Silicon-based Micro pixel headlamp and light and shade cut-off line dimming method and device thereof

Abstract

The invention relates to the field of silicon-based Micro pixel headlamps, and discloses a silicon-based Micro pixel headlamp and a light and shade cut-off line dimming method and device thereof.Position information and speed information of a front opposite vehicle and a front same-direction vehicle are obtained, and according to the position information and the speed information of the front opposite vehicle and the front same-direction vehicle, the light and shade cut-off line of the silicon-based Micro pixel headlamp is dimmed according to the position information and the speed information of the front opposite vehicle and the front same-direction vehicle. The method comprises the following steps: respectively tracking the positions of a front opposite vehicle and a front same-direction vehicle, dynamically controlling a pulse width modulation signal according to the position tracking information of the front opposite vehicle and the front same-direction vehicle, carrying out line-by-line scanning on light-emitting pixels of a silicon-based Micro pixel headlight, and controlling a lighting range; according to the method, the positions and the moving directions of other vehicles can be accurately determined in real time, the lighting range of the silicon-based Micro pixel headlamps can be dynamically controlled, and the stability of the positions of the light and shade cut-off lines is ensured.

Description

Technical Field

[0001] The present invention relates to the field of silicon-based MicroLED pixel headlights, and in particular to a silicon-based MicroLED pixel headlight and a method and device for dimming a light cutoff line thereof. Background Art

[0002] At present, the methods for realizing the light-dark cutoff line of automobile headlights mainly include mechanical shielding design and optical reflection design, most of which use a shading plate to block part of the light to form a light-dark cutoff line. For example, through the combination of a light source, a primary optical element, a roller shading structure and a lens, the shading member arranged on the roller has a cutoff line forming structure, and after blocking the light on the light path, a light shape with a light-dark cutoff line is formed through projection through the lens. The disadvantages of this method are that the structure is complex, the manufacturing and maintenance costs are high, and the adjustment accuracy is limited.

[0003] Another method is to use multiple unique reflective surfaces in the reflective device to reflect and focus the light from the light-emitting element to form light beams with different projection angles, which are then projected through a lens group to produce an illuminated area with a light-dark boundary. However, this design also has problems with complex structure, low integration and high cost. For example, Mercedes-Benz's adaptive headlight system uses a digital elevation model to calculate the rate of change of elevation of the path ahead in specific scenarios, such as crossing mountains, and predicts the turning point where the vehicle tilts around the Y axis, and then adjusts the pitch angle of the pixel headlight system to compensate for the change in the light beam. Although this method solves the problem of light offset to a certain extent, it relies on high-precision elevation detection, requires large changes to the vehicle, and has a complex and costly system, making it difficult to achieve the unity of high precision and low cost in a wide range of applications.

[0004] It can be seen that the existing technologies, whether mechanical occlusion, optical reflection, or adaptive adjustment solutions, all have technical problems of complex structure, low integration, and high cost, and it is difficult to achieve high-precision adjustment in specific scenarios. Summary of the invention

[0005] In view of this, the present invention provides a dimming method based on a light / dark cutoff line of a silicon-based Microled pixel headlight, which solves the technical problems of complex structure, low integration and high cost of realizing a light / dark cutoff line of a pixel headlight in the prior art.

[0006] According to a first aspect of the present invention, a method for dimming a silicon-based Microled pixel headlight at a cutoff line is provided, comprising:

[0007] Obtain the position information and speed information of the oncoming vehicle and the vehicle in the same direction ahead;

[0008] Tracking the positions of the oncoming vehicle and the same-direction vehicle respectively according to the position information and speed information of the oncoming vehicle and the same-direction vehicle;

[0009] According to the position tracking information of the oncoming vehicle and the vehicle in the same direction ahead, the pulse width modulation signal is dynamically controlled, and the luminous pixels of the silicon-based Microled pixel headlights are scanned line by line to control the lighting range, so as to control the position stability of the light and dark cut-off lines in the oncoming area and the same direction area ahead.

[0010] According to a second aspect of the present invention, there is provided a dimming device for a light / dark cutoff line based on a silicon-based Microled pixel headlight, comprising:

[0011] An acquisition module is used to acquire the position information and speed information of the oncoming vehicle and the same-direction vehicle ahead;

[0012] A detection module, used for tracking the positions of the oncoming vehicle ahead and the vehicle ahead in the same direction, respectively, according to the position information and the speed information of the oncoming vehicle ahead and the vehicle ahead in the same direction;

[0013] The control module is used to dynamically control the pulse width modulation signal according to the position tracking information of the oncoming vehicle and the vehicle in the same direction in front, and to scan the light-emitting pixels of the silicon-based Microled pixel headlights line by line to control the lighting range, so as to control the position stability of the light and dark cut-off lines of the oncoming area and the area in the same direction in front.

[0014] According to a third aspect of the present invention, a silicon-based Microled pixel headlight is provided, comprising a silicon-based Microled light source module, a controller, a driving circuit and a power module, wherein the controller is connected to the driving circuit and controls the driving circuit, the power module is used to power the silicon-based Microled light source module, and the controller executes the steps of the above-mentioned dimming method based on the bright and dark cutoff line of the silicon-based Microled pixel headlight.

[0015] By means of the above technical scheme, the present invention provides a silicon-based Microled pixel headlight and a method and device for dimming the light / dark cutoff line thereof. By acquiring the position information and speed information of the oncoming vehicle in front and the vehicle in the same direction in front, the position of the oncoming vehicle in front and the vehicle in the same direction in front are tracked respectively according to the position information and speed information of the oncoming vehicle in front and the vehicle in the same direction in front. According to the position tracking information of the oncoming vehicle in front and the vehicle in the same direction in front, the pulse width modulation signal is dynamically controlled to scan the light-emitting pixels of the silicon-based Microled pixel headlight line by line to control the lighting range, so as to control the position stability of the light / dark cutoff line of the oncoming area in front and the area in the same direction in front. The present invention can determine the position and movement direction of other vehicles in real time and accurately, can dynamically control the lighting range of the silicon-based Microled pixel headlight, ensure the stability of the light / dark cutoff line position, not only can the glare problem of oncoming vehicles be avoided and driving safety be improved, but also the lighting range can be automatically adjusted according to the road and vehicle conditions to optimize the lighting effect of the pixel headlight.

[0016] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and implement it according to the contents of the specification, and to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0018] Figure 1 A schematic diagram of the internal structure of a silicon-based Microled pixel headlight provided in an embodiment of the present invention is shown;

[0019] Figure 2 A schematic flow chart of a method for dimming a light-and-dark cutoff line based on a silicon-based Microled pixel headlight provided in the first embodiment of the present invention is shown;

[0020] Figure 3 A diagram showing an application scenario of a dimming method based on a silicon-based Microled pixel headlight according to the first embodiment of the present invention;

[0021] Figure 4 A schematic diagram of a process flow of a vehicle performing multi-target position tracking and detection method provided in Embodiment 2 of the present invention is shown;

[0022] Figure 5 A schematic diagram of a flow chart of a glare position detection method provided in Embodiment 3 of the present invention is shown;

[0023] Figure 6 A schematic diagram of a process flow of a road curvature detection method provided in a fourth embodiment of the present invention is shown;

[0024] Figure 7 A schematic structural diagram of a cut-off line dimming device based on a silicon-based Microled pixel headlight provided in an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0025] The specific implementation of the present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present invention can be combined with each other without conflict.

[0026] The embodiment of the present invention provides a method for dimming the light cutoff line of a silicon-based Microled pixel headlight, which can be applied to Figure 1 In the hardware control circuit shown, the silicon-based Microled pixel headlight control is divided into the left and right horizontal adjustment function and the height adjustment function of the pixel headlight, which actively adjusts the illumination angle of the headlight to increase the driver's field of vision, ensure driving safety, and comply with regulations and safety requirements. Figure 1 As shown, the control logic of the driving circuit is realized by the control algorithm, the light-emitting mode of the pixel matrix is ​​controlled by rows and columns, different areas are illuminated, and the position adjustment and brightness control of the light-and-dark cutoff line are realized. The driving circuit uses logic control and a pulse width modulation module to set an appropriate ramp time and pulse width for each row / column of pixels, and can control the lighting and closing of each row / column individually. The controller receives the front video stream from the camera / sensor, and the control part applies a light-and-dark cutoff line dimming method of a silicon-based Microled pixel headlight of the present invention, by obtaining the position information and speed information of the front oncoming vehicle and the front same-direction vehicle, and according to the position information and speed information of the front oncoming vehicle and the front same-direction vehicle, Information is collected, and the positions of the oncoming vehicles and the vehicles in the same direction are tracked respectively. According to the position tracking information of the oncoming vehicles and the vehicles in the same direction, the pulse width modulation signal is dynamically controlled to scan the light pixels of the silicon-based Microled pixel headlights line by line to control the lighting range, so as to control the position stability of the light and dark cutoff lines of the oncoming area and the area in the same direction. The present invention can determine the position and movement direction of other vehicles in real time and accurately, and can dynamically control the lighting range of the silicon-based Microled pixel headlights to ensure the stability of the light and dark cutoff line position, which can not only avoid the glare problem of oncoming vehicles and improve driving safety, but also automatically adjust the lighting range according to the road and vehicle conditions to optimize the lighting effect of the pixel headlights. The driving circuit controls the lighting range and light field distribution of the dot matrix, and the light field distribution depends on the processor's processing and control of the illuminated range and shadow range.

[0027] The present invention is described in detail below through specific embodiments.

[0028] Embodiment 1:

[0029] like Figure 2 As shown, a dimming method based on a light-and-dark cutoff line of a silicon-based Microled pixel headlight is provided in an embodiment of the present invention, comprising:

[0030] Step 201, obtaining position information and speed information of the oncoming vehicle and the same-direction vehicle ahead;

[0031] Step 202: Track the positions of the oncoming vehicle ahead and the vehicle ahead in the same direction, respectively, according to the position information and the speed information of the oncoming vehicle ahead and the vehicle ahead in the same direction;

[0032] Step 203: According to the position tracking information of the oncoming vehicle and the vehicle in the same direction ahead, the pulse width modulation signal is dynamically controlled to scan the light-emitting pixels of the silicon-based Microled pixel headlight line by line to control the lighting range, so as to control the position stability of the light-dark cut-off line of the oncoming area ahead and the vehicle in the same direction ahead.

[0033] Wherein, step 203 comprises:

[0034] Step 203-1, according to the real-time position of the oncoming vehicle in front, obtain the angle and distance of the oncoming vehicle in front, and calculate the lateral closing range of the silicon-based Microled pixel headlight matrix according to the angle and distance of the oncoming vehicle in front;

[0035] Step 203-2: Calculate the longitudinal lighting range of the silicon-based Microled pixel headlight matrix according to the distance to the oncoming vehicle in front.

[0036] In combination with a dimming method based on a silicon-based Microled pixel headlight according to the first embodiment of the present invention, Figure 2 In the application scenario shown, the shadow area of ​​the oncoming lane (corresponding to the closed area of ​​the silicon-based Microled pixel headlight dot matrix square) is controlled according to the real-time position of the oncoming vehicle in front, which can avoid glare to the oncoming vehicle in front, and the other areas are illuminated areas, which can illuminate the lane environment for the driver. According to the distance of the oncoming vehicle in front and the height of the pixel headlight, the position of the light and dark cut-off line is adjusted through the longitudinal lighting range of the silicon-based Microled pixel headlight dot matrix to avoid glare to the driver of the front vehicle by illuminating the rearview mirror of the front vehicle.

[0037] In order to properly adjust the sharpness of the light-dark cutoff line and reduce the driver's fatigue, the light-dark grayscale gradient adjustment is adopted in the area near the light-dark cutoff line. Accordingly, the grayscale mapping algorithm is used to control the grayscale value of each point in the light field distribution pattern to achieve the transition from light to dark within the preset range of the edge of the light-dark cutoff line. ,in, is the gray value of the i-th point in the light field distribution pattern, is the actual light intensity at the i-th point, and are the minimum and maximum light intensities in the light field area, respectively. The Microled module is a two-dimensional Micro LED array with high pixel density. Each pixel can be addressed and controlled and independently driven to emit light. The grayscale of the light and dark cutoff line is achieved by controlling the grayscale of each pixel. The method specifically includes:

[0038] For each point in the light field distribution pattern, according to its light intensity Calculate the corresponding gray value ;

[0039] Control each pixel in the MicroLED array so that its luminous intensity is consistent with the calculated grayscale value correspond.

[0040] Through the grayscale mapping algorithm, the grayscale gradient of the edge of the light and dark cutoff line is achieved, thereby reducing the sharpness and alleviating the driver's visual fatigue.

[0041] For a clearer explanation, the grayscale gradient adjustment method in the first embodiment of the present invention is illustrated by taking night driving as an example. By fine-tuning the light field distribution of the headlights, the driver can visually feel a softer light transition, especially near the light-dark cutoff line, to reduce visual fatigue caused by strong contrast. By controlling the brightness of each pixel, the light field is ensured to have a soft gradient at the light-dark cutoff line, thereby improving driving comfort.

[0042] A method for dimming the light / dark cutoff line based on a silicon-based Microled pixel headlight provided in a first embodiment of the present invention obtains position information and speed information of an oncoming vehicle in front and a vehicle in the same direction in front, and tracks the positions of the oncoming vehicle in front and the vehicle in the same direction in front respectively according to the position information and speed information of the oncoming vehicle in front and the vehicle in the same direction in front, and dynamically controls a pulse width modulation signal according to the position tracking information of the oncoming vehicle in front and the vehicle in the same direction in front, and scans the light-emitting pixels of the silicon-based Microled pixel headlight line by line to control the lighting range, so as to control the position stability of the light / dark cutoff line of the oncoming area in front and the area in the same direction in front. The present invention can determine the position and movement direction of other vehicles in real time and accurately, and can dynamically control the lighting range of the silicon-based Microled pixel headlight to ensure the stability of the light / dark cutoff line position, which can not only avoid the glare problem of oncoming vehicles and improve driving safety, but also automatically adjust the lighting range according to the road and vehicle conditions, and optimize the lighting effect of the pixel headlight.

[0043] Embodiment 2:

[0044] In order to obtain the position and speed information of oncoming vehicles and vehicles in the same direction ahead, it is necessary to track the vehicle in real time, obtain the vehicle information ahead on the road in real time, and perform multi-target position tracking and detection on the oncoming vehicles ahead and vehicles in the same direction ahead. Vehicle target tracking methods include multi-target tracking methods based on video image detection and methods based on millimeter-wave radar target detection, both of which can be used for vehicle target tracking.

[0045] like Figure 4 As shown, the multi-target tracking method for video image detection includes:

[0046] Step 401: Obtain a real-time video stream from a vehicle-mounted camera;

[0047] Step 402: preprocessing the real-time video stream;

[0048] Among them, image preprocessing is performed on the video stream, such as denoising and grayscale.

[0049] Step 403: perform vehicle target detection according to the processed video stream;

[0050] Among them, the YOLO deep learning model is used for vehicle target detection ,in, is the output of the YOLO model, is the probability of the vehicle existing (indicates the probability that vehicle c appears in bounding box b), where c represents the target category and in this embodiment represents the vehicle, and b represents the bounding box. This probability needs to be converted into a confidence score M (the confidence score M is an important indicator in the target detection model, which is used to evaluate the confidence and accuracy of the model in the detection results. The confidence score not only reflects the probability of the target existing, but also combines the accuracy of the bounding box positioning). The confidence score M is the probability that a bounding box predicted by the YOLO model contains the target multiplied by the accuracy of the detected bounding box. , IOU represents the intersection-over-union ratio between the predicted bounding box and the actual bounding box. The confidence score is closely related to the probability of the target existing and is used to judge the confidence and accuracy of the model in the detection results. The confidence score not only considers whether the target exists, but also includes the positioning accuracy of the bounding box. It is the sigmoid activation function. The original image is input to the YOLO model. The grid size into which the input image is divided is, for example, a 7×7 grid. The number of bounding boxes predicted for each grid cell is usually 2 or 5.

[0051] Step 404: extracting features of the detected vehicle;

[0052] Among them, features such as color, shape, position, etc. are extracted from the detected vehicle targets. In this step, in the multi-target detection and tracking scenario, there are usually multiple vehicle targets on the road. By extracting features, different targets can be distinguished to prevent target identity exchange during the tracking process. For example, a white car and a black SUV can be distinguished by color, and a small car and a truck can be distinguished by outline and proportion. The feature extraction in this step is equivalent to creating a "fingerprint" for the vehicle, so that even if the target is temporarily obscured (such as blocked by other vehicles), it can be re-identified and matched to the original target through features, thereby improving the robustness of detection and tracking.

[0053] Step 405: Perform multi-target tracking based on the Kalman filter to predict the next position of the vehicle.

[0054] in, , is the predicted state, A is the state transfer matrix, B is the control matrix, is the control vector, k represents the time step or the number of iterations, Represents the state predicted at the kth step.

[0055] When there are multiple oncoming vehicles, it is better to select the closest vehicle as the tracking target to serve as the basis for adjusting the silicon-based Microled pixel headlights.

[0056] The method of target detection based on millimeter wave radar includes:

[0057] Step 401a, obtaining millimeter wave radar data;

[0058] Among them, real-time point cloud data is obtained from the vehicle-mounted millimeter-wave radar.

[0059] Step 402b: pre-process the radar point cloud data:

[0060] Among them, the radar point cloud data is preprocessed, such as denoising and downsampling.

[0061] Step 403b: Detect vehicle targets based on the DBSCAN clustering algorithm;

[0062] Among them, DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is used for target detection. The DBSCAN clustering algorithm detects vehicle targets from the original millimeter-wave radar point cloud data.

[0063] Step 404c, taking each cluster center as the vehicle target position;

[0064] According to the clustering result of 403b, the features of each target are extracted for locating and identifying the vehicle.

[0065] Among them, features such as speed, distance, etc. are extracted from radar targets detected by clustering.

[0066] Step 405d: perform multi-target tracking based on the Kalman filter to predict the next position of the vehicle.

[0067] Among them, the extracted target position and speed are combined with historical tracking information to track the target vehicle in real time. The target position and speed information obtained in step 404c are used to obtain the updated status of each target vehicle in this step, including the predicted position and speed. By combining historical data and current detection results, dynamic prediction and status update of the target are completed.

[0068] The Kalman filter is used to track vehicle A. The Kalman filter assumes that the motion of the target can be described by a linear dynamic model. , is the predicted state vector (estimated value) at the next moment, The state vector includes the key information of the target vehicle (position, speed, etc.). For each target vehicle , is the current position of the target (image pixel coordinates or actual physical coordinates), is the speed of the target (pixels / s or meters / s); is the state transfer matrix; is the state vector at the current moment; B is the control input matrix (which describes how the system responds to the control input, usually related to acceleration or external force); The input vector for control is the acceleration of the target.

[0069] In millimeter-wave radar point cloud data, the target may not be accurately detected every time due to noise, occlusion or data loss. In this step, the Kalman filter makes up for this deficiency through the prediction of historical status. In this step, the Kalman filter can smooth the trajectory of the target, reduce the position jitter caused by detection noise in the point cloud data, provide a more stable target position, and predict the position at the next moment based on the historical motion state of the target. Even if the target disappears in a short time, it can rely on the prediction to maintain continuous tracking. The output predicted position and speed information can be used for real-time decision-making and for adjusting the light and dark cutoff line of the digital headlights.

[0070] Embodiment three:

[0071] In order to realize glare detection in the front environment (for example, reflection from water on rainy days and reflection from traffic signs, etc.) and avoid glare caused by pixel headlights irradiating the position, which affects the driver's driving safety, it is necessary to detect the glare position in order to control the lighting range of the silicon-based Microled pixel headlight. A glare position detection method in Embodiment 3 of the present invention is as follows: Figure 5 As shown, including:

[0072] Step 501: Obtain a video stream in front of the vehicle, and perform Gaussian filtering on the video stream to obtain a continuous image after noise elimination;

[0073] Step 502: Obtain the brightness gradient of each pixel in the continuous image, and mark the image pixels whose brightness gradient is higher than a threshold as glare positions;

[0074] Step 503, counting the number of occurrences of the glare position in the continuous images, and when the number of occurrences is greater than a preset threshold, outputting the pixel coordinates of the glare position;

[0075] Step 504: convert the pixel coordinates of the glare position into the light field coordinates of the silicon-based Microled pixel headlight to control the silicon-based Microled pixel headlight to turn off the corresponding light-emitting pixel point.

[0076] Step 502 specifically includes:

[0077] Step 502-1: Calculate the horizontal brightness gradient of each pixel and vertical brightness gradient ;

[0078] Among them, I represents the brightness or grayscale value of the image, which is a two-dimensional pixel matrix. Indicates that image I is in the horizontal direction The rate of change in the horizontal direction, that is, the gradient in the horizontal direction. Indicates that image I is in the vertical direction The rate of change in the vertical direction, that is, the gradient in the vertical direction.

[0079] Step 502-2: Calculate the brightness gradient amplitude of the pixel as the brightness gradient of a pixel.

[0080] The image brightness gradient detection algorithm can automatically identify the glare position on the image, and can detect multiple glare positions at the same time, making the glare position detection more efficient. The camera collects the video stream, counts the light sources that appear in the image in the video stream, and uses the light source that appears the most times as the glare position, which makes the glare position detection more accurate. It does not distinguish between glare source types and can solve the detection of all glare source positions.

[0081] Embodiment 4:

[0082] In order to detect the change of road curvature and adjust the lighting range of silicon-based Microled pixel headlight dot matrix in real time, adjust the light-dark cut-off line to adapt to the change of road curvature to maintain the position stability of the light-dark cut-off line, a road curvature detection method is provided in the fourth embodiment of the present invention, such as Figure 6 The network structure of the model is shown. The training steps of the model include:

[0083] Step 601: Obtain a video stream in front of the vehicle and detect the curvature of the road surface;

[0084] Among them, real-time video stream is obtained through the on-board camera, and the convolutional neural network (CNN) is used to process the video stream, extract the geometric features of the road surface and calculate the road curvature. The network model extracts the edge and texture features of the road image through multiple layers of convolutional layers and pooling layers, and identifies the changing trend of the road curvature.

[0085] Step 602: Control the lighting range of the silicon-based MicroLED pixel headlight according to the curvature of the road surface, so as to control the position of the light-dark cutoff line to adapt to the change of the curvature of the road surface.

[0086] Among them, according to the detected road curvature information, the lighting range of the silicon-based MicroLED pixel headlights is controlled to adapt to the change of road curvature. For example: when it is detected that the road is curved to the left, the controller sends instructions to the drive circuit to expand the lighting range of the MicroLED dot matrix on the left and reduce the lighting range on the right, so that the light and dark cutoff line moves to the left to ensure that more of the left side of the road is fully illuminated. Conversely, when it is detected that the road is curved to the right, the lighting range on the left is reduced and the right is expanded to ensure that the light and dark cutoff line is adjusted as the road turns right.

[0087] Optionally, to enhance the accuracy of the model, the training set can be expanded through data augmentation techniques to include road curvature change data under various weather conditions. A long short-term memory network (LSTM) is used to capture the temporal changes of road curvature in consecutive frames, thereby improving the response speed to dynamic road curvature.

[0088] Through the method of the fourth embodiment of the present invention, the light field distribution of the headlamp can be intelligently adjusted to ensure the stability of the light and dark cut-off line under various road conditions, thereby improving the safety and comfort of night driving.

[0089] Embodiment five:

[0090] Since it is necessary to simultaneously track the same-direction and opposite-direction vehicles to achieve a better target tracking effect, the fifth embodiment of the present invention comprehensively considers the spatial and temporal characteristics on the basis of the second embodiment, greatly improving the accuracy of vehicle lane change trajectory prediction, especially the target tracking effect in multi-lane and complex traffic scenes. Step 202 specifically includes:

[0091] Step 202-1, pre-processing the position information and speed information of the oncoming vehicle and the same-direction vehicle ahead;

[0092] Among them, in this step, Kalman filtering is used to perform multi-dimensional smoothing on the vehicle's multi-dimensional data (position, speed, acceleration, etc.) to reduce the impact of noise on prediction accuracy. This step enhances the generalization ability of the model in actual complex environments and provides a more reliable data basis for subsequent feature extraction and prediction. Assume that the current state of a vehicle is , after smoothing by Kalman filtering, the position and velocity at the next moment are updated as , which provides more accurate basic data for subsequent feature extraction.

[0093] Step 202-2, extracting features from the pre-processed position information and speed information of the oncoming vehicle and the same-direction vehicle ahead;

[0094] Wherein, step 202-2 specifically includes:

[0095] Step 202-21, calculating the relative position, speed, acceleration and lane change inclination angle of the target vehicle and its surrounding vehicles based on the position information and speed information of the oncoming vehicle and the same-direction vehicle ahead;

[0096] Among them, the feature input vector is composed of the relative position, speed, acceleration and lane change inclination angle of the target vehicle and its surrounding vehicles to construct a new feature vector. The calculation method of the constructed feature vector is as follows:

[0097] Relative position and velocity calculation: ,in, is the target vehicle position, is the position of surrounding vehicles, is the relative speed, capturing the dynamic difference between the target vehicle and surrounding vehicles.

[0098] Lane change tilt angle , reflecting the change in the target vehicle’s driving direction during the lane change process.

[0099] Feature vector fusion combines the relative position, speed, acceleration and tilt angle of the target vehicle and the four vehicles in front, behind, left and right into feature input , these features are fused together and input into the CNN layer.

[0100] Step 202-22, extracting spatial features of the relative position, speed, acceleration and lane change inclination angle of the target vehicle and its surrounding vehicles through the CNN layer to obtain input features of the BiLSTM layer;

[0101] Among them, the one-dimensional convolutional neural network (1D-CNN) extracts the spatial relationship between the target vehicle and the surrounding vehicles through convolution operations, and the convolution output (the output feature value at time step t and channel i after convolution operation) is , where k is the convolution kernel size, is the convolution kernel index, t is the current time position of the input sequence, indicating the starting position of the current sliding window of the convolution kernel, is the convolution kernel weight of the ith channel at position j, For input data, The convolution kernel is used as a bias. In this way, the local area of ​​the input sequence can be weighted and summed to extract representative features. The sliding operation of the convolution kernel enables the network to capture patterns in the time dimension, such as the motion trajectory characteristics or behavior patterns of the target, such as the input relative position and speed characteristics. , a convolution kernel of size 3 is used to perform sliding calculations on these features to extract the complex spatial relationship between the target vehicle and the surrounding vehicles. Then the activation function converts the linear output Mapping to nonlinear space enables the neural network to learn more complex features. The activation function can use ReLU (Rectified Linear Unit), Sigmoid, Tanh, etc. For example, ReLU outputs the convolution formula Perform non-negative restriction and truncate negative output to 0, for example .

[0102] Step 202-3: Input the extracted features into the BiLSTM layer for processing, and use the output of the BiLSTM to track the positions of the oncoming vehicle and the same-direction vehicle in front, respectively.

[0103] The BiLSTM layer is used to capture the dynamic characteristics of the vehicle in the time dimension. BiLSTM uses the forward and backward time series information to model the dynamic interaction between the target vehicle and the surrounding vehicles. BiLSTM combines past and future information to predict the trajectory of the target vehicle at the next moment. According to the BiLSTM state update formula, the forward propagation is , the back propagation is The final output is ,in, is the input vector, is the hidden state at the last moment, is the forward weight matrix, is the backward weight matrix, and are bias terms, is the activation function (usually the Sigmoid function or the Tanh function is used to introduce nonlinearity so that LSTM can learn complex time dependencies).

[0104] Embodiment 5 of the present invention establishes a refined feature input system by combining the spatial, velocity and acceleration information of the target vehicle and surrounding vehicles. The feature input includes the relative position and velocity information of the target vehicle and the surrounding vehicles in front, behind, left and right, so as to capture the interaction between vehicles in the time series data. Through this feature construction, the model can better capture the complex dynamic information during the lane change process and improve the accuracy of trajectory prediction. After extracting the features, the CNN layer is used to extract spatial features from these features, and the BiLSTM layer is used to analyze the time series features. The BiLSTM model can capture the bidirectional relationship between the vehicle's historical and future movements by processing forward and backward time series data.

[0105] Further, as Figures 2 to 6 The specific implementation of the method, the embodiment of the present invention provides a dimming device based on the light and dark cut-off line of a silicon-based Microled pixel headlight, such as Figure 7 As shown, the device comprises:

[0106] An acquisition module 710 is used to acquire position information and speed information of a front oncoming vehicle and a front same-direction vehicle;

[0107] A detection module 720, configured to track the positions of the oncoming vehicle ahead and the vehicle ahead in the same direction, respectively, according to the position information and the speed information of the oncoming vehicle ahead and the vehicle ahead in the same direction;

[0108] The control module 730 is used to dynamically control the pulse width modulation signal according to the position tracking information of the oncoming vehicle and the vehicle in the same direction ahead, and to scan the light-emitting pixels of the silicon-based Microled pixel headlight line by line to control the lighting range, so as to control the position stability of the light and dark cut-off lines of the oncoming area ahead and the vehicle in the same direction ahead.

[0109] In an embodiment of the present invention, a silicon-based Microled pixel headlight is provided, comprising a silicon-based Microled light source module, a controller, a driving circuit and a power module. The controller is connected to the driving circuit and controls the driving circuit. The power module is used to supply power to the silicon-based Microled light source module. The processor executes the steps of a dimming method based on a bright and dark cutoff line of the silicon-based Microled pixel headlight, including:

[0110] Obtain the position information and speed information of the oncoming vehicle and the vehicle in the same direction ahead;

[0111] Tracking the positions of the oncoming vehicle and the same-direction vehicle respectively according to the position information and speed information of the oncoming vehicle and the same-direction vehicle;

[0112] According to the position tracking information of the oncoming vehicle and the vehicle in the same direction ahead, the pulse width modulation signal is dynamically controlled, and the luminous pixels of the silicon-based Microled pixel headlights are scanned line by line to control the lighting range, so as to control the position stability of the light and dark cut-off lines in the oncoming area and the same direction area ahead.

[0113] It should be noted that the above embodiments only use automobile pixel headlight control as an example to illustrate the principles and implementation steps of the embodiments of the present invention, and do not specifically limit the actual application scenarios. For example, electric bicycles, electric motorcycles, etc., the technical solution of the present invention can also be applied to the front lighting control systems of other vehicles. For the functions or steps that can be implemented by the computer-readable storage medium or computer device, please refer to the aforementioned method embodiments. To avoid repetition, they will not be described one by one here.

[0114] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0115] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0116] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for dimming the light cutoff line of a silicon-based Microled pixel headlight, characterized in that: include: Obtain the position information and speed information of the oncoming vehicle and the vehicle in the same direction ahead; Tracking the positions of the oncoming vehicle ahead and the same-direction vehicle ahead respectively according to the position information and the speed information of the oncoming vehicle ahead and the same-direction vehicle ahead; According to the position tracking information of the oncoming vehicle and the vehicle in the same direction ahead, the pulse width modulation signal is dynamically controlled, and the luminous pixels of the silicon-based Microled pixel headlights are scanned line by line to control the lighting range, so as to control the position stability of the light and dark cut-off lines in the oncoming area and the same direction area ahead.

2. The method for dimming the light cutoff line of a silicon-based Microled pixel headlight according to claim 1, characterized in that: The steps of respectively tracking the positions of the oncoming vehicle ahead and the vehicle ahead in the same direction according to the position information and the speed information of the oncoming vehicle ahead and the vehicle ahead in the same direction include: Preprocessing the position information and the speed information of the front oncoming vehicle and the front same-direction vehicle; Extracting features from the preprocessed position information and speed information of the oncoming vehicle and the same-direction vehicle ahead; The extracted features are input into the BiLSTM layer for processing, and the output of BiLSTM is used to track the positions of the oncoming vehicle and the same-direction vehicle in front, respectively.

3. The method for dimming the light cutoff line of a silicon-based Microled pixel headlight according to claim 2, characterized in that: The step of extracting features from the preprocessed position information and speed information of the front oncoming vehicle and the front same-direction vehicle comprises: Calculate the relative position, speed, acceleration and lane change inclination angle of the target vehicle and its surrounding vehicles according to the position information and speed information of the front oncoming vehicle and the front same-direction vehicle; The CNN layer is used to extract spatial features of the relative position, speed, acceleration and lane change inclination angle between the target vehicle and its surrounding vehicles to obtain the input features of the BiLSTM layer.

4. The method for dimming the light cutoff line of a silicon-based Microled pixel headlight according to claim 1, characterized in that: The step of dynamically controlling the pulse width modulation signal to scan the light-emitting pixels of the silicon-based Microled pixel headlight row by row to control the lighting range includes: According to the real-time position of the oncoming vehicle in front, the angle and distance of the oncoming vehicle in front are obtained, and according to the angle and distance of the oncoming vehicle in front, the lateral closing range of the silicon-based Microled pixel headlight matrix is ​​calculated; According to the distance to the oncoming vehicle in front, the longitudinal lighting range of the silicon-based Microled pixel headlight matrix is ​​calculated.

5. The method for dimming the light cutoff line of a silicon-based Microled pixel headlight according to claim 1, characterized in that: After the step of dynamically controlling the pulse width modulation signal to scan the light-emitting pixels of the silicon-based Microled pixel headlight line by line to control the lighting range, the method includes: The grayscale mapping algorithm is used to control the grayscale value of each point in the light field distribution pattern to achieve the transition from light to dark within the preset range of the edge of the light and dark cutoff line. ,in, is the gray value of the i-th point in the light field distribution pattern, is the light intensity at the i-th point, and is the adjusted light intensity at the i-th point.

6. The method for dimming the light cutoff line of a silicon-based Microled pixel headlight according to claim 1, characterized in that: After the step of respectively tracking the positions of the oncoming vehicle ahead and the vehicle ahead in the same direction according to the position information and the speed information of the oncoming vehicle ahead and the vehicle ahead in the same direction, the method includes: Obtain the video stream in front of the vehicle and perform Gaussian filtering on the video stream to obtain a continuous image after noise elimination; Acquire the brightness gradient of each pixel in the continuous image, and mark the image pixels whose brightness gradient is higher than a threshold as glare positions; Counting the number of times the glare position appears in the continuous images, and when the number of occurrences is greater than a preset threshold, outputting the pixel coordinates of the glare position; The pixel coordinates of the glare position are converted into the light field coordinates of the silicon-based Microled pixel headlight to control the silicon-based Microled pixel headlight to turn off the corresponding Microled dot matrix.

7. The method for dimming the light cutoff line of a silicon-based Microled pixel headlight according to claim 5, characterized in that: The step of obtaining the brightness gradient of each pixel in the continuous image comprises: Calculate the horizontal brightness gradient of each pixel and vertical brightness gradient , Represents pixels in the horizontal direction, Represents pixels in the vertical direction; Calculate the brightness gradient magnitude of the pixel as the brightness gradient of a pixel.

8. The method for dimming the light cutoff line of a silicon-based Microled pixel headlight according to claim 1, characterized in that: After the step of respectively tracking the positions of the oncoming vehicle ahead and the vehicle ahead in the same direction according to the position information and the speed information of the oncoming vehicle ahead and the vehicle ahead in the same direction, the method includes: Obtain the video stream in front of the vehicle and detect the curvature of the road surface; According to the curvature of the road surface, the lighting range of the silicon-based Microled pixel headlights is controlled to control the position of the light and dark cut-off line to adapt to changes in the curvature of the road surface.

9. A dimming device for a silicon-based Microled pixel headlight with a cut-off line, characterized in that: include: An acquisition module is used to acquire the position information and speed information of the oncoming vehicle and the same-direction vehicle ahead; A detection module, used for tracking the positions of the oncoming vehicle ahead and the vehicle ahead in the same direction, respectively, according to the position information and the speed information of the oncoming vehicle ahead and the vehicle ahead in the same direction; The control module is used to dynamically control the pulse width modulation signal according to the position tracking information of the oncoming vehicle and the vehicle in the same direction in front, and to scan the light-emitting pixels of the silicon-based Microled pixel headlights line by line to control the lighting range, so as to control the position stability of the light and dark cut-off lines of the oncoming area and the area in the same direction in front.

10. A silicon-based Microled pixel headlight, comprising a silicon-based Microled light source module, a controller, a driving circuit and a power module, wherein the controller is connected to the driving circuit and controls the driving circuit, and the power module is used to supply power to the silicon-based Microled light source module, characterized in that: The controller executes the steps of the cut-off line dimming method for a silicon-based Microled pixel headlight as claimed in any one of claims 1 to 8.

Images