and its value typically ranges from 1.0 to 3.0. Reduction Index The model's design aligns with common observations in traffic engineering regarding the accelerated decline in traffic capacity due to reduced lane width or effective usable area. It more accurately characterizes the additional traffic efficiency loss caused by narrow lanes, obstructed visibility, and forced distortion of vehicle trajectories in construction zones. The model uses a power function to characterize the nonlinear attenuation effect of lane occupancy on traffic efficiency. When the proportion of lanes occupied increases, it indicates that the rate of decrease in traffic capacity will accelerate, which better reflects the phenomenon of worsening congestion in actual engineering projects. When the proportion of lanes occupied... When, the ability remains unchanged; when As the power increases, the capability decreases in a power function manner.

[0117] To ensure that the signal timing model still has a solution under extreme lane occupancy conditions and to avoid zero capability in the calculation, a minimum capability lower limit is set. This value typically ranges from 50 to 200 vehicles per hour. The calculated capacity is then truncated. ;

[0118] This step reconstructs two core dynamic parameters for each lane: effective passable length. and effective traffic capacity These parameters directly and quantitatively reflect the real-time constraints that the dynamic evolution of construction site enclosures places on road space resources. Thus, the road accessibility parameters based on the predicted road occupancy range can be reconstructed in real time.

[0119] S3.4: Based on the reconstructed lane-level parameters, encapsulate them into a dynamic spatial parameter set and aggregate them to generate directional-level traffic capacity constraints.

[0120] The effective passable length of all lanes is obtained through reconstruction. and effective traffic capacity Next, these parameters need to be organized into a data structure that is easy for the subsequent signal timing module to directly call. First, all lane-level parameter pairs are encapsulated into a complete set of dynamic passable space parameters. It is a data structure or list where each entry corresponds to a lane. , containing its binary This ensures that the signal timing algorithm can apply differentiated constraints based on the specific conditions of each lane.

[0121] On the other hand, since the signal phase at intersections is typically controlled on a per-entry-direction basis, it is also necessary to aggregate lane-level capacity to the direction-level to generate a total effective capacity constraint for each entry-direction. For each entry-direction d, its corresponding lane set is denoted as... The total effective capacity constraint in this direction is obtained by summing the results. It is the sum of the capacities of all lanes in that direction: ;

[0122] Total effective capacity constraint The unit is also vehicles per hour. It represents the maximum traffic flow that can pass through in that direction within the predicted time window, and will be directly used to constrain the green light time allocation for that direction in the signal timing.

[0123] In the technical solution of this disclosure, the predicted lane occupancy range is spatially overlapped with the digital lane model to accurately calculate the occupied length and proportion of each lane, thereby deriving the effective passable length and reduced capacity at the lane level, and finally aggregating them into directional capacity constraints. This step transforms the abstract geometric lane occupancy range into a series of quantitative parameters that can be directly used for traffic control engineering calculations, realizing an accurate mapping from the dynamic evolution of road enclosures to the real-time supply capacity of road space resources, and providing differentiated and precise spatial constraints for signal timing.

[0124] S4: Based on the total effective passage capacity constraint of each inlet direction, continuously generate and update the signal timing scheme, and re-execute step S1 to this step before the start of each signal cycle based on the latest UAV video stream to output a timing scheme that matches the dynamic evolution of the enclosure.

[0125] This step follows the basic principle of capacity-oriented signal timing. Its core logic is to allocate green light time resources proportionally based on the real-time changing traffic capacity supply in each approach direction, thereby minimizing overall intersection delays or maximizing traffic efficiency. This overcomes the limitations of traditional methods that can only perform fixed timing or simple sensor control in construction zones. It transforms the dynamic set of passable space parameters and their directional capacity constraints into a phase green light time scheme that can be directly sent to the intersection signal controller, and designs a periodic rolling update mechanism to ensure that the timing scheme always remains consistent with the current and short-term evolution trends of the construction site, thus completely eliminating the control rigidity problem caused by information update lag at the system operation level. This is specifically achieved through the following sub-steps.

[0126] S4.1: Standardize the capacity constraints for each import direction and generate release weights.

[0127] During signal timing, different approaches should receive different proportions of green light time resources due to variations in their traffic capacity. First, the effective traffic capacity for each direction obtained in step S3 is... It is transformed into a set of comparable and assignable standardized weights.

[0128] Directly use effective passage capacity Using this as a basis for allocation may present a problem: when a certain direction's effective traffic capacity is severely reduced due to construction barriers obstructing the road. At extremely low levels, directly calculating the weights in this direction would result in an excessively small weight, potentially leading to a green light duration too short to meet the minimum requirements for safe vehicle passage, rendering the timing scheme unfeasible. To avoid this problem, a minimum controllable capability lower limit is introduced. Its value ranges from 200 to 800 vehicles per hour, which is used to ensure that each direction has the most basic feasibility for passage.

[0129] For each import direction d, where First, calculate its corrected available capacity. : ;

[0130] This operation ensures that the correction capability in each direction is no less than the lower limit of capability. This approach reflects the robustness principle in traffic signal control, preventing timing scheme failures or unacceptable vehicle delays due to a sudden drop in capacity in one direction, and ensuring the basic service functions of the control system under abnormal conditions.

[0131] Subsequently, the correction capabilities in all directions are standardized to obtain the standardized release weight for each direction. : ;

[0132] It is a dimensionless value between 0 and 1, with the sum of the weights for all directions being 1. This weight directly reflects the priority of each approach direction relative to the others under the current and predicted traffic conditions. This weight will become the sole direct basis for subsequently allocating specific green light times, ensuring that the core logic of the signal timing scheme closely follows the changes in spatial capacity caused by the evolution of construction barriers.

[0133] S4.2: Calculate the initial value of the green light duration for each signal phase based on the release weight.

[0134] Traffic signal control at intersections is typically implemented on a phase basis. A phase refers to the combination of signal displays corresponding to a group of traffic flows that simultaneously receive the right-of-way. This involves weighting traffic flow based on direction. Map it to a specific signal phase and calculate the initial value of the green light time for each phase.

[0135] Assume the signal phase set at the intersection is There are J phases in total. Each phase j serves a specific set of inlet directions. For example, a straight-ahead phase may simultaneously serve two opposing north-south approach directions. This phase-to-direction correspondence is predefined by the intersection's signal timing scheme.

[0136] Calculate the overall weight for each phase j Its value is the sum of the weights of all directions served by that phase: ;

[0137] Next, the overall weights of all phases are normalized to obtain the release ratio for each phase. : ; It is also a value between 0 and 1, and the sum of the release ratios of all phases is 1. It represents the share of time that phase j should occupy within a signal cycle.

[0138] Let the current signal period length be... Its value typically ranges from 40 to 180 seconds and can be adjusted according to overall traffic demand. Within a cycle, in addition to the effective green light time, it also includes the yellow light time, all-red time, and other total lost time L, which ranges from approximately 6 to 25 seconds. Therefore, the total effective green light duration available for allocation to each phase within a cycle is... .

[0139] Based on this, the initial value of the green light duration for phase j is... It can be obtained by multiplying its release ratio by the total effective green light duration: Thus, a preliminary timing scheme based on dynamic capability weights has been obtained.

[0140] S4.3: Perform executability correction and cycle duration normalization on the initial value of the green light duration to obtain the final phase green light time.

[0141] The initial value of the green light calculated in sub-step S4.2 While derived purely from capacity ratios, in practical engineering applications, the green light time for each phase must meet executability constraints. Especially in scenarios where construction barriers are dynamically advanced, the capacity in a certain direction may suddenly drop, resulting in a green light time that is too short according to the ratio, making it impossible to guarantee the safe clearing of queuing vehicles; or the capacity in some directions may be relatively high, resulting in a green light time that is too long, causing wasted time and a decrease in overall cycle efficiency.

[0142] Therefore, a minimum green light time for the phase must be introduced. and maximum green light time The constraints are as follows. These are the general safety and efficiency principles for traffic signal control design: the minimum green light time is used to ensure the safe passage of vehicles through the intersection and to meet the minimum time requirements for pedestrians to cross the street; the maximum green light time is used to prevent excessively long cycles, balance the service opportunities of each phase, and avoid excessive saturation of individual phases. The value is usually between 5 and 15 seconds to ensure the minimum safe passage requirement; The value is usually between 25 and 90 seconds to prevent a single phase from monopolizing too much time.

[0143] Initial green light value for each phase j Perform truncation correction to obtain the truncated green light duration. : After this operation, the green light time for each phase is limited to [a certain duration]. Within the range.

[0144] However, the truncation operation changes the sum of the green light times for all phases, which may cause the total green light time to no longer equal the preset total effective green light duration. To ensure a constant cycle length, the truncated green light time needs to be proportionally normalized. First, the total green light time after truncation is calculated. : Then, scale the total time proportionally to obtain the final executable green light duration for each phase. :

[0145]

[0146] This step ensures that the green light time structure of the entire cycle is maintained while meeting the minimum safety requirements and maximum duration limits for each phase, and the resulting solution is fully implementable in engineering.

[0147] S4.4: Trigger a rolling update before the start of each signal cycle to generate and output the complete timing scheme for that cycle.

[0148] To achieve real-time linkage between signal timing and the dynamic evolution of the enclosure, and to avoid using outdated information from the previous cycle, a rolling update and triggering mechanism based on signal cycles must be established. This sub-step specifies the calculation, output, and update timing of the timing scheme.

[0149] Let k be the update index of the signal period, and let k be the start time of each period. In each new signal cycle Before starting, the system automatically triggers a complete timing calculation process. This process will re-execute all calculations from step S1 to step S4.3, but all calculations will rely on the latest current-time inputs. This means that the observational data used to predict the trend of construction site enclosures is up to... The latest video stream, used for computing power constraints, has an occupancy range based on... The latest forecast of time trends, and the resulting timing scheme, are specifically designed for the upcoming cycle. Tailor-made.

[0150] Final output timing scheme It is a complete data structure that encapsulates the length of the current period. and the set of green light durations for each phase. The plan will be implemented within a cycle. The command is immediately sent to the intersection signal controller for execution.

[0151] Through this rolling update mechanism of sensing, predicting, calculating, and distributing before the start of each cycle, the signal timing scheme can continuously and closely follow the dynamic changes of the construction site enclosure. The short-term evolution prediction of the enclosure can always be incorporated into the calculation within its upcoming effective cycle. Thus, in dynamically advancing construction scenarios, it fundamentally solves the problems of signal linkage lag and scheme rigidity caused by reliance on historical physical boundaries, achieving truly continuous, adaptive, and followable signal timing linkage control.

[0152] In the technical solution of this disclosure, standardized release weights are generated based on the dynamic traffic capacity constraints of each inlet direction, and initial green light times for each signal phase are allocated accordingly. After executability correction and cycle duration normalization, a timing scheme that can be directly issued is formed. By triggering rolling updates before the start of each signal cycle, it is ensured that the scheme is always based on the latest construction site enclosure evolution trend. This step establishes a complete, periodically rolling signal generation and update mechanism, enabling signal timing to closely follow the dynamic changes of the construction site enclosure. From the system operation level, continuous adaptive linkage control is achieved, completely solving the control lag and mismatch problems caused by relying on outdated physical boundaries.

[0153] According to embodiments of this disclosure, an electronic device is also provided, which may include a processor, a communications interface, a memory, and a communication bus, wherein the processor, the communications interface, and the memory communicate with each other via the communication bus. The processor can invoke logical instructions stored in the memory to execute the methods provided in the above embodiments.

[0154] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0155] On the other hand, this disclosure also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the methods provided in the above embodiments.

[0156] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0157] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0158] It should be understood that the above embodiments are only used to illustrate the technical solutions of this disclosure, and not to limit them; although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure.

Claims

1. A method for traffic incident detection and signal timing-based linkage control based on UAV video streams, characterized in that, The method includes: S1, constructing a continuous spatial function of the position of the construction site fence in the road space over time based on the UAV video stream; S2, analyzing the movement trend of the fence based on the continuous spatial function, and generating a short-time spatial prediction for the next signal control cycle, wherein the short-time spatial prediction includes the lane occupation range predicted based on the fence movement trend; S3, performing spatial overlap calculation with the predicted lane occupation range and lane-level road network model, reconstructing and outputting the effective passable length and reduced capacity of each lane in real time, and generating the total effective capacity constraint for each approach direction; S4, continuously generating and updating the signal timing scheme based on the total effective capacity constraint for each approach direction, and re-executing steps S1 to this step before the start of each signal cycle based on the latest UAV video stream to output a timing scheme that matches the dynamic evolution of the fence.

2. The traffic event detection and signal timing linkage control method based on UAV video stream according to claim 1, characterized in that, S1 includes: processing the UAV video stream to extract the pixel-level contour of the construction site fence; mapping the pixel-level contour from the image coordinate system to the road coordinate system to obtain a discrete set of location points of the fence in the road space; and constructing a continuous spatial function based on the discrete set of location points to characterize the continuous change of the fence boundary position over time.

3. The traffic event detection and signal timing linkage control method based on UAV video stream according to claim 1, characterized in that, S2 includes: discretizing the continuous spatial function according to the signal period to obtain a position sequence of the fence at multiple consecutive moments that is aligned with the signal period; calculating the average rate of change of the fence in the longitudinal and transverse directions of the road based on the position sequence; predicting the spatial displacement and boundary position of the fence in the next signal period according to the average rate of change, and expanding the predicted boundary position into a rectangular predicted road occupation range.

4. The traffic event detection and signal timing linkage control method based on UAV video stream according to claim 3, characterized in that, The calculation of the average rate of change of the fence in the longitudinal and transverse directions of the road includes: constructing a sliding window sample containing the position coordinates of multiple consecutive time moments based on the position sequence; calculating the position mean in the longitudinal and transverse directions within the sliding window sample; and calculating the average rate of change in the longitudinal and transverse directions based on the position mean of the current sliding window and the position mean of the previous sliding window.

5. The traffic incident detection and signal timing linkage control method based on UAV video stream according to claim 1, characterized in that, S3 includes: instantiating the target road area into multiple lanes with geometric parameters; calculating the overlap length and lane occupancy ratio of the predicted lane occupancy range with each lane; calculating the effective passable length of each lane based on the overlap length, and calculating the reduced capacity of each lane based on the lane occupancy ratio using a nonlinear reduction model; and aggregating and generating the total effective capacity constraint for each approach direction based on the reduced capacity of each lane.

6. The traffic incident detection and signal timing linkage control method based on UAV video stream according to claim 5, characterized in that, The calculation of the lane occupancy ratio includes: calculating the overlap width between the predicted lane occupancy range and the lane in the lateral direction; if the lateral overlap width is greater than zero, then further calculating the overlap length between the two in the longitudinal direction of the road; and determining the lane occupancy ratio based on the ratio of the longitudinal overlap length to the original length of the lane.

7. The traffic event detection and signal timing linkage control method based on UAV video stream according to claim 1, characterized in that, The continuously generated and updated signal timing scheme includes: calculating the standardized release weight for each direction based on the total effective passage capacity constraint for each inbound direction; allocating initial green light duration values ​​for each signal phase serving different directions according to the release weights; applying minimum and maximum green light time constraints to the initial green light duration values ​​and performing periodic duration normalization to obtain a directly executable phase green light time scheme.

8. The traffic event detection and signal timing linkage control method based on UAV video stream according to claim 7, characterized in that, The step of applying minimum and maximum green light time constraints to the initial value of the green light duration and normalizing the cycle duration includes: limiting the initial value of the green light duration for each phase to between a preset minimum and maximum green light time to obtain the truncated green light duration for each phase; calculating the sum of the truncated green light durations for all phases; and scaling the truncated green light durations for each phase proportionally so that the sum of the scaled green light durations for each phase equals the total effective green light duration within one signal cycle.

9. An electronic device, characterized in that, The electronic device includes a memory and at least one processor. The memory stores a computer program, and the processor executes the computer program to implement the traffic event detection and signal timing linkage control method based on UAV video stream as described in any one of claims 1-8.

10. A computer storage medium, characterized in that, It stores a computer program, which, when executed, implements the traffic incident detection and signal timing linkage control method based on UAV video stream according to any one of claims 1-8.

Images