Smart parking lot safety management system and method

Abstract

The invention relates to the field of smart parking lots, and discloses a smart parking lot safety management system and method.The system comprises a gait recognition module, a track prediction module, an intelligent lamplight projection module, a warning lamplight control module, an environment sensing module and a central control unit, and the gait recognition module collects pedestrian gait data and analyzes the walking direction; the trajectory prediction module predicts a pedestrian future path based on the gait features and historical trajectories; the intelligent light projection module projects the real-time state of a pedestrian in the view field of a vehicle driver; the warning light control module adjusts a warning light signal according to the pedestrian risk level; the environment sensing module monitors illumination conditions and optimizes light display; the central control unit comprehensively processes data of all the modules and dynamically adjusts projection information and warning modes, and the system interacts with intelligent light through gait recognition, so that the pedestrian visibility is improved, the blind area risk is reduced, and the driving safety of the underground parking lot is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent parking lots, and particularly relates to an intelligent parking lot safety management system and method. Background Art

[0002] With the development of intelligent transportation and computer vision technologies, the pedestrian safety management in underground parking lots has gradually become an important direction of intelligent research. Traditional monitoring camera systems mainly rely on fixed cameras to collect video streams and perform pedestrian monitoring either manually or based on object detection algorithms. However, due to factors such as complex lighting conditions, insufficient lighting, or light source reflection in underground parking lots, traditional camera systems have certain limitations in the accuracy and response speed of pedestrian detection. For example:

[0003] Limited field of view: Cameras cannot effectively cover all visual blind spots.

[0004] Response delay: Some sensors rely on obstacle proximity detection, which may lead to untimely responses.

[0005] Lack of intelligent prediction ability: Existing technologies are difficult to predict the future movement trajectory of pedestrians in advance, so as to give early warnings before dangerous situations occur. Summary of the Invention

[0006] The first object of the present invention is to provide an intelligent parking lot safety management system, which can accurately obtain the pedestrian movement pattern through gait recognition technology, combine with a trajectory prediction module to infer the future path, and an intelligent light projection module can dynamically display the pedestrian status within the driver's field of view, and cooperate with a warning light control module to output different warning signals, enabling the driver to clearly identify the pedestrian dynamics in low-light and complex environments, achieving the advantage of significantly improving driving safety.

[0007] The above technical object of the present invention is achieved through the following technical solutions:

[0008] An intelligent parking lot safety management system, comprising:

[0009] A gait recognition module, configured to obtain pedestrian gait data in the underground parking lot, extract the gait features of pedestrians based on a preset gait recognition algorithm, and analyze the current position and walking direction of pedestrians;

[0010] A trajectory prediction module, connected to the gait recognition module, configured to predict the future travel path of pedestrians based on the gait features, walking direction, and historical trajectory of the pedestrians;

[0011] An intelligent light projection module, connected to the trajectory prediction module, configured to display the real-time status of pedestrians within the field of view of the vehicle driver in a light projection manner based on the prediction result of the trajectory prediction module;

[0012] A warning light control module, connected to the intelligent light projection module, is configured to output different forms of warning light signals around the pedestrian status projection area according to the predicted path of the pedestrian and the driving risk level;

[0013] An environment perception module, connected to the intelligent light projection module and the warning light control module, is configured to detect the environmental light conditions in the underground parking lot and adjust the brightness of the projection light and the display mode of the warning signal based on the environmental light conditions;

[0014] A central control unit, respectively connected to the gait recognition module, the trajectory prediction module, the intelligent light projection module, the warning light control module and the environment perception module, is configured to receive the pedestrian gait data of the gait recognition module, the travel path prediction result of the trajectory prediction module, and the environmental light data of the environment perception module, and comprehensively process the data based on a preset control strategy to dynamically adjust the projection information of the intelligent light projection module and the light warning mode of the warning light control module.

[0015] Further setting:

[0016] The central control unit includes:

[0017] A multi-source data fusion unit, configured to receive pedestrian detection data, trajectory prediction data and environmental light information from the gait recognition module, the trajectory prediction module and the environment perception module;

[0018] A risk assessment unit, connected to the multi-source data fusion unit, is configured to calculate the risk level of the pedestrian based on the current status and predicted trajectory of the pedestrian;

[0019] An intelligent light projection control unit, connected to the risk assessment unit and the intelligent light projection module, is configured to calculate projection information based on the pedestrian risk level and control the position and display content of the projection image, and send an instruction to the intelligent light projection module when the risk level exceeds the threshold;

[0020] A warning light control unit, connected to the risk assessment unit and the warning light control module, is configured to control the color, brightness and frequency of the warning light based on the risk level.

[0021] Further setting: The risk assessment unit uses a deep learning model based on a long short-term memory network to calculate the pedestrian risk level.

[0022] Further setting: The warning light control unit includes:

[0023] A risk perception interface, connected to the risk assessment unit, is configured to receive pedestrian risk level information and analyze its impact;

[0024] A light adaptation calculation unit, connected to the environment perception module, for calculating the basic brightness of the warning light based on the ambient light;

[0025] A warning light parameter decision-making unit, for calculating the color, brightness, and flashing frequency of the warning light based on the risk level and ambient light, and sending them to the warning light control module;

[0026] A control signal output interface, for sending an adjustment instruction to the warning light control module to dynamically control the warning light state.

[0027] Further settings:

[0028] The gait recognition module includes:

[0029] A gait data acquisition unit, for acquiring pedestrian gait data in the underground parking lot; A gait feature extraction unit, connected to the gait data acquisition unit, for extracting gait features based on the joint point information of the pedestrian; A gait trajectory modeling unit, connected to the gait feature extraction unit, for establishing a trajectory model based on the pedestrian gait features;

[0030] The gait data acquisition unit includes:

[0031] A depth camera, for collecting pedestrian video frames or point cloud data,

[0032] A target detection model, for detecting the position of pedestrians in the video frame and obtaining the pedestrian detection box,

[0033] A data preprocessing module, for denoising and normalizing the collected pedestrian data;

[0034] The gait feature extraction unit includes:

[0035] A key point detection model, for extracting key points such as the legs, knees, and hips of the pedestrian,

[0036] A gait parameter calculation module, for calculating the step length, step frequency, and swing angle based on the key point information;

[0037] The gait trajectory modeling unit includes:

[0038] A trajectory smoothing module, for smoothing the pedestrian detection position based on Kalman filtering,

[0039] A walking direction calculation module, for calculating the pedestrian direction based on past trajectory data.

[0040] Further settings: The trajectory prediction module includes:

[0041] A gait pattern analysis unit, connected to the gait recognition module, for judging the movement pattern of the pedestrian based on the gait feature data;

[0042] A trajectory prediction unit, connected to the gait pattern analysis unit, for predicting the future travel path of a pedestrian;

[0043] The gait pattern analysis unit includes:

[0044] A motion pattern classification module, which determines whether a pedestrian is in a stable gait, an unstable gait, turning, sudden stop, etc. based on step length, step frequency, and swing angle;

[0045] A direction change detection module, which calculates the pedestrian's direction change trend based on the pedestrian's historical direction data and determines whether the pedestrian has a tendency to change the travel direction;

[0046] The trajectory prediction unit includes:

[0047] A trajectory history modeling module, which establishes a trajectory database based on the pedestrian's historical trajectory data;

[0048] A trajectory prediction model, which uses a long short-term memory network for trajectory prediction.

[0049] Further provided: The intelligent light projection module includes:

[0050] A projection position calculation unit, for calculating the position of the projection area based on the prediction result of the trajectory prediction module; A projection image generation unit, connected to the projection position calculation unit, for generating a pedestrian projection image; A dynamic projection adjustment unit, connected to the projection image generation unit, for updating the projection image based on the pedestrian's trajectory;

[0051] The projection position calculation unit includes:

[0052] A trajectory mapping module, which calculates the projection point position according to the trajectory prediction coordinates;

[0053] A projection angle calculation module, which calculates the projection transformation parameters based on the installation height and projection angle of the projection device:

[0054] The projection image generation unit includes:

[0055] A pedestrian gait modeling module, which calculates the dynamic change of the image based on the gait feature parameters;

[0056] An image rendering module, which generates a projection image based on the rendering engine;

[0057] The dynamic projection adjustment unit includes:

[0058] An image update module, which adjusts the projection frame rate based on the pedestrian's speed;

[0059] A light intensity adaptive module, which adjusts the projection light intensity based on the ambient light:

[0060] Further settings: The projection image generation unit includes:

[0061] A human key point extraction module for obtaining key point data of pedestrians from the gait recognition module;

[0062] A skeleton modeling module for constructing a 3D skeleton model based on the key point data, which is composed of joint points and joints;

[0063] A gait animation generation module connected to the skeleton modeling module for generating skeleton animations according to pedestrian gait characteristics;

[0064] An image rendering module connected to the gait animation generation module for rendering the skeleton animation based on the Open Graphics Library and displaying the image through a projection device;

[0065] The projection image generation unit adopts a 3D modeling method to generate projection images through skeleton animations based on human key point data.

[0066] The second objective of the present invention is to provide an intelligent parking lot safety management method, including the following steps:

[0067] S1, gait recognition: Obtain pedestrian gait data in the underground parking lot; Extract pedestrian gait characteristics based on a preset gait recognition algorithm, and analyze the current position and walking direction of the pedestrian;

[0068] S2, trajectory prediction: Through a trajectory prediction model, based on the pedestrian's gait characteristics, walking direction, and historical trajectory, predict the future travel path of the pedestrian;

[0069] S3, intelligent light projection: According to the result of the trajectory prediction, within the visual range of the vehicle driver, display the real-time status of the pedestrian by means of light projection;

[0070] S4, warning light control: According to the future trajectory of the pedestrian and the driving risk level, output different forms of warning light signals around the pedestrian status projection area to remind the vehicle driver to pay attention;

[0071] S5, environment perception and adaptation: Collect the environmental light conditions of the underground parking lot, and adjust the brightness of the projection light and the display mode of the warning signal based on the light conditions.

[0072] In summary, the present invention has the following beneficial effects:

[0073] The present invention accurately obtains the pedestrian motion pattern through gait recognition technology, and combines a trajectory prediction module to infer the future path, thereby breaking through the limitation that traditional systems can only monitor the current state. At the same time, the intelligent light projection module can dynamically display the pedestrian state within the driver's field of vision, and cooperate with the warning light control module to output different warning signals, enabling the driver to clearly identify the pedestrian dynamics in low-light and complex environments. The environmental perception module can adaptively adjust the light signal according to the lighting conditions, enabling the system to adapt to different lighting environments and improving the warning effect. Through the above technical means, the intelligent level of pedestrian guidance and warning in the underground parking lot is significantly improved, which helps to reduce traffic safety accidents caused by difficult recognition of pedestrians in blind spots. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 is a schematic structural diagram of the system of the present invention;

[0075] Figure 2 is a schematic flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0076] The present invention will be further described in detail below with reference to the accompanying drawings.

[0077] Embodiment:

[0078] An intelligent parking lot safety management system, as Figure 1 shown, includes:

[0079] A gait recognition module, including a gait recognition camera, for obtaining pedestrian gait data in the underground parking lot, extracting pedestrian gait features based on a preset gait recognition algorithm, and analyzing the current position and walking direction of the pedestrian;

[0080] A trajectory prediction module, connected to the gait recognition module, for predicting the future travel path of the pedestrian based on the pedestrian's gait features, walking direction, and historical trajectory;

[0081] An intelligent light projection module, including an LED projection device and a high-brightness indicator light, connected to the trajectory prediction module, for displaying the real-time state of the pedestrian in the form of light projection within the field of vision of the vehicle driver based on the prediction result of the trajectory prediction module;

[0082] A warning light control module, connected to the intelligent light projection module, for outputting different forms of warning light signals around the pedestrian state projection area according to the predicted path of the pedestrian and the driving risk level;

[0083] An environmental perception module, connected to the intelligent light projection module and the warning light control module, for detecting the environmental light conditions in the underground parking lot and adjusting the brightness of the projection light and the display mode of the warning signal based on the environmental light conditions;

[0084] The central control unit is respectively connected to the gait recognition module, the trajectory prediction module, the intelligent light projection module, the warning light control module and the environment perception module, and is used to receive the pedestrian gait data of the gait recognition module, the predicted travel path result of the trajectory prediction module, and the environmental light data of the environment perception module, and comprehensively process the data based on a preset control strategy to dynamically adjust the projection information of the intelligent light projection module and the light warning method of the warning light control module.

[0085] The central control unit includes:

[0086] The multi-source data fusion unit is used to receive the pedestrian detection data, trajectory prediction data and environmental light information from the gait recognition module, the trajectory prediction module and the environment perception module;

[0087] The risk assessment unit is connected to the multi-source data fusion unit and is used to calculate the risk level of the pedestrian based on the current state and predicted trajectory of the pedestrian;

[0088] The intelligent light projection control unit is connected to the risk assessment unit and the intelligent light projection module, and is used to calculate the projection information based on the pedestrian risk level, and control the position and display content of the projection image. When the risk level exceeds the threshold, an instruction is sent to the intelligent light projection module;

[0089] The warning light control unit is connected to the risk assessment unit and the warning light control module, and is used to control the color, brightness and frequency of the warning light based on the risk level.

[0090] The central control unit is the core of the entire system, responsible for data fusion, decision-making analysis, control output, and interacting with other modules to achieve the functions of dynamic pedestrian guidance and warning;

[0091] The main functions of the central control unit include:

[0092] 1. Obtain data input:

[0093] Receive the pedestrian movement data provided by the gait recognition module.

[0094] Receive the future pedestrian path data provided by the trajectory prediction module.

[0095] Receive the light and vehicle dynamic information provided by the environment perception module.

[0096] 2. Fusion data processing:

[0097] The Kalman filtering algorithm is used for data denoising and smoothing processing, and the formula is as follows:

[0098] X t = AX t-1 + BU t + Wt

[0099] Z t = HX t + V t

[0100] Where:

[0101] X t : System state vector, including pedestrian position, speed, gait parameters, etc.

[0102] A: State transition matrix (describing the motion law).

[0103] B: Control input matrix.

[0104] U t : Control input (such as ambient light change).

[0105] W t : System noise.

[0106] Z t : Observation value (pedestrian position, trajectory measured by sensors).

[0107] H: Observation matrix.

[0108] V t : Observation noise.

[0109] Prediction update (predicting the current state):

[0110]

[0111] Measurement update (correcting the prediction):

[0112]

[0113] Where:

[0114] Final state estimate value (smoothed pedestrian position).

[0115] P t : State error covariance matrix.

[0116] K t : Kalman gain (determining the degree of correction).

[0117] Q, R: Covariance matrices of process noise and measurement noise.

[0118] Calculating the pedestrian risk level:

[0119] The central control unit calculates whether the pedestrian is in a dangerous area and decides whether to trigger warning lights or projection enhancement.

[0120] The pedestrian risk probability is calculated as follows:

[0121]

[0122] Where:

[0123] P(R|X t ): The probability that a pedestrian is at risk at the current location X t .

[0124] P(X t |R): The probability of observing the pedestrian state X t in a dangerous situation.

[0125] P(R): The prior probability, i.e., the probability that a pedestrian is in a dangerous state by default.

[0126] P(X t ): The normalization factor to ensure that the sum of probabilities is 1.

[0127] The risk level assessment is calculated as follows:

[0128]

[0129] Where:

[0130] L: The comprehensive risk level (the higher the value, the more likely a warning will be triggered).

[0131] w i : The weights of different risk factors (such as pedestrian speed, distance, and environmental lighting).

[0132] 3. Control output:

[0133] Send projection data to the intelligent lighting projection module to update the position of the pedestrian image.

[0134] Send a warning signal to the warning light control module to control the color and intensity of the light.

[0135] The risk assessment unit calculates the pedestrian risk level using a deep learning model based on a long short-term memory network.

[0136] The central control unit, as the core control module, communicates with the gait recognition module, trajectory prediction module, intelligent lighting projection module, warning light control module, and environmental perception module, processes all data, and comprehensively adjusts the output modes of light projection and warning signals according to the preset control strategy.

[0137] The gait recognition module includes:

[0138] A gait data acquisition unit for acquiring pedestrian gait data in an underground parking lot; a gait feature extraction unit connected to the gait data acquisition unit for extracting gait features based on the joint point information of pedestrians; a gait trajectory modeling unit connected to the gait feature extraction unit for establishing a trajectory model based on pedestrian gait features;

[0139] The gait data acquisition unit includes:

[0140] A depth camera for collecting pedestrian video frames or point cloud data,

[0141] A target detection model for detecting the position of pedestrians in video frames and obtaining pedestrian detection boxes,

[0142] A data preprocessing module for denoising and normalizing the collected pedestrian data;

[0143] The gait feature extraction unit includes:

[0144] A key point detection model for extracting key points such as the legs, knees, and hips of pedestrians,

[0145] A gait parameter calculation module for calculating step length, step frequency, and swing angle based on key point information;

[0146] The gait trajectory modeling unit includes:

[0147] A trajectory smoothing module for smoothing the detected pedestrian positions based on Kalman filtering,

[0148] A walking direction calculation module for calculating the pedestrian direction based on past trajectory data.

[0149] Through gait recognition technology, pedestrian gait data in an underground parking lot is collected, and gait features are extracted based on deep learning algorithms to analyze the current position and walking direction of pedestrians. Compared with traditional static image-based human detection methods, gait recognition can accurately identify pedestrians in different postures, lighting conditions, and occlusion situations, improving robustness; it can achieve different from face recognition or clothing color recognition, and gait recognition is not affected by clothing and lighting changes, enhancing the recognition stability of pedestrians. Even when pedestrians enter an occlusion area or a dimly lit area, they can be tracked through continuous gait data.

[0150] The trajectory prediction module includes:

[0151] A gait pattern analysis unit connected to the gait recognition module for judging the motion pattern of pedestrians based on gait feature data;

[0152] A trajectory prediction unit connected to the gait pattern analysis unit for predicting the future walking path of pedestrians;

[0153] The gait pattern analysis unit includes:

[0154] A motion pattern classification module that determines whether a pedestrian is in a stable gait, an unstable gait, turning, suddenly stopping, etc. based on step length, step frequency, and swing angle;

[0155] A direction change detection module that calculates the pedestrian's direction change trend based on the pedestrian's historical direction data and determines whether the pedestrian has a tendency to change the traveling direction;

[0156] The trajectory prediction unit includes:

[0157] A trajectory history modeling module that establishes a trajectory database based on the pedestrian's historical trajectory data;

[0158] A trajectory prediction model that uses a long short-term memory network for trajectory prediction.

[0159] The trajectory prediction module calculates the traveling path within a short future time based on prediction algorithms such as the long short-term memory network by analyzing the pedestrian's gait characteristics, historical trajectory, and current walking direction.

[0160] Different from the traditional method that can only detect the current position, this module can predict the possible moving direction of the pedestrian, enabling the system to take preventive measures before a danger occurs; for example, when a pedestrian suddenly crosses the lane or looks down at the mobile phone, etc., the system can identify the pedestrian's state based on the gait characteristics and send a warning signal to the vehicle driver in advance.

[0161] The intelligent light projection module includes:

[0162] A projection position calculation unit for calculating the position of the projection area based on the prediction result of the trajectory prediction module; a projection image generation unit connected to the projection position calculation unit for generating a pedestrian projection image; a dynamic projection adjustment unit connected to the projection image generation unit for updating the projection image based on the pedestrian's trajectory;

[0163] The projection position calculation unit includes:

[0164] A trajectory mapping module for calculating the projection point position according to the trajectory prediction coordinates;

[0165] A projection angle calculation module for calculating the projection transformation parameters based on the installation height and projection angle of the projection device:

[0166] The projection image generation unit includes:

[0167] A pedestrian gait modeling module for calculating the dynamic change of the image based on the gait characteristic parameters;

[0168] An image rendering module for generating a projection image based on a rendering engine;

[0169] The dynamic projection adjustment unit includes:

[0170] The image update module adjusts the projection frame rate based on the pedestrian speed;

[0171] The light adaptive module adjusts the projection light intensity based on the ambient light:

[0172] Within the field of vision of the vehicle driver, based on the pedestrian trajectory prediction result, the dynamic image of the pedestrian is displayed in real time by means of light projection, enabling the driver to observe the pedestrian state in the visual blind area.

[0173] Traditional underground parking lot camera monitoring relies on the driver to observe the display screen, while the present invention can directly project the pedestrian image on the ground or wall, enabling the driver to perceive the pedestrian information in the blind area without shifting the line of sight; it is particularly suitable for corners, ramps, and areas blocked by obstacles, effectively reducing traffic accidents caused by the limited driver's perspective.

[0174] The projection image generation unit includes:

[0175] The human key point extraction module is used to obtain the key point data of the pedestrian from the gait recognition module;

[0176] The skeleton modeling module is used to construct a 3D skeleton model based on the key point data, and the model consists of joint points and joints;

[0177] The gait animation generation module is connected to the skeleton modeling module and is used to generate a skeleton animation according to the pedestrian gait characteristics;

[0178] The image rendering module is connected to the gait animation generation module and is used to render the skeleton animation based on the Open Graphics Library and display the image through a projection device;

[0179] The projection image generation unit adopts a 3D modeling method to generate a projection image through a skeleton animation based on human key point data.

[0180] The core task of the projection image generation unit is to generate a skeleton animation projection image based on the pedestrian gait characteristics and human key point data through a 3D modeling method, so as to project a realistic dynamic pedestrian image on the ground or wall of the underground parking lot, enabling the driver to perceive the pedestrian state in the blind area. The generation process of the projection image is as follows:

[0181] (1) Extraction of human key point data

[0182] First, use OpenPose or HRNet to detect the key points of the pedestrian and extract N = 25 main human key points:

[0183] P t = {(x i , y i , z i ) | i ∈ [1, N]}

[0184] Wherein:

[0185] P t : The set of key points at time t;

[0186] x i ,y i ,z i : The coordinates of the i-th key point in three-dimensional space;

[0187] N: The total number of human key points (including head, shoulders, elbows, wrists, hips, knees, ankles, etc.);

[0188] (2) Skeleton model establishment

[0189] Construct a 3D skeleton model based on the key point data, and the skeleton is formed by connecting multiple joint points:

[0190] J t = {(P i , P j )|(i, j) ∈ E}

[0191] Wherein:

[0192] J t : The set of skeleton joints at time t;

[0193] P i , P j : The two connected key points respectively;

[0194] E: The edge set of the skeleton topology structure (defining the connection of key points, such as hip and knee, knee and ankle, etc.);

[0195] Perform 3D transformation on the skeleton joints, and adopt a rigid transformation matrix:

[0196]

[0197] Wherein:

[0198] T i，j : The transformation matrix of joints i, j;

[0199] R i，j : The rotation matrix of the joint;

[0200] S i，j : The scaling matrix of the joint;

[0201] Initial skeleton transformation matrix;

[0202] (3) Gait animation generation

[0203] The generation of the skeletal animation is based on the gait characteristics (step length, step frequency, swing angle) of pedestrians, and the inverse kinematics is used to calculate the joint movement:

[0204]

[0205] Where:

[0206] The position of the key point at time t;

[0207] The position of the key point at time t+1;

[0208] ΔP i : The movement increment of the key point, which is determined by the gait parameters;

[0209] Calculation of the gait movement increment:

[0210]

[0211] Where:

[0212] S t : Step length;

[0213] T s : Gait cycle (the time interval between two steps);

[0214] v i : The movement direction vector of the key point.

[0215] (4) Image rendering and projection

[0216] The open graphics library is used for skeletal rendering, and the image projection is based on the optical projection transformation:

[0217] I t =P t ·M view ·M proj

[0218] Where:

[0219] I t : The projection image coordinates at time t;

[0220] M view : View transformation matrix (used to adjust the viewing angle);

[0221] M proj : Projection transformation matrix (used for projection transformation to make the image adapt to the ground or wall).

[0222] The warning light control unit includes:

[0223] A risk perception interface, connected to the risk assessment unit, for receiving pedestrian risk level information and analyzing its impact;

[0224] A light adaptation calculation unit, connected to the environmental perception module, for calculating the base brightness of the warning light based on the environmental light;

[0225] A warning light parameter decision unit, for calculating the color, brightness, and flashing frequency of the warning light based on the risk level and environmental light, and sending them to the warning light control module;

[0226] A control signal output interface, for sending adjustment instructions to the warning light control module to dynamically control the warning light status.

[0227] The warning light control unit includes:

[0228] A risk perception interface, connected to the risk assessment unit, for receiving pedestrian risk level information and analyzing its impact;

[0229] A light adaptation calculation unit, connected to the environmental perception module, for calculating the base brightness of the warning light based on the environmental light;

[0230] A warning light parameter decision unit, for calculating the color, brightness, and flashing frequency of the warning light based on the risk level and environmental light, and sending them to the warning light control module;

[0231] A control signal output interface, for sending adjustment instructions to the warning light control module to dynamically control the warning light status.

[0232] The warning light control unit communicates with the risk assessment unit and the warning light control module, and is used to calculate and adjust the brightness, color, and flashing frequency of the warning light based on the pedestrian risk level, environmental light conditions, and historical warning data, so as to enhance the pedestrian warning effect;

[0233] Calculate the comprehensive output intensity I of the warning light using environmental light, risk assessment, and time history information light , specifically as follows:

[0234]

[0235] Among them:

[0236] Integral term Dynamically accumulate the brightness of the warning light based on time T;

[0237] Summation term Used to aggregate multiple historical data dimensions (such as different time periods, pedestrian risk factors);

[0238] Exponential function term Used to simulate the attenuation of pedestrian risk over time.

[0239] Gamma distribution term Γ(v)(t + τ n ) v : Used to model the changes in risk signals at different time periods.

[0240] Complex filtering function H(E ambient , L, T, σ t ): Custom non - linear interaction function of light influence and risk assessment, as follows:

[0241]

[0242] Where:

[0243] E ambient : Current ambient light intensity;

[0244] L: Pedestrian risk level;

[0245] T: Time parameter;

[0246] σ t : Light adaptability parameter;

[0247] The normalization summation term in the denominator part Is used to balance short - term risk peaks.

[0248] Warning light color adjustment

[0249] To enhance the dynamic adaptability of warning light colors, construct the following color mapping function:

[0250]

[0251] Where:

[0252] argmax selects the color mapping to ensure the optimal color is selected at different light and risk levels:

[0253] C light = Green: Low risk;

[0254] C light = Yellow: Medium risk;

[0255] C light = Red: High risk.

[0256] Warning light frequency adjustment

[0257] To adapt to the light flashing frequencies in different risk scenarios, construct the following adjustment model:

[0258]

[0259] Where:

[0260] Summation term : Combine the pedestrian risk level L and the environmental light to calculate the adjustment of the base frequency.

[0261] Integral term Calculate the dynamic change of frequency caused by risk decay based on time integration.

[0262] Exponential decay function e -μt : Used to simulate the impact of risk change on frequency.

[0263] The warning light control module outputs different forms of warning light signals around the projection area of the pedestrian state according to the trajectory prediction result and the driving risk level;

[0264] Realize multi-level warning and improve vigilance:

[0265] Low risk (constant yellow light): Remind the driver that there may be pedestrians ahead.

[0266] Medium risk (flashing yellow light): Remind the driver to slow down.

[0267] High risk (flashing red light + alarm sound): Prompt emergency braking to avoid collision.

[0268] Realize dynamic adaptation to the pedestrian state: The light signal is adjusted in real time according to the movement state of the pedestrian and the distance to the vehicle, enhancing the driver's perception ability of potential risks.

[0269] The method corresponding to this system configures the following steps, as Figure 2 shown:

[0270] S1, Gait recognition: Obtain the pedestrian gait data in the underground parking lot; Extract the gait features of the pedestrian based on the preset gait recognition algorithm, and analyze the current position and walking direction of the pedestrian;

[0271] S2, Trajectory prediction: Through the trajectory prediction model, based on the gait features, walking direction and historical trajectory of the pedestrian, predict the future movement path of the pedestrian;

[0272] S3, Intelligent light projection: According to the result of the trajectory prediction, within the visual range of the vehicle driver, display the real-time state of the pedestrian by means of light projection;

[0273] S4, Warning light control: According to the future trajectory of the pedestrian and the driving risk level, output different forms of warning light signals around the projection area of the pedestrian state to remind the vehicle driver to pay attention;

[0274] S5, Environmental perception and adaptation: Collect the environmental light conditions in the underground parking lot, and adjust the brightness of the projection light and the display mode of the warning signal based on the light conditions.

[0275] The present invention can accurately obtain the pedestrian motion pattern through gait recognition technology, and combine with the trajectory prediction module to infer the future path. The intelligent light projection module can dynamically display the pedestrian status within the driver's field of vision, and cooperate with the warning light control module to output different warning signals, enabling the driver to clearly identify the pedestrian dynamics in low-light and complex environments, achieving the advantage of significantly improving driving safety.

[0276] The above-described embodiments do not constitute a limitation on the protection scope of the technical solution. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the above embodiments shall be included within the protection scope of the technical solution.

Claims

1. A smart parking lot safety management system, characterized in that: include: The gait recognition module is used to obtain the gait data of pedestrians in the underground parking lot, extract the gait characteristics of pedestrians based on the preset gait recognition algorithm, and analyze the current position and walking direction of pedestrians; A trajectory prediction module, connected to the gait recognition module, for predicting the pedestrian's future path based on the pedestrian's gait characteristics, walking direction and historical trajectory; An intelligent light projection module, connected to the trajectory prediction module, for displaying the real-time status of pedestrians in the field of vision of the vehicle driver by light projection based on the prediction result of the trajectory prediction module; A warning light control module, connected to the intelligent light projection module, is used to output different forms of warning light signals around the pedestrian status projection area according to the pedestrian's predicted path and driving risk level; An environment perception module, connected to the intelligent light projection module and the warning light control module, for detecting the ambient light conditions of the underground parking lot and adjusting the brightness of the projection light and the display mode of the warning signal based on the ambient light conditions; The central control unit is respectively connected to the gait recognition module, the trajectory prediction module, the intelligent light projection module, the warning light control module and the environmental perception module, and is used to receive the pedestrian gait data of the gait recognition module, the travel path prediction result of the trajectory prediction module, and the ambient light data of the environmental perception module, and comprehensively process the data based on a preset control strategy to dynamically adjust the projection information of the intelligent light projection module and the light warning mode of the warning light control module.

2. According to claim 1, a smart parking lot safety management system is characterized in that: The central control unit comprises: A multi-source data fusion unit, used to receive pedestrian detection data, trajectory prediction data and ambient lighting information from a gait recognition module, a trajectory prediction module and an environmental perception module; a risk assessment unit connected to the multi-source data fusion unit and configured to calculate the risk level of the pedestrian based on the pedestrian's current state and predicted trajectory; An intelligent light projection control unit, connected to the risk assessment unit and the intelligent light projection module, is used to calculate projection information based on the pedestrian risk level, and control the position and display content of the projected image, and send instructions to the intelligent light projection module when the risk level exceeds a threshold; The warning light control unit is connected to the risk assessment unit and the warning light control module, and is used to control the color, brightness and frequency of the warning light based on the risk level.

3. According to claim 2, a smart parking lot safety management system is characterized in that: The risk assessment unit calculates the pedestrian risk level using a deep learning model based on a long short-term memory network.

4. According to claim 2, a smart parking lot safety management system is characterized in that: The warning light control unit comprises: A risk perception interface, connected to the risk assessment unit, for receiving pedestrian risk level information and analyzing its impact; A light adaptation calculation unit, connected to the environment perception module, is used to calculate the basic brightness of the warning light based on the ambient light; A warning light parameter decision unit, used to calculate the color, brightness and flashing frequency of the warning light based on the risk level and ambient light, and send it to the warning light control module; The control signal output interface is used to send adjustment instructions to the warning light control module to dynamically control the warning light status.

5. According to claim 1, the smart parking lot safety management system is characterized in that: The gait recognition module comprises: a gait data acquisition unit, used to acquire gait data of pedestrians in the underground parking lot; a gait feature extraction unit, connected to the gait data acquisition unit, used to extract gait features based on the joint point information of the pedestrians; a gait trajectory modeling unit, connected to the gait feature extraction unit, used to establish a trajectory model based on the gait features of the pedestrians; The gait data acquisition unit comprises: Depth camera, used to collect pedestrian video frames or point cloud data, The target detection model is used to detect the position of pedestrians in the video frame and obtain the pedestrian detection frame. Data preprocessing module, which performs denoising and standardization on the collected pedestrian data; The gait feature extraction unit comprises: Key point detection model, used to extract key points such as legs, knees, hips, etc. of pedestrians. A gait parameter calculation module calculates the step length, step frequency, and swing angle based on the key point information; The gait trajectory modeling unit comprises: The trajectory smoothing module smoothes the pedestrian detection position based on Kalman filtering. The walking direction calculation module calculates the pedestrian direction based on past trajectory data.

6. A smart parking lot safety management system according to claim 5, characterized in that: The trajectory prediction module comprises: A gait pattern analysis unit, connected to the gait recognition module, for determining the movement pattern of the pedestrian based on the gait feature data; a trajectory prediction unit, connected to the gait pattern analysis unit, for predicting the future travel path of the pedestrian; The gait pattern analysis unit comprises: The motion pattern classification module determines whether the pedestrian is in a stable gait, unstable gait, turning, sudden stop, etc. based on the step length, step frequency and swing angle; The direction change detection module calculates the pedestrian's direction change trend based on the pedestrian's historical direction data and determines whether the pedestrian has a tendency to change the direction of travel; The trajectory prediction unit comprises: Trajectory history modeling module, which builds a trajectory database based on pedestrian historical trajectory data; Trajectory prediction model,uses long short-term memory network for trajectory prediction.

7. The smart parking lot safety management system according to claim 1 is characterized in that: The intelligent light projection module comprises: A projection position calculation unit, used to calculate the position of the projection area based on the prediction result of the trajectory prediction module; a projection image generation unit, connected to the projection position calculation unit, used to generate a pedestrian projection image; a dynamic projection adjustment unit, connected to the projection image generation unit, used to update the projection image based on the pedestrian trajectory; The projection position calculation unit comprises: The trajectory mapping module calculates the projection point position according to the trajectory prediction coordinates; The projection angle calculation module calculates the projection transformation parameters based on the installation height of the projection device and the projection angle: The projection image generating unit comprises: Pedestrian gait modeling module, which calculates the dynamic changes of images based on gait characteristic parameters; An image rendering module generates projection images based on a rendering engine; The dynamic projection adjustment unit comprises: Image update module, which adjusts the projection frame rate based on pedestrian speed; The illumination adaptation module adjusts the projection light intensity based on the ambient lighting.

8. The intelligent parking lot safety management system according to claim 7 is characterized in that: The projection image generating unit comprises: A human key point extraction module is used to obtain the key point data of pedestrians from the gait recognition module; The skeleton modeling module is used to build a 3D skeleton model based on key point data. The model consists of joint points and joints. A gait animation generation module, connected to the skeleton modeling module, for generating skeleton animation according to pedestrian gait characteristics; An image rendering module, connected to the gait animation generation module, is used to render skeleton animation based on an open graphics library and display images through a projection device; The projection image generation unit adopts a 3D modeling method to generate a projection image through a skeleton animation based on key point data of the human body.

9. A smart parking lot safety management method, characterized in that: The following steps are involved: S1, gait recognition: obtaining pedestrian gait data in the underground parking lot; Extract the pedestrian's gait characteristics based on the preset gait recognition algorithm and analyze the pedestrian's current position and walking direction; S2, trajectory prediction: The trajectory prediction model is used to predict the pedestrian’s future path based on the pedestrian’s gait characteristics, walking direction, and historical trajectory; S3, intelligent light projection: according to the result of the trajectory prediction, the real-time status of the pedestrian is displayed in the field of vision of the vehicle driver by means of light projection; S4, warning light control: according to the pedestrian's future trajectory and driving risk level, different forms of warning light signals are output around the pedestrian status projection area to alert the vehicle driver; S5, environmental perception and adaptation: collecting the ambient lighting conditions of the underground parking lot, and adjusting the brightness of the projection light and the display mode of the warning signal based on the lighting conditions.

Images