Method and system for preventing car following based on rolling shutter coded structured light

Abstract

The application discloses a method and system for preventing car following based on rolling shutter coded structured light, and relates to the technical field of Internet of Things, which comprises the following steps: collecting original data of a barrier area through a LED coded light source, a rolling shutter camera and an ambient light sensor of a sensing layer; obtaining time information and three-dimensional spatial coordinates by time decoding and 3D reconstruction through the original data to construct a three-dimensional point cloud model; realizing the identification of a vehicle target and multi-target tracking by segmenting and clustering the point cloud model; obtaining a control strategy according to the results of identification and multi-target tracking; and controlling the equipment of the barrier area according to the control strategy. The application can predict the collision risk before contact by 3D position and speed, dynamically adjust the movement trajectory of the barrier to actively avoid, accurately measure the real physical distance and relative closing speed of the two vehicles, and establish a reliable car following behavior model. The active light emitting and specific coding scheme can eliminate the dependence on ambient light.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, and in particular to a method and system for preventing following vehicles based on rolling shutter encoded structured light. Background Technology

[0002] Vehicle entrances and exits (parking lots, highway toll stations, park and residential area access control, ETC lanes, etc.) are core nodes in smart transportation and smart community management. As the basic execution unit for vehicle passage control, the barrier gate's anti-following capability directly determines the security of entrance and exit management, vehicle passage efficiency, and toll management compliance. Stable and reliable anti-following technology can effectively curb behaviors such as vehicles closely following to evade tolls and illegally entering, while also avoiding accidents caused by barrier gate malfunctions. It is a key technological support for balancing entrance and exit passage efficiency and security control, and for achieving intelligent management of vehicle entrances and exits.

[0003] Currently, the main technologies for preventing vehicles from following each other and ensuring safety in the context of barrier gates include presence detection schemes based on inductive loops and infrared beams, active ranging and sensing schemes based on millimeter-wave radar and lidar, machine vision detection schemes based on binocular stereo vision, ToF cameras, and pure monocular vision, as well as mechanical and electrical passive safety protection schemes that rely on obstacle detection and rebound, torque limiting, and anti-smashing rubber strips. These schemes constitute the mainstream technology system for preventing vehicles from following each other and preventing vehicles from smashing in the industry.

[0004] These solutions cannot simultaneously meet the comprehensive requirements of low cost, high precision, strong environmental adaptability, and easy installation and maintenance. Ground loop coils and infrared beam detectors can only output binary signals indicating the presence or absence of objects, resulting in limited information dimensions, requiring road construction, and failing to prevent vehicles from following each other. Millimeter wave and lidar hardware are expensive, susceptible to rain, fog, multipath reflection, and installation obstruction, and their insufficient angular resolution makes it difficult to distinguish vehicles that are close at hand. Binocular vision and ToF cameras are complex to install and calibrate, rely on high-power supplementary lighting, consume a lot of computing power, and their accuracy drops significantly in strong light and low-reflection object scenarios. Pure monocular vision has scale uncertainty, cannot obtain absolute 3D information, and has poor environmental robustness. Mechanical and electrical passive protection can only trigger responses, cannot predict risks in advance, and has no ability to prevent vehicles from following each other. Summary of the Invention

[0005] This invention provides a method and system for preventing following vehicles based on rolling shutter encoded structured light. This addresses the problems of existing technologies where inductive loops and infrared beam detectors only output binary signals indicating the presence or absence of objects, resulting in limited information dimensions, the need for road excavation, and ineffective follow-prevention capabilities; millimeter-wave and lidar hardware is costly, susceptible to rain, fog, multipath reflection, and installation obstruction, and lacks sufficient angular resolution to distinguish closely following vehicles; binocular vision and ToF cameras are complex to install and calibrate, rely on high-power supplementary lighting, consume significant computing power, and experience a sharp drop in accuracy under strong light and low-reflectivity conditions; pure monocular vision suffers from scale uncertainty, cannot acquire absolute 3D information, and has poor environmental robustness; and mechanical and electrical passive protection only provides triggered responses, cannot predict risks in advance, and completely lacks follow-prevention capabilities.

[0006] On one hand, embodiments of the present invention provide a method for preventing following vehicles based on rolling shutter encoded structured light, including: Raw data of the barrier gate area is collected through LED coded light sources, a rolling shutter camera, and an ambient light sensor in the sensing layer. The raw data is temporally decoded and 3D reconstructed using an edge controller based on the rolling shutter line-by-line exposure timing sequence to obtain temporal information and three-dimensional spatial coordinates. The edge controller constructs a three-dimensional point cloud model using the time information and three-dimensional spatial coordinates; The edge controller segments and clusters the 3D point cloud model to achieve vehicle target identification and multi-target tracking; The cloud server's intelligent decision engine generates a control strategy based on the results of vehicle target identification and multi-target tracking; The barrier gate controller controls the equipment in the barrier gate area according to the control strategy.

[0007] In one possible implementation, the raw data of the barrier gate area is acquired through an LED coded light source in the sensing layer, a rolling shutter camera, and an ambient light sensor, including: The LED coded light source, rolling shutter camera, and ambient light sensor are activated after the triggering device detects that a vehicle has entered the barrier area. The parameters of the LED coded light source and the rolling shutter camera are adjusted using the ambient light sensor. The LED coded light source emits a high-frequency time-coded light signal to the vehicle. The rolling shutter camera captures coded light images reflected from the vehicle by exposing each line sequentially according to a preset line cycle. The original data consists of samples taken from different rows of the same frame of the encoded optical image at different times corresponding to the high-frequency time-coded optical signal.

[0008] In one possible implementation, obtaining temporal information and three-dimensional spatial coordinates by performing temporal decoding and 3D reconstruction on the raw data through an edge controller includes: Standard data is obtained by performing image preprocessing on the original data on the edge controller, including denoising, distortion correction, and brightness normalization. The standard data is calculated by time decoding to obtain relative time data with encoded information; Unambiguous absolute time parameters are obtained by correcting the coded relative time data through phase expansion; The three-dimensional spatial coordinates are calculated using the unambiguous absolute time parameters, LED coded light source, and camera geometric calibration parameters.

[0009] In one possible implementation, the edge controller constructs a 3D point cloud model using the temporal information and 3D spatial coordinates, including: The three-dimensional point cloud model is obtained by performing motion compensation and temporal filtering on the three-dimensional spatial coordinates.

[0010] In one possible implementation, the edge controller segments and clusters the 3D point cloud model to achieve vehicle target identification and multi-target tracking, including: Ground segmentation and Euclidean distance-based clustering are performed on the 3D point cloud model to obtain aggregated candidate targets; The target type of the candidate target is obtained through target identification; The vehicle target in the target type is continuously tracked using a multi-hypothesis tracking algorithm; The multi-target tracking is accomplished by continuously tracking and predicting the future position of the vehicle target.

[0011] In one possible implementation, the target type is a vehicle spatial feature; the target type includes leading edge position, trailing edge position, body box, vehicle center, and vehicle length.

[0012] In one possible implementation, the control strategy includes a normal passage control strategy, a following interception control strategy, an anti-collision protection control strategy, and a convoy passage control strategy. The normal passage control strategy is to identify a single authorized vehicle and generate a control command for the barrier gate to lower normally when the vehicle maintains a safe distance from the vehicles in front and behind. The following vehicle interception control strategy is to detect when two vehicles are close together and the distance between them is less than a safety threshold, determine it as following vehicle behavior, and generate a control command for the barrier gate to quickly lower and intercept the vehicle. The anti-collision protection control strategy is to detect that the vehicle is still in the area below the barrier gate and there is a risk of collision with the barrier gate arm, and generate a protection command to pause the lowering or raising of the barrier gate arm. The convoy passage control strategy identifies multiple authorized vehicles passing through consecutively while maintaining a safe distance between them, and generates an efficient passage command to keep the gate open.

[0013] In one possible implementation, the safety threshold is dynamically set based on the current distance, relative speed or approach trend, occupancy status of the area below the gate arm, and predicted occupancy status. The anti-smashing protection control strategy is based on the spatial relationship between the vehicle's three-dimensional position and the gate arm's movement area.

[0014] On the other hand, embodiments of the present invention provide a vehicle-following prevention system based on rolling shutter encoded structured light, including: The data acquisition module is used to collect raw data of the barrier gate area through the LED coded light source, rolling shutter camera and ambient light sensor of the sensing layer; The data processing module is used to perform time decoding and 3D reconstruction on the raw data based on the rolling shutter line-by-line exposure timing to obtain time information and three-dimensional spatial coordinates through the edge controller; The model building module is used by the edge controller to build a three-dimensional point cloud model using the time information and three-dimensional spatial coordinates; The target recognition module is used by the edge controller to segment and cluster the 3D point cloud model to achieve vehicle target recognition and multi-target tracking; The intelligent decision-making module is used by the intelligent decision-making engine of the cloud server to obtain control strategies based on the identification of the vehicle target and the results of multi-target tracking; The local control module is used by the barrier gate controller to control the equipment in the barrier gate area according to the control strategy.

[0015] The anti-following method and system based on rolling shutter encoded structured light in this invention has the following advantages: (1) By measuring the 3D position and speed of all objects under the barrier gate in real time and with high precision, the risk of collision can be predicted before the barrier gate arm comes into contact with the vehicle, and the movement trajectory of the barrier gate can be dynamically adjusted to actively avoid collision, thus achieving zero-contact anti-smashing.

[0016] (2) Accurately measure the actual physical distance and relative approach speed between the two vehicles to establish a reliable following behavior model. Once following is determined, the system can decisively and safely execute the barrier drop at the gap between the vehicles to effectively prevent toll evasion and intrusion.

[0017] (3) By adopting an active light emission and specific coding scheme, the dependence on ambient light is fundamentally eliminated, and it can work stably regardless of day or night, or whether it is in front of or behind the light. In addition, no road construction is required, which greatly simplifies installation and maintenance.

[0018] (4) By utilizing a rolling shutter camera and an coded light source, the dynamic 3D sensing function, which originally required expensive hardware, is achieved through spatiotemporal coding and decoding algorithms, achieving the performance of traditional expensive solutions at extremely low hardware cost. By using the line exposure time as the temporal coding sampling dimension, the high-speed modulation information of the LED coded light source is unfolded within a single frame image, obtaining stable temporal parameters and 3D measurement results without relying on expensive ToF sensors or mechanical scanning mechanisms. Attached Figure Description To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart of a method for preventing following other vehicles based on rolling shutter encoded structured light provided in this application embodiment; Figure 2 A schematic diagram of the anti-following system based on rolling shutter coded structured light provided in an embodiment of this application; Figure 3 A schematic diagram illustrating the rolling shutter line-by-line exposure principle of the anti-following method based on rolling shutter encoded structured light provided in this application embodiment. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Figure 1 This is a flowchart illustrating the anti-following method and system based on rolling shutter encoded structured light provided in an embodiment of the present invention; the embodiment of the present invention provides an anti-following method and system based on rolling shutter encoded structured light, including: Raw data of the barrier gate area is collected through LED coded light sources, a rolling shutter camera, and an ambient light sensor in the sensing layer. The raw data is temporally decoded and 3D reconstructed using an edge controller based on the rolling shutter line-by-line exposure timing sequence to obtain temporal information and three-dimensional spatial coordinates. The edge controller constructs a three-dimensional point cloud model using the time information and three-dimensional spatial coordinates; The edge controller segments and clusters the 3D point cloud model to achieve vehicle target identification and multi-target tracking; The cloud server's intelligent decision engine generates a control strategy based on the results of vehicle target identification and multi-target tracking; The barrier gate controller controls the equipment in the barrier gate area according to the control strategy.

[0022] Raw data of the barrier gate area is collected through LED coded light sources, a rolling shutter camera, and an ambient light sensor in the sensing layer, including: The LED coded light source, rolling shutter camera, and ambient light sensor are activated after the triggering device detects that a vehicle has entered the barrier area. The parameters of the LED coded light source and the rolling shutter camera are adjusted using the ambient light sensor. The LED coded light source emits a high-frequency time-coded light signal to the vehicle. The rolling shutter camera captures coded light images reflected from the vehicle by exposing each line sequentially according to a preset line cycle. The original data consists of samples taken from different rows of the same frame of the encoded optical image at different times corresponding to the high-frequency time-coded optical signal.

[0023] The time information and three-dimensional spatial coordinates obtained by performing time decoding and 3D reconstruction on the raw data through the edge controller include: Standard data is obtained by performing image preprocessing on the original data on the edge controller, including denoising, distortion correction, and brightness normalization. The standard data is calculated by time decoding to obtain relative time data with encoded information; Unambiguous absolute time parameters are obtained by correcting the coded relative time data through phase expansion; The three-dimensional spatial coordinates are calculated using the unambiguous absolute time parameters, LED coded light source, and camera geometric calibration parameters.

[0024] The edge controller constructs a 3D point cloud model using the time information and 3D spatial coordinates, including: The three-dimensional point cloud model is obtained by performing motion compensation and temporal filtering on the three-dimensional spatial coordinates.

[0025] The edge controller performs segmentation and clustering of the 3D point cloud model to achieve vehicle target identification and multi-target tracking, including: Ground segmentation and Euclidean distance-based clustering are performed on the 3D point cloud model to obtain aggregated candidate targets; The target type of the candidate target is obtained through target identification; The vehicle target in the target type is continuously tracked using a multi-hypothesis tracking algorithm; The multi-target tracking is accomplished by continuously tracking and predicting the future position of the vehicle target.

[0026] The target type is a vehicle spatial feature; the target type includes leading edge position, trailing edge position, body box, vehicle center, and vehicle length.

[0027] The control strategies include normal passage control strategy, following and interception control strategy, anti-smashing protection control strategy and convoy passage control strategy. The normal passage control strategy is to identify a single authorized vehicle and generate a control command for the barrier gate to lower normally when the vehicle maintains a safe distance from the vehicles in front and behind. The following vehicle interception control strategy is to detect when two vehicles are close together and the distance between them is less than a safety threshold, determine it as following vehicle behavior, and generate a control command for the barrier gate to quickly lower and intercept the vehicle. The anti-collision protection control strategy is to detect that the vehicle is still in the area below the barrier gate and there is a risk of collision with the barrier gate arm, and generate a protection command to pause the lowering or raising of the barrier gate arm. The convoy passage control strategy identifies multiple authorized vehicles passing through consecutively while maintaining a safe distance between them, and generates an efficient passage command to keep the gate open.

[0028] The safety threshold is dynamically set based on the current distance, relative speed or approach trend, occupancy status of the area below the gate arm, and predicted occupancy status. The anti-smashing protection control strategy is based on the spatial relationship between the vehicle's three-dimensional position and the gate arm's movement area.

[0029] For example, raw data of the barrier gate area is acquired through an LED coded light source, a rolling shutter camera, and an ambient light sensor in the sensing layer. The LED coded light source emits a high-frequency time-coded light pattern using a specific wavelength LED array to emit modulated structured light patterns. The encoding method uses a binary pseudo-random sequence, which has excellent autocorrelation characteristics, facilitating decoding and recognition. The rolling shutter camera is a CMOS camera using a rolling shutter sensor to capture time-coded light pattern images modulated by objects. The camera's line-by-line exposure characteristic is the core utilization point of this invention. The ambient light sensor detects the ambient light intensity and dynamically adjusts the emission power of the LED light source to ensure clear coded images are obtained under different lighting conditions. The trigger module detects vehicles entering the detection area and triggers the system to start the coded light source and image acquisition.

[0030] The specific process involves an LED coded light source emitting a high-frequency time-coded light signal (such as binary code). The light signal is reflected after hitting the vehicle surface. A rolling shutter camera exposes the image line by line, sampling each line at different times. By utilizing the inter-line exposure delay characteristics, the time information is decoded from a single frame image. Through spatiotemporal correlation calculations, pixel-level 3D position and velocity information is reconstructed.

[0031] In this system, the LED light source emits time-coded light signals. The encoding uses a pseudo-random binary sequence (PRBS), which exhibits good autocorrelation characteristics. The encoding frequency is much higher than the camera's frame rate, resulting in different encoding states corresponding to different lines of exposure time within a single frame. Rolling shutter cameras expose line by line, with a fixed exposure time difference (typically tens of microseconds) between adjacent lines. This characteristic is often considered a source of motion artifacts, but this application utilizes this deficiency as a means of ultra-high-speed time sampling. By precisely controlling the synchronization of the encoded signal and the exposure timing, each line (e.g., ...) can be made to have different encoding states. Figure 3 As shown, rows 1-8 of the image record the encoded state at a specific moment. Since the inter-line delay of the rolling shutter is known, the encoded sequence corresponding to each pixel can be reconstructed by analyzing the brightness variation pattern of each row in the image. By performing correlation operations between the reconstructed sequence and the known emission code, the precise timestamp corresponding to the pixel can be determined. Combining the geometric relationship between the light source and the camera, and using the principle of triangulation, the three-dimensional spatial coordinates corresponding to each pixel can be calculated.

[0032] In this application, the rolling shutter is not eliminated as an imaging error, but rather used as a time-coded sampler. The exposure center time of the y-th row of the camera is determined by... Determine and calculate the exposure center time of a pixel or pixel block. ;in, It is the frame trigger moment; It is a row cycle; It refers to the exposure time; These are pixel row numbers. When the LED coded light source follows a time-coded sequence... During high-frequency modulation, the reflected light from the vehicle surface is recorded on different image lines. Different time segments. Therefore, the same frame image forms an encoded strip along the row direction generated by the rolling shutter row timing. The edge controller recovers the relative time offset, absolute time parameters and the corresponding structured light emission state through this encoded strip, thereby completing the 3D reconstruction.

[0033] The processing layer, implemented by the edge controller, is responsible for data processing and decision-making. It includes: a spatiotemporal decoding algorithm, the core module, which utilizes the inter-line exposure delay characteristic of the rolling shutter to decode temporal information from single or multiple frames of images; a 3D reconstruction engine, which converts the decoded temporal information into pixel-level 3D spatial coordinates based on triangulation principles; vehicle detection and tracking, which segments and clusters 3D point clouds to identify vehicle targets and establish tracking; an intelligent decision-making module, which analyzes vehicle status and scene to determine the current situation and output control decisions; and a trajectory planning algorithm, which generates a barrier gate movement trajectory that meets safety constraints.

[0034] The process involves several key steps: Temporal decoding analyzes the brightness variation patterns of each line of the image, identifies the corresponding encoded sequence through correlation operations, and determines the exposure time of each line. Phase unwrapping addresses the periodic ambiguity of the encoded data by determining the absolute phase through multi-frame fusion or spatial continuity constraints. Based on known light source-camera geometric parameters, the temporal information is converted into three-dimensional spatial coordinates, leading to point cloud generation and outputting pixel-level 3D point cloud data containing the spatial coordinates and confidence level of each point. Motion compensation addresses the issue of vehicle position changes during single-frame image acquisition, which can cause deviations in 3D reconstruction. This invention employs multi-frame fusion and motion compensation techniques to track vehicle motion across consecutive frames, compensating for and correcting 3D measurement results to improve the accuracy of dynamic measurements. Temporal filtering utilizes techniques such as Kalman filtering to fuse multi-frame measurement results, smoothing noise while maintaining effective tracking of fast-moving objects.

[0035] The relative time data with encoded information is obtained by calculating the standard data through time decoding, which is the frame trigger time of the rolling shutter camera. , row cycle Exposure time and pixel row number Calculate the exposure center time of a pixel or pixel block The formula is as follows: ; Normalized brightness sequences of the same column or local window in standard data Known time-coded sequences of LED coded light sources Perform sliding correlation, as shown in the following equation: ; Take The largest As relative time offset and output relative phase and related peak confidence levels The relative time data with encoded information is obtained; wherein, It is the time offset to be determined. This refers to the row numbers of all pixels within the same column or a partial window. Summation; It is the first Row weighting coefficients; It is a relative time offset; It is the signal period of the LED coded light source, also known as the coding period.

[0036] By correcting the coded relative time data through phase expansion, unambiguous absolute time parameters are obtained, based on the relative phase. As input, combined with the encoding period Multi-frame historical prediction Spatial neighborhood continuity constraints and two sets of coprime periodic codes 、 Determine the integer period number m such that the energy Minimum; among which, It is the objective function; middle, It refers to the temporal state within the neighborhood; This is the set of spatial neighborhood nodes, representing the set of neighboring pixels surrounding the current pixel; It is the first in the neighborhood The relative time offset of each time state; It is the first in the neighborhood Integer period number of each time state; As an unambiguous absolute time parameter, low-confidence points are eliminated or interpolated based on neighborhood consistency. The three-dimensional spatial coordinates are calculated using the unambiguous absolute time parameters, the LED coded light source, and the camera geometric calibration parameters, including: based on the correspondence between the calibrated absolute time parameters and the structured light emission state. Determine the light plane or emission direction corresponding to the pixel, where Indicates the encoded light source in the first Light plane at a given time state ;in, It is the first The unit normal vector of a plane. It is a row vector; It is the first The offset parameters of each plane; the camera ray is calculated based on the camera intrinsic parameter K and pixel coordinates (u,v). ;make To obtain three-dimensional spatial coordinates = rWhen using multi-LED arrays or dot matrix structured light, the midpoint between the closest point of the camera ray and the corresponding emitted ray is used as the reference point. and output =(x,y,z) and confidence level .

[0037] The 3D point cloud model is obtained by performing motion compensation and temporal filtering on the 3D spatial coordinates, which involves setting each 3D point to its exposure center time. Confidence level Binding to the frame number; speed estimated from the vehicle tracking state. and acceleration Point Unified compensation to reference time The motion compensation formula is shown below: ； The point sets after compensation for consecutive frames are fused according to voxel grids or spatial neighborhoods, and the position and confidence of the points are updated by Kalman filtering or exponential weighted filtering. Outliers are removed to form a three-dimensional point cloud model containing three-dimensional coordinates, timestamps, velocities and confidence.

[0038] Point cloud segmentation involves performing ground segmentation and clustering analysis on the 3D point cloud obtained from spatiotemporal decoding to identify independent vehicle targets. A clustering algorithm based on Euclidean distance is used to aggregate spatially continuous point clouds into candidate targets. Vehicle recognition involves extracting features from the clustered point cloud, including size, shape, and height, and matching them with vehicle models to distinguish different target types such as vehicles, pedestrians, and non-motorized vehicles. Then, multiple hypothesis tracking (MHT) or joint probabilistic data association (JPDA) algorithms are used to continuously track multiple detected targets, establish their motion trajectories, predict their future positions, and provide a basis for decision-making.

[0039] The execution layer includes a barrier gate controller, a servo motor drive, a barrier gate arm, and an audible and visual alarm. The barrier gate controller receives control commands from the processing layer and drives the barrier gate to perform corresponding actions. The servo motor drive precisely controls the speed and position of the barrier gate arm. The barrier gate arm is the actuator used to perform opening or closing actions. The audible and visual alarm is used to issue warnings in abnormal situations.

[0040] Figure 2 This is a schematic diagram of a vehicle-following prevention system based on rolling shutter coded structured light, provided in an embodiment of the present invention. The embodiment of the present invention provides a vehicle-following prevention system based on rolling shutter coded structured light, comprising: The data acquisition module is used to collect raw data of the barrier gate area through the LED coded light source, rolling shutter camera and ambient light sensor of the sensing layer; The data processing module is used to perform time decoding and 3D reconstruction on the raw data based on the rolling shutter line-by-line exposure timing to obtain time information and three-dimensional spatial coordinates through the edge controller; The model building module is used by the edge controller to build a three-dimensional point cloud model using the time information and three-dimensional spatial coordinates; The target recognition module is used by the edge controller to segment and cluster the 3D point cloud model to achieve vehicle target recognition and multi-target tracking; The intelligent decision-making module is used by the intelligent decision-making engine of the cloud server to obtain control strategies based on the identification of the vehicle target and the results of multi-target tracking; The local control module is used by the barrier gate controller to control the equipment in the barrier gate area according to the control strategy.

[0041] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0042] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for preventing car following based on rolling shutter encoded structured light, characterized in that, include: Raw data of the barrier gate area is collected through LED coded light sources, a rolling shutter camera, and an ambient light sensor in the sensing layer. The raw data is temporally decoded and 3D reconstructed using an edge controller based on the rolling shutter line-by-line exposure timing sequence to obtain temporal information and three-dimensional spatial coordinates. Standard data is obtained by performing image preprocessing on the original data on the edge controller, including denoising, distortion correction, and brightness normalization. The relative time data with encoded information obtained by time decoding of the standard data includes: Frame triggering time of a rolling shutter camera , row cycle Exposure time and pixel row number Calculate the exposure center time of the pixel block The exposure center time The solution formula is as follows: ； Normalized luminance sequence of the same column in the standard data Known time-coded sequences of LED coded light sources The sliding correlation calculation is performed as shown in the following formula: ； Take The largest As relative time offset ; Output relative phase and related peak confidence The relative time data with encoded information is obtained; in, It is the time offset to be determined. This refers to the row numbers of all pixels within the same column or a partial window. Summation; It is the first Row weighting coefficients; It is a relative time offset; It is the signal period of the LED coded light source, i.e., the coding period; Unambiguous absolute time parameters are obtained by correcting the coded relative time data through phase expansion, including: With the relative phase As input, combined with the encoding period Multi-frame historical prediction Spatial neighborhood continuity constraints and two sets of coprime periodic codes , Determine the integer period number m such that the energy The minimum is shown in the following formula: ； in, It is the objective function; middle, It refers to the temporal state within the neighborhood; This is the set of spatial neighborhood nodes, representing the set of neighboring pixels surrounding the current pixel; It is the first in the neighborhood The relative time offset of each time state; It is the first in the neighborhood Integer period number of each time state; Will As an unambiguous absolute time parameter, points with low confidence are eliminated or interpolated based on neighborhood consistency. The three-dimensional spatial coordinates are calculated using the unambiguous absolute time parameters, LED coded light source, and camera geometric calibration parameters, including: Based on the correspondence between the absolute time parameters and the structured light emission state Determine the light plane or emission direction corresponding to this pixel; make ,in, It is the first The unit normal vector of a plane. It is a row vector; It is the first The bias parameters of each plane; According to camera internal parameters K and pixel coordinates ( u,v The camera ray is calculated as follows: ； make To obtain three-dimensional spatial coordinates = r; When using multi-LED arrays or dot matrix structured light, the midpoint between the nearest point of the camera ray and the corresponding emitted ray is used as... and output =(x,y,z) and confidence level ; The edge controller constructs a three-dimensional point cloud model using the time information and three-dimensional spatial coordinates; The edge controller segments and clusters the 3D point cloud model to achieve vehicle target identification and multi-target tracking; The cloud server's intelligent decision engine generates a control strategy based on the results of vehicle target identification and multi-target tracking; The barrier gate controller controls the equipment in the barrier gate area according to the control strategy.

2. The anti-following method based on rolling shutter encoded structured light according to claim 1, characterized in that, Raw data of the barrier gate area is collected through LED coded light sources, a rolling shutter camera, and an ambient light sensor in the sensing layer, including: The LED coded light source, rolling shutter camera, and ambient light sensor are activated after the triggering device detects that a vehicle has entered the barrier area. The parameters of the LED coded light source and the rolling shutter camera are adjusted using the ambient light sensor. The LED coded light source emits a high-frequency time-coded light signal to the vehicle. The rolling shutter camera captures coded light images reflected from the vehicle by exposing each line sequentially according to a preset line cycle. The original data consists of samples taken from different rows of the same frame of the encoded optical image at different times corresponding to the high-frequency time-coded optical signal.

3. The anti-following method based on rolling shutter encoded structured light according to claim 1, characterized in that, The edge controller constructs a 3D point cloud model using the time information and 3D spatial coordinates, including: The three-dimensional point cloud model is obtained by performing motion compensation and temporal filtering on the three-dimensional spatial coordinates.

4. The anti-following method based on rolling shutter encoded structured light according to claim 2, characterized in that, The edge controller performs segmentation and clustering of the 3D point cloud model to achieve vehicle target identification and multi-target tracking, including: Ground segmentation and Euclidean distance-based clustering are performed on the 3D point cloud model to obtain aggregated candidate targets; The target type of the candidate target is obtained through target identification; The vehicle target in the target type is continuously tracked using a multi-hypothesis tracking algorithm; The multi-target tracking is accomplished by continuously tracking and predicting the future position of the vehicle target.

5. The anti-following method based on rolling shutter encoded structured light according to claim 4, characterized in that, The target type is a vehicle spatial feature; the target type includes leading edge position, trailing edge position, body box, vehicle center, and vehicle length.

6. The anti-following method based on rolling shutter encoded structured light according to claim 1, characterized in that, The control strategies include normal passage control strategy, following and interception control strategy, anti-smashing protection control strategy and convoy passage control strategy. The normal passage control strategy is to identify a single authorized vehicle and generate a control command for the barrier gate to lower normally when the vehicle maintains a safe distance from the vehicles in front and behind. The following vehicle interception control strategy is to detect when two vehicles are close together and the distance between them is less than a safety threshold, determine it as following vehicle behavior, and generate a control command for the barrier gate to quickly lower and intercept the vehicle. The anti-collision protection control strategy is to detect that the vehicle is still in the area below the barrier gate and there is a risk of collision with the barrier gate arm, and generate a protection command to pause the lowering or raising of the barrier gate arm. The convoy passage control strategy identifies multiple authorized vehicles passing through consecutively while maintaining a safe distance between them, and generates an efficient passage command to keep the gate open.

7. The anti-following method based on rolling shutter encoded structured light according to claim 6, characterized in that, The safety threshold is dynamically set based on the current distance, relative speed or approach trend, occupancy status of the area below the gate arm, and predicted occupancy status. The anti-smashing protection control strategy is based on the spatial relationship between the vehicle's three-dimensional position and the gate arm's movement area.

8. A vehicle-following prevention system based on rolling shutter encoded structured light, characterized in that, include: The data acquisition module is used to collect raw data of the barrier gate area through the LED coded light source, rolling shutter camera and ambient light sensor of the sensing layer; The data processing module is used to perform time decoding and 3D reconstruction on the raw data based on the rolling shutter line-by-line exposure timing to obtain time information and three-dimensional spatial coordinates through the edge controller; Standard data is obtained by performing image preprocessing on the original data on the edge controller, including denoising, distortion correction, and brightness normalization. The relative time data with encoded information obtained by time decoding of the standard data includes: Frame triggering time of a rolling shutter camera , row cycle Exposure time and pixel row number Calculate the exposure center time of the pixel block The exposure center time The solution formula is as follows: ； Normalized luminance sequence of the same column in the standard data Known time-coded sequences of LED coded light sources The sliding correlation calculation is performed as shown in the following formula: ； Take The largest As relative time offset ; Output relative phase and related peak confidence The relative time data with encoded information is obtained; in, It is the time offset to be determined. This refers to the row numbers of all pixels within the same column or a partial window. Summation; It is the first Row weighting coefficients; It is a relative time offset; It is the signal period of the LED coded light source, i.e., the coding period; Unambiguous absolute time parameters are obtained by correcting the coded relative time data through phase expansion, including: With the relative phase As input, combined with the encoding period Multi-frame historical prediction Spatial neighborhood continuity constraints and two sets of coprime periodic codes , Determine the integer period number m such that the energy The minimum is shown in the following formula: ； in, It is the objective function; middle, It refers to the temporal state within the neighborhood; This is the set of spatial neighborhood nodes, representing the set of neighboring pixels surrounding the current pixel; It is the first in the neighborhood The relative time offset of each time state; It is the first in the neighborhood Integer period number of each time state; Will As an unambiguous absolute time parameter, points with low confidence are eliminated or interpolated based on neighborhood consistency. The three-dimensional spatial coordinates are calculated using the unambiguous absolute time parameters, LED coded light source, and camera geometric calibration parameters, including: Based on the correspondence between the absolute time parameters and the structured light emission state Determine the light plane or emission direction corresponding to this pixel; make ,in, It is the first The unit normal vector of a plane. It is a row vector; It is the first The bias parameters of each plane; According to camera internal parameters K and pixel coordinates ( u,v The camera ray is calculated as follows: ； make To obtain three-dimensional spatial coordinates = r; When using multi-LED arrays or dot matrix structured light, the midpoint between the nearest point of the camera ray and the corresponding emitted ray is used as... and output =(x,y,z) and confidence level ; The model building module is used by the edge controller to build a three-dimensional point cloud model using the time information and three-dimensional spatial coordinates; The target recognition module is used by the edge controller to segment and cluster the 3D point cloud model to achieve vehicle target recognition and multi-target tracking; The intelligent decision-making module is used by the intelligent decision-making engine of the cloud server to obtain control strategies based on the identification of the vehicle target and the results of multi-target tracking; The local control module is used by the barrier gate controller to control the equipment in the barrier gate area according to the control strategy.

Images