Fatigue driving detection method and device

Abstract

The present invention relates to the technical field of vehicle driving control, and specifically discloses a fatigue driving detection method and device, the method comprising: image clarity value determination, fatigue feature abnormal value evaluation and fatigue driving state detection, determining the clarity value of a facial recognition image, analyzing the facial recognition image by a vehicle-mounted terminal, and evaluating the fatigue feature abnormal value of the driver's facial recognition image within a fatigue evaluation cycle; matching environmental influencing factors and driving behavior influencing factors, and combining the fatigue feature abnormal value of the driver's facial recognition image within the fatigue evaluation cycle to warn the driver's driving fatigue state, once the fatigue driving state is detected, the fatigue driving detection system will immediately issue a warning and take preventive measures, thereby effectively avoiding traffic accidents caused by fatigue driving, thereby completing the driver's fatigue driving detection, so as to optimize the driving experience and improve traffic efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle driving control, and in particular to a fatigue driving detection method and device. Background Art

[0002] With the rapid development of social economy and the booming transportation industry, road traffic conditions are becoming increasingly complex and traffic safety issues are becoming increasingly prominent. Fatigue driving, as one of the important factors causing traffic accidents, poses a serious threat to people's lives and property. Therefore, developing an effective fatigue driving detection method is of great significance to improving road traffic safety.

[0003] For example, the invention patent with the announcement number CN114670847B announces a driving fatigue monitoring method, controller, system and vehicle, wherein the method includes: when a driving fatigue monitoring activation signal is detected, obtaining the value of the vehicle's lateral motion parameter and the vehicle's entire driving mode; performing fatigue feature identification according to the value of the lateral motion parameter and the vehicle's entire driving mode, and obtaining a driving fatigue judgment result; when the driving fatigue judgment result is fatigue driving, outputting a fatigue driving warning signal; without adding physiological monitoring equipment to the vehicle, when driving fatigue monitoring is required, fatigue driving identification and warning are performed by obtaining the value of the vehicle's lateral motion parameter and the vehicle's current entire driving mode, thereby avoiding increasing vehicle costs while ensuring driving fatigue monitoring of the driver, and combining the vehicle's entire driving mode for judgment, so that the same value may have different judgment results under different vehicle driving modes.

[0004] For example, the invention patent with announcement number CN107117175B announces a fatigue driving reminder method and terminal device, the method includes: detecting whether the user is in a driving state, when it is detected that the user is in a driving state, starting the physical fatigue detection function of the terminal device, and under the physical fatigue detection function, monitoring the user's body heart rate value through the heart rate sensor on the terminal device, and determining the user's current heart rate fluctuation pattern based on the monitored body heart rate value, when the determined current heart rate fluctuation pattern matches the preset heart rate fluctuation pattern, determining that the user is in a fatigue driving state, and outputting a warning message for the fatigue driving state.

[0005] In combination with the above technical solutions, it is found that the following shortcomings are still present in the technical solutions for fatigue driving detection: on the one hand, it only relies on vehicle driving data to judge the fatigue state, and on the other hand, it focuses on the driver himself, but the judgment dimension is relatively single. For complex and changeable situations during vehicle driving, it will lead to errors in judgment, thereby reducing the accuracy of fatigue driving detection and failing to provide strong protection for driving safety. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a fatigue driving detection method and device, which can effectively solve the problems involved in the above-mentioned background technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: The first aspect of the present invention provides a fatigue driving detection method, including: S1. Performing facial recognition on the driver through a vehicle-mounted camera, obtaining the driver's facial recognition image data, determining the clarity value of the facial recognition image, and comparing it with a predefined clarity preset value to obtain a comparison result; S2. Based on the comparison result, the vehicle-mounted terminal analyzes the facial recognition image and evaluates the fatigue feature abnormality value of the driver's facial recognition image within the fatigue assessment cycle; S3. Obtaining the environmental data inside the vehicle during the fatigue assessment cycle and the driving behavior data of the driver during the fatigue assessment cycle, thereby matching the environmental influencing factors and the driving behavior influencing factors, and combining the fatigue feature abnormality value of the driver's facial recognition image during the fatigue assessment cycle, the fatigue monitoring component comprehensively determines the driver's driving fatigue state, and warns the driver of the driving fatigue state, thereby completing the driver's fatigue driving detection.

[0008] As a further method, the clarity value of the facial recognition image is determined, and the specific determination process is as follows: the facial recognition image data of the driver specifically includes the number of feature recognition points of the eye area recognition image, the contrast of the eye area recognition image at each pixel point, and the noise value of the eye area recognition image at each pixel point; the number of feature recognition points of the eye area recognition image, the contrast of the eye area recognition image at each pixel point, and the noise value of the eye area recognition image at each pixel point are comprehensively processed to determine the first clarity value of the facial recognition image; the grayscale value of each pixel point of the eye recognition image is obtained, and the standard deviation is processed to obtain the grayscale change index of the eye recognition image, and the grayscale change is comprehensively processed with the predefined grayscale change definition index to obtain the second clarity value of the facial recognition image; according to the first clarity value of the facial recognition image and the second clarity value of the facial recognition image, the clarity value of the facial recognition image is comprehensively determined, and the determination method is as follows:

[0009]

[0010] in

[0011] In the formula, is the clarity value of the facial recognition image, is the first clarity value of the face recognition image, is the second clarity value of the facial recognition image, is the weight factor corresponding to the first clarity value predefined in the detection information database, is a weight factor corresponding to the second clarity value predefined in the detection information database, The number of feature points for eye region recognition images, The contrast of the eye area recognition image at the yth pixel, where y is the number of each pixel. , Y is the total number of pixels, The noise value of the eye area recognition image at the y-th pixel, It is the grayscale change index of the eye recognition image. Define indicators for grayscale changes, To detect the impact factor corresponding to a single feature recognition point predefined in the information database, To detect the impact factor corresponding to the contrast unit value predefined in the information database, To detect the impact factor corresponding to the noise value unit value predefined in the information database, It is the influence factor corresponding to the grayscale change index difference unit value predefined in the detection information library, and e is a natural constant.

[0012] As a further method, based on the comparison result, the vehicle-mounted terminal analyzes the facial recognition image, and the specific analysis process is: if the comparison result shows that the clarity value of the facial recognition image is greater than the preset clarity value, the vehicle-mounted terminal directly analyzes the facial recognition image; if the comparison result shows that the clarity value of the facial recognition image is less than or equal to the preset clarity value, the clarity of the facial recognition image is optimized, so that the clarity value of the facial recognition image after the clarity optimization is greater than the preset clarity value, and finally the vehicle-mounted terminal analyzes the facial recognition image.

[0013] As a further method, the fatigue feature abnormality value of the driver's facial recognition image during the fatigue assessment cycle is evaluated, and the specific analysis process is: the facial recognition image is analyzed by the on-board terminal to obtain data of the facial recognition image analyzed by the on-board terminal during the fatigue assessment cycle, specifically including the number of blinks of the driver during the fatigue assessment cycle and the maximum angle value of the driver's head tilt during the fatigue assessment cycle; the number of blinks of the driver during the fatigue assessment cycle is ratio-processed with the duration corresponding to the fatigue assessment cycle to obtain the driver's blinking frequency during the fatigue assessment cycle; based on the driver's blinking frequency during the fatigue assessment cycle, the driver's head tilt maximum angle during the fatigue assessment cycle and the number of feature recognition points of the eye area recognition image, a comprehensive analysis is performed to obtain and record the fatigue feature abnormality value of the driver's facial recognition image during the fatigue assessment cycle.

[0014] As a further method, the environmental influencing factors and driving behavior influencing factors are matched, and the specific matching process is: the environmental data of the interior of the vehicle during the fatigue assessment cycle, specifically the carbon dioxide content of the interior of the vehicle at each fatigue assessment time point; the driving behavior data of the driver during the fatigue assessment cycle, specifically the steering angle of the driver's steering wheel at each fatigue assessment time point; the carbon dioxide content of the interior of the vehicle at each fatigue assessment time point is compared with the predefined carbon dioxide reference content, and if the carbon dioxide content at a certain fatigue assessment time point is greater than the carbon dioxide reference content, then the fatigue assessment time point is recorded as a point where the carbon dioxide content exceeds the standard, so as to count the number of points where the carbon dioxide content exceeds the standard; The number of points at which the carbon dioxide content exceeds the standard is matched with the environmental impact factors corresponding to the predefined intervals of the number of points at which the carbon dioxide content exceeds the standard, so as to obtain the environmental impact factors; the steering wheel steering angle of the driver at each fatigue assessment time point is compared with the predefined preset intervals of the steering wheel steering angle; if the steering wheel steering angle at a certain fatigue assessment time point does not belong to the preset interval of the steering wheel steering angle, the fatigue assessment time point is recorded as a steering wheel steering angle exceeding standard point, so as to count the number of steering wheel steering angle exceeding standard points; the number of steering wheel steering angle exceeding standard points is matched with the driving behavior influencing factors corresponding to the predefined intervals of the number of steering wheel steering angle exceeding standard points, so as to obtain the driving behavior influencing factors.

[0015] As a further method, the driver's driving fatigue state is determined, and the specific determination process is: synchronously comparing the fatigue characteristic abnormality value of the driver's facial recognition image within the fatigue assessment cycle with the predefined fatigue characteristic abnormality preset value, comparing the environmental impact factor with the predefined environmental impact factor reference value, and comparing the driving behavior impact factor with the predefined driving behavior impact factor reference value to obtain a driving fatigue state comparison result; if the driving fatigue state comparison result shows that the fatigue characteristic abnormality value of the driver's facial recognition image within the fatigue assessment cycle is less than or equal to the fatigue characteristic abnormality preset value, the environmental impact factor is greater than or equal to the environmental impact factor reference value, and the driving behavior impact factor is greater than or equal to the driving behavior impact factor reference value, then the driver's driving fatigue state is determined. The driving fatigue state is the first fatigue driving state; if the driving fatigue state comparison result shows that the fatigue characteristic abnormal value of the driver's facial recognition image within the fatigue assessment period is less than or equal to the fatigue characteristic preset value, the environmental impact factor is greater than or equal to the environmental impact factor reference value, and the driving behavior impact factor is less than the driving behavior impact factor reference value, then the driver's driving fatigue state is determined to be the second fatigue driving state; if the driving fatigue state comparison result shows that the fatigue characteristic abnormal value of the driver's facial recognition image within the fatigue assessment period is less than or equal to the fatigue characteristic preset value, the environmental impact factor is less than the environmental impact factor reference value, and the driving behavior impact factor is greater than or equal to the driving behavior impact factor reference value, then the driver's driving fatigue state is determined to be the third fatigue driving state.

[0016] As a further method, the driver's driving fatigue state is warned, and the specific warning process is: warning is issued according to the driver's driving fatigue state, and the warning is specifically matched with an automatic intervention strategy or a warning prompt strategy; if the driver's driving fatigue state is a first fatigue driving state, the automatic intervention strategy is matched; if the driver's driving fatigue state is a second fatigue driving state or a third fatigue driving state, the warning prompt strategy is matched.

[0017] The second aspect of the present invention provides a device for a fatigue driving detection method, comprising: a vehicle-mounted camera, a vehicle-mounted terminal and a fatigue monitoring component; the vehicle-mounted camera is used to perform facial recognition on the driver; the vehicle-mounted terminal is used to analyze the facial recognition image, the environmental data inside the vehicle during the fatigue assessment cycle, and the driving behavior data of the driver during the fatigue assessment cycle, and obtain the fatigue characteristic abnormal values, environmental influencing factors and driving behavior influencing factors of the driver's facial recognition image during the fatigue assessment cycle; the fatigue monitoring component is used to connect to the vehicle-mounted terminal via wireless communication technology, receive the fatigue characteristic abnormal values, environmental influencing factors and driving behavior influencing factors of the driver's facial recognition image in the vehicle-mounted terminal during the fatigue assessment cycle, and warn the driver of the driving fatigue state.

[0018] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects:

[0019] (1) The present invention provides a fatigue driving detection method and device, determines the clarity value of a facial recognition image, analyzes the facial recognition image by a vehicle-mounted terminal, and evaluates the fatigue feature abnormality value of the driver's facial recognition image within a fatigue evaluation cycle, so as to timely detect signs of driver fatigue; matches environmental influencing factors and driving behavior influencing factors, and combines the fatigue feature abnormality value of the driver's facial recognition image within the fatigue evaluation cycle to warn the driver of the driving fatigue state. Once the fatigue driving state is detected, the fatigue driving detection system will immediately issue a warning and take preventive measures, thereby effectively avoiding traffic accidents caused by fatigue driving, thereby completing the driver's fatigue driving detection to optimize the driving experience and improve traffic efficiency.

[0020] (2) The present invention uses a vehicle-mounted camera to perform facial recognition on the driver, obtains facial recognition image data of the driver, and determines the clarity value of the facial recognition image. A clear image helps to extract facial features more accurately, and to a great extent avoids possible occlusion of the driver's face during the facial recognition process, thereby improving the accuracy of facial recognition. By screening clear images, the false recognition rate can be reduced and the reliability of facial recognition can be improved.

[0021] (3) The present invention can more comprehensively evaluate the driver's fatigue state by combining the fatigue feature abnormal values ​​of the driver's facial recognition image within the fatigue assessment cycle, environmental influencing factors, and driving behavior influencing factors. This comprehensive consideration avoids the one-sidedness of single-factor evaluation and improves the accuracy of detection. Single-factor evaluation may lead to false positives or false negatives. By comprehensively considering multiple factors, such errors can be reduced and the reliability of the fatigue driving detection system can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0023] Figure 1 The figure is a schematic flow chart of the method steps of the present invention.

[0024] Figure 2 This is the P80 standard schematic diagram. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0026] The fatigue driving detection system is one of the key technologies to ensure road traffic safety. As people pay more attention to driving safety, the importance of the fatigue driving detection system has become increasingly prominent. The embodiment of the present invention aims to monitor the driver's state, the vehicle's driving conditions, and the environmental quality inside the car, to conduct a comprehensive and continuous detection of the driver's driving state and issue warnings in time according to the driving state, so as to help the driver avoid potential traffic accidents. The embodiment of the present invention can be used in the process of driving for a long time or driving at night. By monitoring the driver's eye activities or the vehicle's deviation from the lane due to lack of concentration, or detecting the air density in the car in real time, the system can detect the driver's fatigue state in time and issue a warning, thereby effectively preventing traffic accidents caused by fatigue driving.

[0027] Reference Figure 1 As shown, the first aspect of the present invention provides a fatigue driving detection method, including: S1. performing facial recognition on the driver through a vehicle-mounted camera, obtaining facial recognition image data of the driver, determining the clarity value of the facial recognition image, and comparing it with a predefined clarity preset value to obtain a comparison result.

[0028] The above-mentioned vehicle-mounted camera is used for facial recognition of the driver. Specifically, the vehicle-mounted camera captures the driver's facial image, including the opening and closing of the eyes, and then compares and analyzes it with a pre-set state, thereby realizing the monitoring of the driver's state.

[0029] Specifically, the process of determining the clarity value of the facial recognition image is as follows:

[0030] The driver's facial recognition image data specifically includes the number of feature recognition points of the eye region recognition image, the contrast of the eye region recognition image at each pixel point, and the noise value of the eye region recognition image at each pixel point.

[0031] The above-mentioned driver's facial recognition image data can be extracted from the operation data report of the vehicle-mounted camera.

[0032] A first clarity value of the facial recognition image is determined based on the facial recognition image data of the driver.

[0033] The grayscale value of each pixel of the eye recognition image is obtained, wherein the grayscale value of each pixel of the eye recognition image is extracted by using an open source computer vision library. The open source computer vision library provides a wealth of image processing and computer vision algorithms. Standard deviation processing is performed to obtain a grayscale change index of the eye recognition image, and the grayscale change definition index predefined in the detection information library is comprehensively processed to obtain a second clarity value of the facial recognition image.

[0034] According to the first clarity value of the facial recognition image and the second clarity value of the facial recognition image, the clarity value of the facial recognition image is comprehensively determined in the following manner:

[0035]

[0036] in

[0037] In the formula, is the clarity value of the facial recognition image, which in this embodiment refers to the clarity of the face part in the image, which is an important indicator for measuring image quality.

[0038] is the first clarity value of the facial recognition image, which means that the first clarity value of the facial recognition image is obtained by comprehensive analysis of the number of feature recognition points of the eye area recognition image, the contrast of the eye area recognition image at each pixel, and the noise value of the eye area recognition image at each pixel.

[0039] is the second clarity value of the facial recognition image, indicating that the second clarity value of the facial recognition image is obtained by comprehensive processing of the grayscale change index and the grayscale change definition index of the eye recognition image.

[0040] It is a weight factor corresponding to the first clarity value predefined in the detection information library.

[0041] In a specific embodiment, the weight factor corresponding to the first clarity value can be directly obtained from the detection information library during use. The factor represents the numerical value of the influence of the first clarity value on the clarity value of the facial recognition image during the clarity determination of the facial recognition image. The corresponding relationship can be a pre-set mapping relationship. For example, the first clarity value and the weight factor corresponding to the first clarity value preset in the detection information library form a mapping set, and the real-time first clarity value is brought into the mapping set to obtain the weight factor corresponding to the first clarity value. The mapping relationship can be a one-to-one correspondence or a many-to-one relationship. In this embodiment, its value range is (0, 1).

[0042] It is a weight factor corresponding to the second clarity value predefined in the detection information library.

[0043] In a specific embodiment, the weight factor corresponding to the second clarity value can be directly obtained from the detection information library during use. The factor represents the numerical value of the influence of the second clarity value on the clarity value of the facial recognition image during the clarity determination of the facial recognition image. The corresponding relationship can be a pre-set mapping relationship. For example, the second clarity value and the weight factor corresponding to the second clarity value preset in the detection information library form a mapping set, and the real-time second clarity value is brought into the mapping set to obtain the weight factor corresponding to the second clarity value. The mapping relationship can be a one-to-one correspondence or a many-to-one relationship. In this embodiment, its value range is (0, 1).

[0044] The number of feature recognition points for the eye region recognition image refers to the number of points with unique local properties that can be detected in the processing task of the eye region recognition image.

[0045] The contrast of the eye region recognition image at the yth pixel point refers to the degree of difference in brightness corresponding to each pixel point in the eye region recognition image.

[0046] y is the number of each pixel, , Y is the total number of pixels.

[0047] The noise value of the yth pixel of the eye region recognition image refers to analyzing the signal of each pixel in the eye region recognition image to distinguish which pixel values ​​are abnormal values ​​caused by noise interference and which pixel values ​​represent the real signal of the image. Noise usually appears as unwanted, random, and interfering pixel values ​​in the image, which may come from various factors in the acquisition, transmission or processing of the image.

[0048] It is the grayscale change index of the eye recognition image, which refers to the degree of change of the grayscale value of the pixels in the image during the eye recognition process.

[0049] Defining an index for grayscale change refers to setting the minimum value of the grayscale change index within a reasonable range.

[0050] It is the impact factor corresponding to a single feature recognition point predefined in the detection information database.

[0051] In a specific embodiment, the influence factor corresponding to a single feature identification point can be directly obtained from the detection information library during use. The factor represents the numerical value of the influence of the number of feature identification points on the first clarity value during the first clarity determination process. The corresponding relationship can be a pre-set mapping relationship. For example, the number of feature identification points and the influence factors corresponding to the single feature identification point preset in the detection information library form a mapping set, and the real-time number of feature identification points is brought into the mapping set to obtain the influence factor corresponding to the single feature identification point. The mapping relationship can be a one-to-one correspondence or a many-to-one relationship. In this embodiment, its value range is (0, 1).

[0052] It is the impact factor corresponding to the contrast unit value predefined in the detection information library.

[0053] In a specific embodiment, the influence factor corresponding to the contrast unit value can be directly obtained from the detection information library during use. The factor represents the value of the influence of the contrast on the first clarity value in the process of determining the first clarity. The corresponding relationship can be a preset mapping relationship. For example, the contrast and the influence factor corresponding to the preset contrast unit value in the detection information library form a mapping set, and the real-time contrast is brought into the mapping set to obtain the influence factor corresponding to the contrast unit value. The mapping relationship can be a one-to-one correspondence or a many-to-one relationship. In this embodiment, its value range is (0, 1).

[0054] It is the impact factor corresponding to the noise value unit value predefined in the detection information database.

[0055] In a specific embodiment, the influence factor corresponding to the unit value of the noise value can be directly obtained from the detection information library during use. The factor represents the influence degree of the noise value on the first clarity value in the process of first clarity determination. The corresponding relationship can be a pre-set mapping relationship. For example, the noise value and the influence factor corresponding to the unit value of the noise value preset in the detection information library form a mapping set, and the real-time noise value is brought into the mapping set to obtain the influence factor corresponding to the unit value of the noise value. The mapping relationship can be a one-to-one correspondence or a many-to-one relationship. In this embodiment, its value range is (0, 1).

[0056] It is the influence factor corresponding to the grayscale change index difference unit value predefined in the detection information library, and e is a natural constant.

[0057] In a specific embodiment, the influence factor corresponding to the grayscale change index difference unit value can be directly obtained from the detection information library when in use. The factor represents the numerical value of the influence of the grayscale change index difference on the second clarity value in the process of the second clarity determination. The corresponding relationship can be a pre-set mapping relationship. For example, the grayscale change index difference and the influence factor corresponding to the grayscale change index difference unit value preset in the detection information library form a mapping set, and the real-time grayscale change index difference is brought into the mapping set to obtain the influence factor corresponding to the grayscale change index difference unit value. The mapping relationship can be a one-to-one correspondence or a many-to-one relationship. In this embodiment, its value range is (0, 1).

[0058] In this embodiment, image contrast refers to the brightness difference between different areas in the image. High contrast means that the brightness difference between the bright and dark parts of the image is large, which makes the features in the image more prominent, which is conducive to the detection of feature points. In high-contrast images, feature points, such as corners, edges, etc., are usually more obvious and therefore easier to detect, thereby increasing the clarity of the image. On the contrary, in low-contrast images, feature points may become blurred or difficult to identify, resulting in a reduction in the number of feature points, which affects the clarity of the image. Image noise refers to unwanted and random brightness changes in the image. Noise may interfere with the detection of feature points, resulting in a reduction in the number of feature points or false detection. In high-noise images, some feature points that do not originally exist may be mistakenly detected (i.e., false detection) due to the interference of noise, and some real feature points may become blurred or difficult to identify (i.e., missed detection) due to the interference of noise, which will also have a negative impact on the determination of image clarity. In summary, there is a close relationship between the number of feature recognition points, the contrast of each pixel point, and the noise value of each pixel point. In the image processing process, these factors need to be comprehensively considered to obtain an image with a larger clarity value.

[0059] In this embodiment, by performing standard deviation processing on the grayscale values ​​of each pixel of the image, the grayscale change of the image can be reflected. The larger the standard deviation is and the larger it is than the grayscale change definition index, the more drastic the grayscale change of the image is, which means the clearer the image is. Specifically, the larger the standard deviation is, the more dispersed the grayscale distribution of the image is, and the more drastic the grayscale change of the image is. This usually means that the image has a large contrast and a significant contrast, and can present more details and texture information, so the image will look clearer. On the contrary, if the standard deviation is small, the grayscale distribution of the image is concentrated, the grayscale changes smoothly, the contrast of the image is small, the contrast is not high, and the tone may appear to be single, and not much information can be seen.

[0060] Furthermore, the clarity value of the facial recognition image is compared with a predefined clarity preset value to obtain a comparison result. The specific comparison process is as follows:

[0061] The clarity value of the facial recognition image is compared with a preset clarity value predefined in the detection information library to obtain a comparison result, ie, a first clarity comparison result or a second clarity comparison result.

[0062] The first clarity comparison result is that the clarity value of the facial recognition image is greater than a preset clarity value.

[0063] The second clarity comparison result is that the clarity value of the facial recognition image is less than or equal to a preset clarity value.

[0064] S2. Based on the comparison result, the vehicle terminal analyzes the facial recognition image and evaluates the fatigue feature abnormality of the driver's facial recognition image within the fatigue assessment cycle.

[0065] Specifically, based on the comparison result, the vehicle terminal analyzes the facial recognition image, and the specific analysis process is:

[0066] If the comparison result shows that the clarity value of the facial recognition image is greater than the preset clarity value, the vehicle-mounted terminal directly analyzes the facial recognition image.

[0067] The above-mentioned vehicle-mounted terminal analyzes the facial recognition image, specifically, the vehicle-mounted terminal extracts features from the facial recognition image. This process mainly extracts key features that can represent the face by analyzing the texture, feature points and other information of the face. Feature point positioning is to locate the key areas of the face, such as eyes, nose, mouth and other facial features and contour points, based on a given face image. After that, the driver's state is judged by analyzing facial features, such as fatigue driving. If the driver is detected to be in a fatigued state, the vehicle-mounted terminal will issue a warning in time to improve driving safety.

[0068] If the comparison result shows that the clarity value of the facial recognition image is less than or equal to the preset clarity value, the clarity of the facial recognition image is optimized so that the clarity value of the facial recognition image after clarity optimization is greater than the preset clarity value, and finally the vehicle-mounted terminal analyzes the facial recognition image.

[0069] The above-mentioned clarity optimization of facial recognition images is specifically image enhancement processing of the original facial recognition images. It is achieved by "adjusting" the grayscale values ​​using a cumulative function through histogram equalization to enhance contrast. The "central idea" of the histogram equalization processing is to change the grayscale histogram of the original image from a relatively concentrated grayscale interval to a uniform distribution within the entire grayscale range. By improving the contrast of the image, the grayscale value distribution is made more uniform, thereby increasing the grayscale standard deviation. At the same time, a low-pass filter is used to remove image noise, thereby increasing the clarity value of the image.

[0070] It should be explained that the above-mentioned clarity optimization of facial recognition images is aimed at improving the visual quality of images and making them clearer through a series of technical means, such as denoising, contrast enhancement, sharpening, etc. These optimization measures can enhance the detail performance of the image and make facial features more obvious, thereby improving the accuracy and efficiency of facial recognition. Removing noise from the image can reduce the interference information in the image and make the facial features more prominent. The denoising image is smoother and the details are clearer, thereby improving the clarity value. Enhancing the contrast of the image can make facial features more obvious, especially in low light or strong light conditions. The contrast-enhanced image is visually clearer and the clarity value will be increased accordingly, thereby ensuring that the clarity value of the facial recognition image is greater than the preset clarity value.

[0071] Furthermore, the fatigue feature abnormal value of the driver's facial recognition image during the fatigue assessment cycle is evaluated, and the specific analysis process is as follows:

[0072] The facial recognition image is analyzed by the vehicle-mounted terminal to obtain data of the facial recognition image analysis performed by the vehicle-mounted terminal during the fatigue assessment cycle, specifically including the number of times the driver blinks during the fatigue assessment cycle and the maximum angle value of the driver's head tilt during the fatigue assessment cycle.

[0073] The data of facial recognition image analysis performed by the above-mentioned vehicle-mounted terminal during the fatigue assessment cycle can be extracted from the operation report of the vehicle-mounted terminal, where the fatigue assessment cycle is obtained by a comprehensive analysis by the management personnel based on factors such as the monitoring environment, terminal status, and driver status.

[0074] The number of times the driver blinks during the fatigue assessment period is compared with the duration of the fatigue assessment period to obtain the blink frequency of the driver during the fatigue assessment period.

[0075] According to the driver's blinking frequency during the fatigue assessment period, the driver's maximum head tilt angle during the fatigue assessment period, and the number of feature recognition points of the eye area recognition image, a comprehensive analysis is performed and recorded as the fatigue feature abnormal value of the driver's facial recognition image during the fatigue assessment period. The specific analysis method is as follows:

[0076]

[0077] In the formula, It is the fatigue feature abnormal value of the driver's facial recognition image during the fatigue assessment cycle. In this embodiment, it means the index detected in the process of continuous monitoring of the driver's facial features that is significantly deviated or abnormal compared with the normal driving state.

[0078] The blinking frequency of the driver during the fatigue assessment cycle refers to the number of times the driver blinks during the fatigue assessment cycle. Blinking refers to the process of opening-closing-opening the eyes.

[0079] It should be explained that the main indicator for judging whether the driver is fatigued in the embodiment of the present invention is the eye feature. The specific implementation process is: the judgment is completed through the vehicle-mounted terminal, and the core algorithm adopted is based on PERCLOS (the proportion of the time occupied by the degree of eye closure per unit time), and is improved according to the actual application situation. Specifically, the time periods of different eye opening and closing states of the original algorithm are improved to capture the blinking frequency.

[0080] Among them, the PERCLOS algorithm is a method for evaluating fatigue status by calculating the percentage of eyelid closure time in the total observation time. Its principle is to reflect eye characteristics through eye closure time, and it has several detection standards for different eye closure states. Among them, the P80 standard (the percentage of time when the eye closure degree exceeds 80% per unit time in the total time) has the best correlation with the driver's fatigue level. Taking P80 as an example, the algorithm principle is as follows Figure 2 As shown, this is the P80 standard principle diagram, where the horizontal axis is the detection time point t, and the vertical axis is the eye opening z.

[0081] Depend on Figure 2 It can be seen that when the time point is And the time point is When the eye opening and closing state is at 80% pupil state, when the time point is And the time point is When the eyes are open and closed, they are in a 20% pupil state. Therefore, the PERCLOS algorithm needs to calculate the time periods of different eye opening and closing states, so it has high requirements on the performance of the processor. In the embodiment of the present invention, the PERCLOS algorithm is improved and the blinking frequency is used to replace the closing time indicator in the original algorithm. The improved algorithm sets the threshold of eye opening and closing, and then captures each blinking action of the driver for accumulation. Compared with the original algorithm, the advantage is that there is no need to measure the exact time points of each different degree of closure during each blink, and the real-time requirements are lower.

[0082] It is the maximum tilt angle of the driver's head during the fatigue assessment cycle, which refers to the maximum tilt angle that the driver's head can reach relative to the normal driving posture during the fatigue assessment cycle.

[0083] The number of feature points in the image for eye region recognition.

[0084] It is the weight factor corresponding to the blink frequency predefined in the detection information library.

[0085] In a specific embodiment, the weight factor corresponding to the blinking frequency can be directly obtained from the detection information library during use. The factor represents the numerical value of the influence of the blinking frequency on the fatigue feature abnormality value of the driver's facial recognition image within the fatigue assessment period during the process of determining the fatigue feature abnormality of the driver's facial recognition image within the fatigue assessment period. The corresponding relationship can be a pre-set mapping relationship. For example, the blinking frequency and the weight factor corresponding to the blinking frequency preset in the detection information library form a mapping set, and the real-time blinking frequency is brought into the mapping set to obtain the weight factor corresponding to the blinking frequency. The mapping relationship can be a one-to-one correspondence or a many-to-one relationship. In this embodiment, its value range is (0, 1).

[0086] It is the weight factor corresponding to the maximum head tilt angle predefined in the detection information library.

[0087] In a specific embodiment, the weight factor corresponding to the maximum head tilt angle can be directly obtained from the detection information library during use. The factor represents the numerical value of the influence of the maximum head tilt angle on the fatigue feature abnormality value of the driver's facial recognition image within the fatigue assessment period during the fatigue feature abnormality judgment of the driver's facial recognition image within the fatigue assessment period. The corresponding relationship can be a pre-set mapping relationship. For example, the maximum head tilt angle and the weight factor corresponding to the maximum head tilt angle preset in the detection information library form a mapping set, and the real-time maximum head tilt angle is brought into the mapping set to obtain the weight factor corresponding to the maximum head tilt angle. The mapping relationship can be a one-to-one correspondence or a many-to-one relationship. In this embodiment, its value range is (0, 1).

[0088] It is the impact factor corresponding to a single feature recognition point predefined in the detection information library, and e is a natural constant.

[0089] In this embodiment, this abnormal value is usually used to determine whether the driver is in a fatigue driving state, wherein the greater the blinking frequency within a reasonable range and the smaller the maximum head tilt angle, the driver is in a normal driving state, then the greater the fatigue characteristic abnormal value, the driver is not in a fatigue driving state, on the contrary, the smaller the blinking frequency and the larger the maximum head tilt angle, the driver may be in a fatigue driving state, then the smaller the fatigue characteristic abnormal value, the more the driver belongs to a fatigue driving state; therefore, there is a certain correlation between the blinking frequency and the maximum head tilt angle. When the driver feels tired, they may experience a decrease in blinking frequency and an increase in head tilt angle. This is because fatigue affects the driver's attention and body control ability, making it difficult for them to maintain a normal driving posture and eye activity; at the same time, the increase in the number of eye feature recognition points means that more eye detail information can be captured. If the number of feature recognition points is insufficient, then the vehicle-mounted terminal may find it difficult to accurately capture subtle changes in the eye area, which will result in a smaller captured blinking frequency, and the measurement of the head tilt angle may also become smaller. When the blinking frequency and the head tilt angle are measured inaccurately, this will cause the accuracy of the fatigue assessment system to decrease, and thus it may not be possible to provide timely and effective feedback on the driver's true state.

[0090] S3. Obtain the environmental data of the vehicle's interior during the fatigue assessment cycle and the driver's driving behavior data during the fatigue assessment cycle, so as to match the environmental influencing factors and the driving behavior influencing factors, and combine the fatigue characteristic abnormal values ​​of the driver's facial recognition image during the fatigue assessment cycle. The fatigue monitoring component comprehensively determines the driver's driving fatigue state and issues an early warning of the driver's driving fatigue state, so as to complete the driver's fatigue driving detection.

[0091] Specifically, the environmental influencing factors and driving behavior influencing factors are matched, and the specific matching process is:

[0092] The environmental data inside the vehicle during the fatigue assessment cycle is specifically the carbon dioxide content inside the vehicle at each fatigue assessment time point, wherein the carbon dioxide content can be detected by a carbon dioxide sensor belonging to the vehicle-mounted terminal.

[0093] The above fatigue assessment time points are obtained by dividing the fatigue assessment cycle.

[0094] The driving behavior data of the driver during the fatigue assessment cycle is specifically the steering wheel steering angle of the driver at each fatigue assessment time point, wherein the steering wheel steering angle can be detected by a steering angle sensor belonging to the vehicle-mounted terminal.

[0095] The carbon dioxide content inside the vehicle at each fatigue assessment time point is compared with the predefined carbon dioxide reference content in the detection information library. If the carbon dioxide content at a certain fatigue assessment time point is greater than the carbon dioxide reference content, the fatigue assessment time point is recorded as a point where the carbon dioxide content exceeds the standard, and the number of points where the carbon dioxide content exceeds the standard is counted.

[0096] The number of points where the carbon dioxide content exceeds the standard is matched with the environmental impact factors corresponding to each predefined interval of the number of points where the carbon dioxide content exceeds the standard in the detection information database. The specific matching method is: obtain a mapping set of the number of points where the carbon dioxide content exceeds the standard and the environmental impact factors from the detection information database, first determine the interval to which the number of points where the carbon dioxide content exceeds the standard belongs, and assign the environmental impact factor corresponding to the interval to the number of points where the carbon dioxide content exceeds the standard, so as to obtain the environmental impact factor.

[0097] The driver's steering wheel angle at each fatigue assessment time point is compared with the predefined steering wheel steering angle preset interval in the detection information library. If the steering wheel steering angle at a fatigue assessment time point does not belong to the preset steering wheel steering angle interval, the fatigue assessment time point is recorded as a steering wheel steering angle exceeding the standard point, and the number of steering wheel steering angle exceeding the standard points is counted.

[0098] In a specific embodiment, the preset range of the steering wheel steering angle may be determined by the vehicle terminal according to the driving direction of the vehicle navigation. If the steering wheel steering angle is not within the preset range, it means that the driver may be in a fatigue driving state.

[0099] The number of steering wheel steering angle exceeding standard points is matched with the driving behavior influencing factors corresponding to each steering wheel steering angle exceeding standard point number interval predefined in the detection information library. The specific matching method is: obtain a mapping set of the number of steering wheel steering angle exceeding standard points and the driving behavior influencing factors from the detection information library, first determine the interval to which the number of steering wheel steering angle exceeding standard points belongs, and assign the driving behavior influencing factor corresponding to the interval to the number of steering wheel steering angle exceeding standard points, so as to obtain the driving behavior influencing factor.

[0100] Furthermore, the driver's driving fatigue state is determined by the following specific determination process:

[0101] The fatigue feature abnormality value of the driver's facial recognition image within the fatigue assessment cycle is synchronously compared with the fatigue feature abnormality preset value predefined in the detection information library, the environmental influencing factor is compared with the environmental influencing factor reference value predefined in the detection information library, and the driving behavior influencing factor is compared with the driving behavior influencing factor reference value predefined in the detection information library to obtain a driving fatigue status comparison result.

[0102] If the driving fatigue status comparison result shows that the fatigue characteristic abnormality value of the driver's facial recognition image during the fatigue assessment period is less than or equal to the fatigue characteristic abnormality preset value, the environmental impact factor is greater than or equal to the environmental impact factor reference value, and the driving behavior impact factor is greater than or equal to the driving behavior impact factor reference value, then the driver's driving fatigue status is determined to be the first fatigue driving state.

[0103] The first fatigue driving state mentioned above is specifically that the driver's facial features show a fatigue state, and the in-car environment and driving behavior also increase the possibility that the driver is in a fatigue state, so it may be necessary to intervene in the driver's driving state.

[0104] If the driving fatigue status comparison result shows that the fatigue characteristic abnormal value of the driver's facial recognition image during the fatigue assessment period is less than or equal to the fatigue characteristic preset value, the environmental impact factor is greater than or equal to the environmental impact factor reference value, and the driving behavior impact factor is less than the driving behavior impact factor reference value, then the driver's driving fatigue status is determined to be the second fatigue driving state.

[0105] The above-mentioned second fatigue driving state, specifically, the driver's facial features show a fatigue state, and the in-car environment also increases the possibility that the driver is in a fatigue state, while the driving behavior is in a normal state, so it may be necessary to warn the driver of the driving state.

[0106] If the driving fatigue status comparison result shows that the fatigue characteristic abnormal value of the driver's facial recognition image during the fatigue assessment period is less than or equal to the fatigue characteristic preset value, the environmental impact factor is less than the environmental impact factor reference value, and the driving behavior impact factor is greater than or equal to the driving behavior impact factor reference value, then the driver's driving fatigue status is determined to be the third fatigue driving state.

[0107] The third fatigue driving state mentioned above is specifically that the driver's facial features show a fatigue state, and the driving behavior also increases the possibility that the driver is in a fatigue state, while the interior environment of the car is in a normal state. Therefore, it may be necessary to warn the driver of the driving state.

[0108] In this embodiment, if the comparison result of the driving fatigue state shows that the fatigue feature abnormality value of the driver's facial recognition image during the fatigue assessment period is less than or equal to the fatigue feature compliance preset value, and the environmental impact factor is less than the environmental impact factor reference value and the driving behavior impact factor is less than the driving behavior impact factor reference value, it may mean that although the driver's facial features show certain signs of fatigue, these signs are not significant, and the environment and driving behavior have little effect on the driver's fatigue state, which helps the driver to stay relatively awake and focused, thereby reducing fatigue, so that the driver is in a relatively safe and controllable driving state as a whole. The driver's fatigue state can be continuously monitored and managed, and the driver can be reminded of fatigue using an on-board display screen or other prompting methods.

[0109] In this embodiment, if the driving fatigue status comparison result shows that the fatigue feature abnormal value of the driver's facial recognition image within the fatigue assessment cycle is greater than the fatigue feature compliance preset value, it means that when the fatigue feature abnormal value is greater than the fatigue feature compliance preset value, regardless of the environmental influencing factors and the numerical relationship between the driving behavior influencing factors and their factor reference values, the driver does not show a fatigue driving state, and the driver's status can be continuously monitored subsequently to ensure the driver's driving safety.

[0110] Specifically, the warning of the driver's driving fatigue state is carried out in the following specific process:

[0111] A warning is issued according to the driver's driving fatigue state, wherein the warning is specifically matched with an automatic intervention strategy or a warning prompt strategy.

[0112] If the driver's driving fatigue state is the first fatigue driving state, an automatic intervention strategy is matched.

[0113] If the driver's driving fatigue state is the second fatigue driving state or the third fatigue driving state, a warning prompt strategy is matched.

[0114] The specific matching process of the above matching is: obtain a mapping set of driving fatigue status and warning strategies from the detection information library, first determine the category to which the driver's driving fatigue status belongs, and assign the warning strategy corresponding to the category to the corresponding driving fatigue status.

[0115] Furthermore, the automatic intervention strategy or early warning prompt strategy is specifically:

[0116] The automatic intervention strategy may be that the fatigue monitoring component automatically slows down the vehicle, pulls over to the side of the road, and flashes the parking lights.

[0117] The early warning prompt strategy may be that the fatigue monitoring component prompts the driver to open windows for ventilation or prompts the driver to deviate from the path.

[0118] The above-mentioned fatigue monitoring component warns the driver of the driving fatigue state. Specifically, the fatigue monitoring component compares the fatigue characteristic abnormal values, environmental influencing factors and driving behavior influencing factors of the driver's facial recognition image received from the on-board terminal during the fatigue assessment cycle with the corresponding parameters under the predefined fatigue driving state. If the fatigue monitoring component detects that the driver is in a fatigue state, corresponding measures are taken on the driver's driving process according to the corresponding strategy.

[0119] A second aspect of the present invention provides a device for a fatigue driving detection method, comprising: a vehicle-mounted camera, a vehicle-mounted terminal and a fatigue monitoring component.

[0120] The vehicle-mounted camera is used to perform facial recognition on the driver.

[0121] The vehicle-mounted terminal is used to analyze the facial recognition image, the environmental data inside the vehicle during the fatigue assessment cycle, and the driving behavior data of the driver during the fatigue assessment cycle to obtain the fatigue feature abnormal value, environmental influencing factor, and driving behavior influencing factor of the driver's facial recognition image during the fatigue assessment cycle.

[0122] The fatigue monitoring component is used to connect to the vehicle terminal through wireless communication technology, receive fatigue feature abnormal values, environmental influencing factors and driving behavior influencing factors of the driver's facial recognition image in the vehicle terminal during the fatigue assessment cycle, and issue an early warning of the driver's driving fatigue state.

[0123] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the present invention, they should all fall within the protection scope of the present invention.

Claims

1. A method for detecting fatigue driving, characterized in that: include: S1. Perform facial recognition on the driver through the vehicle camera, obtain the driver's facial recognition image data, determine the clarity value of the facial recognition image, and compare it with the predefined clarity preset value to obtain a comparison result; S2. Based on the comparison results, the vehicle terminal analyzes the facial recognition image and evaluates the fatigue feature abnormality of the driver's facial recognition image within the fatigue assessment cycle; S3. Obtain the environmental data of the vehicle and the driving behavior data of the driver during the fatigue assessment cycle, so as to match the environmental influencing factors and the driving behavior influencing factors, and combine the fatigue characteristic abnormal values ​​of the driver's facial recognition image during the fatigue assessment cycle, so that the fatigue monitoring component comprehensively determines the driver's driving fatigue state, and issues an early warning of the driver's driving fatigue state, so as to complete the driver's fatigue driving detection; The environmental influencing factors and driving behavior influencing factors are matched, and the specific matching process is: The environmental data of the vehicle interior during the fatigue assessment cycle, specifically the carbon dioxide content of the vehicle interior at each fatigue assessment time point; The driving behavior data of the driver during the fatigue assessment period, specifically the steering wheel steering angle of the driver at each fatigue assessment time point; The carbon dioxide content inside the vehicle at each fatigue assessment time point is compared with the predefined carbon dioxide reference content. If the carbon dioxide content at a fatigue assessment time point is greater than the carbon dioxide reference content, the fatigue assessment time point is recorded as a carbon dioxide content exceeding standard point, and the number of carbon dioxide content exceeding standard points is counted; Matching the number of points where the carbon dioxide content exceeds the standard with the environmental impact factors corresponding to the predefined ranges of the number of points where the carbon dioxide content exceeds the standard, thereby obtaining the environmental impact factors; The driver's steering wheel angle at each fatigue assessment time point is compared with the predefined steering wheel angle preset interval. If the steering wheel angle at a fatigue assessment time point does not belong to the steering wheel angle preset interval, the fatigue assessment time point is recorded as a steering wheel angle exceeding the standard point, thereby counting the number of steering wheel steering angle exceeding the standard point; The number of steering wheel steering angle exceeding the standard point is matched with the driving behavior influencing factors corresponding to the predefined steering wheel steering angle exceeding the standard point number intervals, so as to obtain the driving behavior influencing factors.

2. A fatigue driving detection method according to claim 1, characterized in that: The specific determination process of determining the clarity value of the facial recognition image is as follows: The driver's facial recognition image data specifically includes the number of feature recognition points of the eye region recognition image, the contrast of the eye region recognition image at each pixel point, and the noise value of the eye region recognition image at each pixel point; Thus, the number of feature recognition points of the eye region recognition image, the contrast of the eye region recognition image at each pixel point, and the noise value of the eye region recognition image at each pixel point are comprehensively processed to determine a first clarity value of the facial recognition image; Obtaining the grayscale value of each pixel of the eye recognition image, performing standard deviation processing to obtain a grayscale change index of the eye recognition image, and performing comprehensive processing with a predefined grayscale change definition index to obtain a second clarity value of the face recognition image; According to the first clarity value of the facial recognition image and the second clarity value of the facial recognition image, the clarity value of the facial recognition image is comprehensively determined in the following manner: in In the formula, is the clarity value of the facial recognition image, is the first clarity value of the face recognition image, is the second clarity value of the facial recognition image, is the weight factor corresponding to the first clarity value predefined in the detection information database, is a weight factor corresponding to the second clarity value predefined in the detection information database, The number of feature points for eye region recognition images, The contrast of the eye area recognition image at the yth pixel, where y is the number of each pixel. , Y is the total number of pixels, The noise value of the eye area recognition image at the y-th pixel, It is the grayscale change index of the eye recognition image. Define indicators for grayscale changes, To detect the impact factor corresponding to a single feature recognition point predefined in the information database, To detect the impact factor corresponding to the contrast unit value predefined in the information database, To detect the impact factor corresponding to the noise value unit value predefined in the information database, It is the influence factor corresponding to the grayscale change index difference unit value predefined in the detection information library, and e is a natural constant.

3. A fatigue driving detection method according to claim 2, characterized in that: The definition value of the facial recognition image is compared with a predefined definition preset value to obtain a comparison result. The specific comparison process is as follows: Comparing the clarity value of the facial recognition image with a predefined clarity preset value to obtain a comparison result, namely a first clarity comparison result or a second clarity comparison result; The first clarity comparison result is that the clarity value of the facial recognition image is greater than a preset clarity value; The second clarity comparison result is that the clarity value of the facial recognition image is less than or equal to a preset clarity value.

4. A fatigue driving detection method according to claim 3, characterized in that: Based on the comparison result, the vehicle terminal analyzes the facial recognition image, and the specific analysis process is as follows: If the comparison result shows that the clarity value of the facial recognition image is greater than the preset clarity value, the vehicle-mounted terminal directly analyzes the facial recognition image; If the comparison result shows that the clarity value of the facial recognition image is less than or equal to the preset clarity value, the clarity of the facial recognition image is optimized so that the clarity value of the facial recognition image after clarity optimization is greater than the preset clarity value, and finally the vehicle-mounted terminal analyzes the facial recognition image.

5. A fatigue driving detection method according to claim 1, characterized in that: The specific analysis process of evaluating the fatigue feature abnormal value of the driver's facial recognition image within the fatigue evaluation cycle is as follows: Analyze the facial recognition image through the vehicle terminal to obtain data of the facial recognition image analyzed by the vehicle terminal during the fatigue assessment cycle, specifically including the number of blinks of the driver during the fatigue assessment cycle and the maximum angle value of the driver's head tilt during the fatigue assessment cycle; The number of blinks of the driver during the fatigue assessment period is processed by ratio with the duration of the fatigue assessment period to obtain the blink frequency of the driver during the fatigue assessment period; Based on the driver's blinking frequency during the fatigue assessment period, the driver's maximum head tilt angle during the fatigue assessment period, and the number of feature recognition points of the eye area recognition image, a comprehensive analysis is performed to obtain and record the fatigue feature abnormal value of the driver's facial recognition image during the fatigue assessment period.

6. A fatigue driving detection method according to claim 1, characterized in that: The specific process of determining the driver's driving fatigue state is as follows: Synchronously comparing the fatigue characteristic abnormality value of the driver's facial recognition image within the fatigue assessment cycle with a predefined fatigue characteristic abnormality preset value, comparing the environmental influencing factor with a predefined environmental influencing factor reference value, and comparing the driving behavior influencing factor with a predefined driving behavior influencing factor reference value, to obtain a driving fatigue state comparison result; If the driving fatigue state comparison result shows that the fatigue characteristic abnormality value of the driver's facial recognition image within the fatigue assessment period is less than or equal to the fatigue characteristic abnormality preset value, the environmental influence factor is greater than or equal to the environmental influence factor reference value, and the driving behavior influence factor is greater than or equal to the driving behavior influence factor reference value, then the driver's driving fatigue state is determined to be the first fatigue driving state; If the driving fatigue state comparison result shows that the fatigue feature abnormality value of the driver's facial recognition image within the fatigue assessment period is less than or equal to the fatigue feature compliance preset value, the environmental impact factor is greater than or equal to the environmental impact factor reference value, and the driving behavior impact factor is less than the driving behavior impact factor reference value, then the driver's driving fatigue state is determined to be the second fatigue driving state; If the driving fatigue status comparison result shows that the fatigue characteristic abnormal value of the driver's facial recognition image during the fatigue assessment period is less than or equal to the fatigue characteristic preset value, the environmental impact factor is less than the environmental impact factor reference value, and the driving behavior impact factor is greater than or equal to the driving behavior impact factor reference value, then the driver's driving fatigue status is determined to be the third fatigue driving state.

7. A fatigue driving detection method according to claim 6, characterized in that: The specific warning process of the driver's driving fatigue state is as follows: Providing a warning according to the driver's driving fatigue state, wherein the warning specifically includes matching an automatic intervention strategy or a warning prompt strategy; If the driver's driving fatigue state is the first fatigue driving state, an automatic intervention strategy is matched; If the driver's driving fatigue state is the second fatigue driving state or the third fatigue driving state, a warning prompt strategy is matched.

8. A fatigue driving detection method according to claim 7, characterized in that: The automatic intervention strategy or early warning prompt strategy is specifically: The automatic intervention strategy may be that the fatigue monitoring component automatically slows down the vehicle, pulls over to the side of the road, and flashes the parking light; The early warning prompt strategy may be that the fatigue monitoring component prompts the driver to open windows for ventilation or prompts the driver to deviate from the path.

9. A device using a fatigue driving detection method as claimed in any one of claims 1 to 8, characterized in that: include: On-board cameras, on-board terminals and fatigue monitoring components; The vehicle-mounted camera is used to perform facial recognition on the driver; The vehicle-mounted terminal is used to analyze the facial recognition image, the environmental data inside the vehicle during the fatigue assessment period, and the driving behavior data of the driver during the fatigue assessment period to obtain the fatigue characteristic abnormal value, environmental influencing factor, and driving behavior influencing factor of the driver's facial recognition image during the fatigue assessment period; The fatigue monitoring component is used to connect to the vehicle terminal through wireless communication technology, receive fatigue feature abnormal values, environmental influencing factors and driving behavior influencing factors of the driver's facial recognition image in the vehicle terminal during the fatigue assessment cycle, and issue an early warning of the driver's driving fatigue state.

Images