A driving fatigue detection system and method based on video monitoring

Abstract

The application discloses a driving fatigue detection system and method based on video monitoring, and relates to the technical field of safe driving of automobiles.The system comprises a camera module and an image processing module, a fatigue detection module and a warning module.The camera module is used for collecting video data in the driving process of a vehicle and inputting the video data into the image processing module.The image processing module comprises a gray processing submodule, a noise processing submodule and an image enhancement submodule.The image processing module is used for extracting driving track features of the vehicle based on the video data.The fatigue detection module comprises a lane line detection submodule, a vehicle motion feature detection submodule and a driving behavior feature detection submodule.The fatigue detection module is used for detecting whether a driver is in a fatigue state according to the extracted driving track features.The application can more accurately detect the fatigue state of a driver.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of safe driving of automobiles, and particularly relates to a driving fatigue detection system and method based on video monitoring. BACKGROUND

[0002] With the rapid development of the transportation industry, the number of vehicles has increased dramatically, and traffic accidents have also occurred frequently. Fatigue driving is one of the important reasons for traffic accidents. Fatigue driving refers to the phenomenon that after a long time of continuous driving, the driver's physiological and psychological functions decline, leading to a decrease in driving skills, and thus causing traffic accidents. Fatigue driving can seriously affect the driver's attention, reaction ability, judgment ability, etc., and increase the risk of operation failure and accidents. According to relevant statistics, traffic accidents caused by fatigue driving account for a high proportion, which poses a serious threat to people's life and property safety. SUMMARY

[0003] The purpose of the present application is to provide a driving fatigue detection system and method based on video monitoring, which can significantly improve the accuracy of fatigue driving detection.

[0004] To achieve the above purpose, the present application provides the following solutions:

[0005] In a first aspect, the present application provides a driving fatigue detection system based on video monitoring, comprising: a camera module and an image processing module, a fatigue detection module and a warning module.

[0006] The camera module is used to collect video data during vehicle driving, and input the video data to the image processing module.

[0007] The image processing module comprises a gray processing submodule, a noise processing submodule and an image enhancement submodule; the image processing module is used to extract the driving track features of the vehicle based on the video data.

[0008] The fatigue detection module comprises a lane line detection submodule, a vehicle motion feature detection submodule and a driving behavior feature detection submodule; the fatigue detection module is used to detect whether the driver is in a fatigue state according to the extracted driving track features.

[0009] Optionally, it further comprises a warning module, which is used to issue a warning signal to the driver when it is detected that the driver is in a fatigue state.

[0010] Optionally, the gray processing submodule is used to convert the color image in the video data into a gray image based on the weighted average method and the average value method.

[0011] Optionally, the noise processing submodule is configured to remove noise in the grayscale image based on a filtering method; the filtering method includes mean filtering, median filtering, Gaussian filtering, and bilateral filtering.

[0012] Optionally, the image enhancement submodule is configured to perform image enhancement on the grayscale image after noise removal in a manner of histogram equalization, adaptive histogram equalization, logarithmic transformation, and gamma correction, to obtain a processed driving image.

[0013] Optionally, the lane line detection submodule is configured to:

[0014] perform edge detection on the processed driving image based on a Canny edge detection method to determine an edge region in the driving image; the edge region is a region in which a grayscale value or a color value in the driving image changes.

[0015] determine a straight lane line in the driving image based on the edge region in the driving image by using a Hough transform.

[0016] determine a curved lane line in the driving image based on the edge region in the driving image by using polynomial fitting.

[0017] Optionally, the vehicle motion feature detection submodule is configured to determine a motion trajectory of the vehicle based on each frame of image in the video data by using an optical flow method.

[0018] Optionally, the driving behavior feature detection submodule is configured to: acquire time series data of vehicle speed and time series data of vehicle steering wheel angle; calculate vehicle acceleration based on the time series data of vehicle speed; calculate a vehicle steering angle change rate based on the time series data of vehicle steering wheel angle; and determine a driving behavior of the driver based on the vehicle acceleration and the vehicle steering angle change rate.

[0019] In a second aspect, the present application provides a driving fatigue detection method based on video monitoring, which comprises:

[0020] acquiring video data in a vehicle driving process.

[0021] performing image processing on the video data to extract a driving trajectory feature of the vehicle.

[0022] detecting whether a driver is in a fatigue state based on a lane line detection submodule, a vehicle motion feature detection submodule, and a driving behavior feature detection submodule according to the driving trajectory feature.

[0023] Optionally, when it is detected that the driver is in a fatigue state, a warning signal is sent to the driver.

[0024] According to the specific embodiments provided in the application, the application discloses the following technical effects:

[0025] The application provides a driving fatigue detection system and method based on video monitoring. The system adopts a mode in which multiple modules such as a camera module, an image processing module, a fatigue detection module, and a warning module work cooperatively. This multi-module design enables the system to analyze the state of the driver from multiple angles and levels, thereby improving the comprehensiveness and accuracy of detection. The image processing module pre-processes the collected video data through sub-modules such as grayscale processing, noise processing, and image enhancement. Based on the video data, the driving trajectory features of the vehicle are extracted. These features can reflect the driving stability of the vehicle and the driving behavior of the driver, providing an important basis for fatigue detection. The fatigue detection module includes a lane line detection sub-module, a vehicle motion feature detection sub-module, and a driving behavior feature detection sub-module. These sub-modules detect the driving state of the driver from different angles and fuse the detection results. Multi-feature fusion detection can more comprehensively evaluate the fatigue degree of the driver, thereby improving the accuracy of detection. The driving fatigue detection system based on video monitoring can more accurately detect the fatigue state of the driver. BRIEF DESCRIPTION OF DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings described below are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0027] Figure 1 A structural schematic diagram of a driving fatigue detection system based on video monitoring is provided for the first embodiment of the application.

[0028] Figure 2 A flowchart of a driving fatigue detection method based on video monitoring is provided for the second embodiment of the application. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only some of the embodiments of the application, not all. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0030] In order to make the above-mentioned purposes, features and advantages of the application more obvious and easy to understand, the application will be further described in detail below with reference to the drawings and specific embodiments.

[0031] Embodiment one

[0032] As shown in the embodiment, a driving fatigue detection system based on video monitoring is provided, characterized in that it comprises a camera module and an image processing module, a fatigue detection module and a warning module. Figure 1

[0033] The camera module is used to collect video data during vehicle driving and input the video data to the image processing module.

[0034] The image processing module comprises a grayscale processing submodule, a noise processing submodule and an image enhancement submodule; the image processing module is used to extract the driving trajectory features of the vehicle based on the video data.

[0035] The fatigue detection module comprises a lane line detection submodule, a vehicle motion feature detection submodule and a driving behavior feature detection submodule; the fatigue detection module is used to detect whether the driver is in a fatigue state according to the extracted driving trajectory features.

[0036] In the embodiment, the camera module comprises a front camera of the vehicle and an optional rear camera.

[0037] The installation position of the front camera is located at the front of the vehicle, for example, mounted on the front bumper or the front windshield of the vehicle. Its function is to capture real-time driving video data in front of the vehicle in order to detect the front lane line and driving trajectory.

[0038] The installation position of the rear camera is located at the rear of the vehicle, for example, mounted on the rear bumper or the rear windshield of the vehicle. Its function is to supplement the capture of driving video data behind the vehicle to achieve more comprehensive analysis of the driving trajectory.

[0039] In some embodiments of the embodiment, the image processing module can be specifically as follows:

[0040] The image processing module comprises a grayscale processing submodule, a noise processing submodule and an image enhancement submodule, and further comprises a video input interface for receiving the original video data collected by the front and rear cameras.

[0041] The grayscale processing submodule is used to convert the color image in the video data into a grayscale image based on the weighted average method and the average value method.

[0042] Specifically, the grayscale conversion is the process of converting a color image into a grayscale image, and the following methods are usually used:

[0043] Weighted average method: weighted average according to the weight of RGB channel, such as the commonly used formula:

[0044] ​Gray = 0.299 x R + 0.587 x G + 0.114 x B Gray = 0.299 x R + 0.587 x G + 0.114 x B Gray = 0.299 x R + 0.587 x G + 0.114 x B.

[0045] Mean value method: directly take the average value of RGB channel: Gray = R + G + B 3 Gray = 3 R + G + B.

[0046] The noise processing submodule is configured to remove noise in the grayscale image based on a filtering method, and the filtering method includes mean filtering, median filtering, Gaussian filtering, and bilateral filtering.

[0047] Specifically, the commonly used methods include: mean filtering: using the average value of the neighborhood pixels to replace the center pixel value, which can smooth the image but will blur the edges. The algorithm is that each element of the convolution kernel is 1, and the size is, for example, 3x3 or 5x5. Median filtering: using the median value of the neighborhood pixels to replace the center pixel value, which is suitable for removing salt and pepper noise; the algorithm is to sort the pixel values in the neighborhood, and take the middle value. Gaussian filtering: using Gaussian distribution weight to weight the average of the neighborhood pixels, which has good smoothing effect and can preserve some edge information; the algorithm is that the convolution kernel weight is calculated according to the Gaussian function, and the commonly used size is, for example, 3x3 or 5x5. Bilateral filtering: considering both spatial proximity and pixel value similarity, which can well preserve the edges; the algorithm: the convolution kernel weight combines spatial distance and pixel value difference.

[0048] The image enhancement submodule is configured to perform image enhancement on the grayscale image after noise removal in a histogram equalization, adaptive histogram equalization, logarithmic transformation, and gamma correction manner to obtain a processed driving image. The commonly used methods include:

[0049] 1) Histogram equalization: adjusting the gray histogram of the image to enhance the contrast. Algorithm: re-distribute the pixel values to be uniformly distributed on all gray levels.

[0050] 2) Adaptive histogram equalization (CLAHE): applying histogram equalization in local regions to avoid over-enhancement. Algorithm: divide the image into small blocks, perform histogram equalization on each block, and smooth the boundaries.

[0051] 3) Logarithmic transformation: using a logarithmic function to compress the dynamic range of the image to enhance details. Algorithm:

[0052] where c is a constant.

[0053] 4) Gamma Correction: Adjusts the brightness of an image by applying a non-linear transformation through a gamma function. Algorithm: Output = Input^γ, where γ is the gamma value.

[0054] Here is a code example (using OpenCV library):

[0055] python

[0056] Copy the code

[0057] import cv2

[0058] import numpy as np

[0059] # Read the image

[0060] image = cv2.imread('input_image.jpg')

[0061] # Grayscale

[0062] gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

[0063] # Mean filtering

[0064] mean_filtered_image = cv2.blur(gray_image, (3, 3))

[0065] # Median filtering

[0066] median_filtered_image = cv2.medianBlur(gray_image, 3)

[0067] # Gaussian filtering

[0068] gaussian_filtered_image = cv2.GaussianBlur(gray_image, (5, 5), 0)

[0069] # Bilateral filtering

[0070] bilateral_filtered_image = cv2.bilateralFilter(gray_image, 9, 75, 75)

[0071] # Histogram equalization

[0072] equalized_image = cv2.equalizeHist(gray_image)

[0073] # Adaptive Histogram Equalization (CLAHE)

[0074] clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))

[0075] clahe_image = clahe.apply(gray_image)

[0076] # Log Transformation

[0077] c = 255 / np.log(1 + np.max(gray_image))

[0078] log_transformed_image = c * np.log(1 + gray_image)

[0079] log_transformed_image = np.array(log_transformed_image, dtype=np.uint8) # Gamma Correction

[0080] gamma = 2.0

[0081] gamma_corrected_image = np.array(255 * (gray_image / 255) ** gamma, dtype='uint8')

[0082] # Display Images

[0083] cv2.imshow('Gray Image', gray_image)

[0084] cv2.imshow('Mean Filtered Image', mean_filtered_image)

[0085] cv2.imshow('Median Filtered Image', median_filtered_image)

[0086] cv2.imshow('Gaussian Filtered Image', gaussian_filtered_image)

[0087] cv2.imshow('Bilateral Filtered Image', bilateral_filtered_image)

[0088] cv2.imshow('Equalized Image', equalized_image)

[0089] cv2.imshow('CLAHE Image', clahe_image)

[0090] cv2.imshow('Log Transformed Image', log_transformed_image)

[0091] cv2.imshow('Gamma Corrected Image', gamma_corrected_image)

[0092] cv2.waitKey(0)

[0093] cv2.destroyAllWindows()

[0094] The above is the image preprocessing using the OpenCV library.

[0095] In some embodiments of the present embodiment, the fatigue detection module comprises a lane line detection submodule, a vehicle motion feature detection submodule, and a driving behavior feature detection submodule.

[0096] The lane line detection submodule is configured to:

[0097] Based on the Canny edge detection method, the edge region in the processed driving image is determined by performing edge detection on the processed driving image; the edge region is a region in which the gray value or color value in the driving image changes.

[0098] Based on the edge region in the driving image, the straight lane line in the driving image is determined by using Hough transform.

[0099] Based on the edge region in the driving image, the curved lane line in the driving image is determined by using polynomial fitting.

[0100] In the field of image processing and computer vision, edge detection is a commonly used technique to identify the boundaries of objects in an image. Specifically, Canny edge detection is a widely used method for edge detection that is based on a multi-stage detection algorithm, including steps such as noise filtering, gradient calculation, non-maximum suppression, and double thresholding. This method is effective in detecting edges in an image, especially those with significant intensity changes. In edge detection, an edge refers to a place in an image where the intensity or color value changes significantly, which often corresponds to the boundary of an object.

[0101] For lane line detection, edge features typically appear as distinct intensity contrasts between the road and lane lines. To detect straight lines in an image, the Hough transform can be used. However, for curved lane lines, traditional Hough transforms may not accurately describe their shape. In such cases, polynomial fitting (e.g., quadratic polynomials) can be used to describe the curved portions of the lane lines. By extracting points on the lane lines from the edge detection results, curve fitting can be performed to obtain a more accurate lane line model.

[0102] In addition to traditional image processing methods, deep learning methods have shown great potential in lane line detection. Convolutional Neural Networks (CNNs) are powerful deep learning models that are particularly suitable for image recognition and classification tasks. In lane line detection, end-to-end CNN models such as U-Net and SegNet can learn directly from raw images and predict the positions of lane lines. These models can extract features from images through multiple layers of convolution and pooling operations, and can be trained through backpropagation algorithms to achieve high-precision lane line detection.

[0103] Furthermore, instance segmentation methods such as Mask R-CNN are also applied to lane line detection. Instance segmentation not only identifies lane lines in an image, but also distinguishes them from other objects. Mask R-CNN generates a mask for each pixel belonging to a lane line by classifying at the pixel level, thereby achieving accurate segmentation of lane lines. This method not only detects the presence of lane lines, but also provides detailed shape and position information of lane lines, which has important application value for autonomous driving and intelligent transportation systems.

[0104] The following is an OpenCV-based code example

[0105]

[0106] Among them, the vehicle motion feature detection submodule of the fatigue detection module is used to determine the motion trajectory of the vehicle based on each frame of image in the video data by using the optical flow method.

[0107] The extraction of vehicle motion features typically involves elements such as speed, acceleration, steering angle, etc., and methods include but are not limited to the following:

[0108] Kalman filter technology, which is widely used for the estimation and tracking of vehicle position, speed and acceleration. Optical flow method, which estimates the movement of pixels in the image and then deduces the vehicle's motion trajectory. Inertial Measurement Unit (IMU), containing sensors such as accelerometers and gyroscopes, used to accurately detect the vehicle's acceleration and angular velocity.

[0109] The following is an example code using OpenCV and the optical flow method

[0110]

[0111]

[0112] Among them, the driving behavior feature detection submodule of the fatigue detection module is used to: acquire time series data of vehicle speed and time series data of vehicle steering wheel angle; calculate the vehicle acceleration according to the time series data of vehicle speed; calculate the vehicle steering angle change rate according to the time series data of vehicle steering wheel angle; determine the driving behavior of the driver according to the vehicle acceleration and the vehicle steering angle change rate.

[0113] During the process of driving behavior feature extraction, some key behavior features are usually focused on, such as sudden acceleration, sudden braking and sudden turning, etc. In order to accurately extract these features, the following methods can be used:

[0114] First of all, the rate of change of speed is an important indicator. By analyzing the time series data of vehicle speed, the change of acceleration can be calculated. This method can help detect whether the driver has sudden acceleration or sudden braking behavior during driving. Specifically, by differentiating the speed data, the time series of acceleration can be obtained, and then a threshold value is set. When the acceleration exceeds this threshold value, it can be judged as sudden acceleration or sudden braking.

[0115] Secondly, the steering angle change rate is also a key feature. By analyzing the angle data of the steering wheel, the change rate of the steering angle can be calculated. This method can help detect whether the driver has sudden turning behavior during driving. Specifically, by differentiating the steering wheel angle data, the time series of steering angle change rate can be obtained, and then a threshold value is set. When the steering angle change rate exceeds this threshold value, it can be judged as sudden turning.

[0116] Finally, driving mode classification is a more comprehensive approach. By using machine learning algorithms, the driving behavior of the driver can be classified. Common machine learning algorithms include decision trees, random forests, and support vector machines (SVM), among others. These algorithms can learn the characteristics of the driver's behavior through training data and classify new driving behavior. Specifically, a large amount of driving behavior data can be collected, including speed, steering angle, acceleration, and other information, and then used to train a machine learning model. Through the trained model, the driving behavior of the driver can be classified, thereby identifying different driving modes, such as aggressive driving, smooth driving, etc.

[0117] The following is an example of code for acceleration and sudden acceleration detection

[0118] python

[0119] Copy the code

[0120] import numpy as np

[0121] # Example speed data (in m / s)

[0122] speed_data = [0, 2, 4, 6, 8, 10, 12, 15, 20, 22, 20, 18, 16, 14, 12, 10, 8, 6, 4, 2, 0]

[0123] # Calculate acceleration

[0124] acceleration_data = np.diff(speed_data)

[0125] # Set sudden acceleration threshold (in m / s^2)

[0126] acceleration_threshold = 3

[0127] # Detect sudden acceleration

[0128] sudden_acceleration_events = np.where(acceleration_data > acceleration_threshold)[0]

[0129] print("Sudden acceleration events occurred at time points:", sudden_acceleration_events).

[0130] The above algorithms and methods can help build an effective driving fatigue detection system. In practical applications, a variety of features and algorithms can be combined, and machine learning or deep learning models can be used to further improve detection accuracy and reliability.

[0131] In addition, in some embodiments of the present embodiment, the driving fatigue detection system further comprises a warning module for issuing a warning signal to the driver when the driver is detected to be in a fatigue state.

[0132] Embodiment two

[0133] The present embodiment provides a driving fatigue detection method based on video monitoring, which comprises:

[0134] Step 201: Collecting video data during vehicle driving.

[0135] Step 202: Image processing of the video data to extract the driving trajectory features of the vehicle.

[0136] Step 203: According to the driving trajectory features, based on the lane line detection sub-module, vehicle motion feature detection sub-module and driving behavior feature detection sub-module, detecting whether the driver is in a fatigue state.

[0137] Step 204: When the driver is detected to be in a fatigue state, issuing a warning signal to the driver.

[0138] In summary, the present application has the following technical effects:

[0139] 1) Non-contact detection: The detection system based on video and vehicle motion features in the present application collects data by installing a camera and using existing vehicle sensors (such as speed sensors, accelerometers, etc.), without direct contact with the driver. This feature makes it more convenient and comfortable in practical applications, reducing the disturbance to the driver.

[0140] 2) Comprehensive analysis: The method provided by the present application can comprehensively analyze the driving trajectory of the vehicle, the vehicle motion features (such as speed, acceleration, steering angle, etc.) and the driving behavior features (such as sudden acceleration, sudden braking, sudden turning, etc.). Through comprehensive analysis of multiple data sources, the fatigue state of the driver can be more accurately detected, reducing the possibility of false positives and false negatives.

[0141] 3) Strong real-time performance: The present application uses cameras and vehicle sensors for data collection and analysis, which can realize real-time detection and warning. When the system detects that the driver may be in a fatigue state, an alarm can be issued immediately to help the driver take timely measures to prevent accidents.

[0142] 4) Strong environmental adaptability: The detection system based on video and vehicle motion characteristics in this application can adapt to different driving environments. Whether it's day or night, urban roads or highways, this method can effectively perform detection. Furthermore, by analyzing environmental characteristics such as road type and traffic flow, the accuracy of detection can be further improved.

[0143] 5) Relatively low cost: Compared to some high-precision physiological signal detection devices (such as electroencephalogram (EEG) detectors and electrocardiogram (ECG) monitoring devices, the detection system based on video and vehicle motion characteristics has a relatively low cost. Modern vehicles are typically equipped with multiple sensors and cameras, which can be directly used for driver fatigue detection, reducing the investment in additional equipment.

[0144] 6) Privacy Protection: This application has an advantage in terms of privacy protection because it does not require the collection of the driver's physiological signals. The driver does not need to worry about the leakage or misuse of physiological data.

[0145] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0146] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A driver fatigue detection system based on video monitoring, characterized in that, include: Camera module and image processing module, fatigue detection module and early warning module; The camera module is used to collect video data during the vehicle's movement and input the video data into the image processing module; The image processing module includes a grayscale processing submodule, a noise processing submodule, and an image enhancement submodule; the image processing module is used to extract the vehicle's driving trajectory features based on the video data; The fatigue detection module includes a lane line detection submodule, a vehicle motion feature detection submodule, and a driving behavior feature detection submodule; The fatigue detection module is used to detect whether the driver is fatigued based on the extracted driving trajectory features. The driving behavior feature detection submodule is used to: acquire time-series data of vehicle speed and time-series data of vehicle steering wheel angle; calculate vehicle acceleration based on the time-series data of vehicle speed; calculate the rate of change of vehicle steering angle based on the time-series data of vehicle steering wheel angle; and determine the driver's driving behavior based on vehicle acceleration and the rate of change of vehicle steering angle.

2. The video-based driver fatigue detection system according to claim 1, characterized in that, It also includes a warning module, which issues a warning signal to the driver when driver fatigue is detected.

3. The video-based driver fatigue detection system according to claim 2, characterized in that, The grayscale processing submodule is used to convert color images in the video data into grayscale images based on weighted average and average methods.

4. The video-based driver fatigue detection system according to claim 3, characterized in that, The noise processing submodule is used to remove noise from grayscale images based on filtering methods; the filtering methods include mean filtering, median filtering, Gaussian filtering, and bilateral filtering.

5. A video-based driver fatigue detection system according to claim 4, characterized in that, The image enhancement submodule is used to enhance the noise-removed grayscale image by employing histogram equalization, adaptive histogram equalization, logarithmic transformation, and gamma correction to obtain the processed driving image.

6. A video-based driver fatigue detection system according to claim 5, characterized in that, The lane line detection submodule is used for: Based on the Canny edge detection method, edge detection is performed on the processed driving image to determine the edge regions in the driving image; the edge regions are areas in the driving image where the grayscale value or color value changes. Based on the edge regions in the driving image, the Hough transform is used to determine the straight lane lines in the driving image; Based on the edge regions in the driving image, polynomial fitting is used to determine the curved lane lines in the driving image.

7. A video-based driver fatigue detection system according to claim 6, characterized in that, The vehicle motion feature detection submodule is used to determine the vehicle's motion trajectory based on each frame of the video data using optical flow.

8. A method for detecting driver fatigue using a video-based driver fatigue detection system according to any one of claims 1-7, characterized in that, The driving fatigue detection method includes: Collect video data during vehicle operation; The video data is processed to extract the vehicle's driving trajectory features; Based on the driving trajectory characteristics, and using the lane line detection submodule, vehicle motion feature detection submodule, and driving behavior feature detection submodule, it is determined whether the driver is fatigued.

9. The method for detecting driver fatigue based on video monitoring according to claim 8, characterized in that, When driver fatigue is detected, a warning signal is issued to the driver.

Images