User driving behavior detection method, system and storage medium
Through image processing and convolutional neural network model detection, combined with physiological data and behavioral scores, the problem of insufficient accuracy of drivers' long-term driving behavior detection in the existing technology is solved, and accurate detection of user driving behavior and risk level scores are realized. Improved driving safety.
Patent Information
- Application Number
- CN202411119912.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-15
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-08-15
AI Technical Summary
Existing driving behavior detection technology cannot effectively detect physical changes caused by drivers during long driving due to physiological and habit differences, resulting in a high probability of error contact and monitoring accuracy cannot meet the needs of long-term driving.
By acquiring the user images in the cab for image processing, determining the user's current status, establishing the initial state and fluctuation threshold box, periodically collecting images for comparison, using the convolutional neural network model to detect user behavior, combining user physiological data and behavior scores, and judging Whether there is dangerous driving behavior.
Customized and accurate detection of user driving behavior is realized, detection accuracy is improved through continuous comparison, and the degree of danger is evaluated based on behavior scores to ensure driving safety.
Smart Images

Figure CN118658148B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of image processing technology, and in particular to a user driving behavior detection method, system and storage medium. Background Art
[0002] As vehicles become more and more popular, it is becoming increasingly important to detect driver behavior. For example, if a driver makes a phone call while driving, it will distract the driver and bring potential safety hazards. In addition, if the driver's hands leave the steering wheel or the driver drives fatigued, it may lead to traffic accidents.
[0003] The current related technology is to install a monitoring device in the cockpit. When the driver is detected to have the above behaviors, the car will issue a prompt to remind the driver to stop the abnormal behavior so as to regulate the driver's driving behavior.
[0004] However, due to differences in the physiology and habits of drivers, each user's driving posture in the cockpit will be different to a certain extent, and the current monitoring device can only perform simple detection. For example, a slight deviation of the user's body will not affect the driving of the user at all, nor will it interfere with the driving, resulting in a high probability of false triggering of the monitoring device. In addition, the monitoring rules of the existing monitoring device are fixed. When the user drives for a long time, the user's movement amount will be larger over time (the driver's position on the seat will move, and the body changes will be correspondingly greater), which results in the monitoring accuracy being unable to meet the needs of long-term driving. Summary of the invention
[0005] 1. Technical issues to be resolved
[0006] In view of the deficiencies in the prior art, the present invention provides a user driving behavior detection method, system and storage medium to solve the problems raised in the above background technology.
[0007] (II) Technical solution
[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: a user driving behavior detection method, comprising:
[0009] Acquire a user image in the cab and perform image processing on the user image;
[0010] Determine the user's current state based on the processed user image, judge whether the user's current state is a dangerous behavior, and if not, determine it as an initial state. If so, remind the user to adjust to a normal driving state, and establish an initial state based on the adjusted state;
[0011] Establishing a fluctuation threshold frame based on the processing result of the image of the user's initial state;
[0012] The user's driving images are collected periodically, processed, and compared with the fluctuation threshold frame based on the processed user driving images to determine whether the user has dangerous driving behavior.
[0013] As a preferred embodiment of this embodiment, the acquiring of the user image in the cab and performing image processing on the user image comprises:
[0014] Collect the user's current state image through the electronic device to obtain the relevant contour and feature point positions of the user's appearance;
[0015] Obtain a binary mask of the ROI of the head, torso and arm regions of the user image to obtain a first image;
[0016] Based on the binary mask and edge curve of the region of interest of the user's head, torso and arm in the first image, a second image is obtained, where the second image is a ROI image of the region of interest of the user's head, torso and arm of the color channel image of the first image, where the color channel is a color channel that can reflect the user's behavior;
[0017] Based on the binary mask of the second image, the connected domain of the user behavior is extracted, and the circumscribed rectangular frame corresponding to each connected domain is obtained;
[0018] Based on the circumscribed rectangular frame and the user image, image blocks at corresponding positions of the two are obtained.
[0019] As a preferred embodiment of this embodiment, determining the current state of the user based on the processed user image and judging whether the current state of the user belongs to dangerous behavior includes:
[0020] Detecting the image blocks through a convolutional neural network model to determine the behavior type of the image blocks in the user's area of interest, the behavior type including the eye state, the torso position, and the arm position, and determining the user's physiological state when determining the eye state;
[0021] Analyze the user's current behavior status based on the user's behavior type.
[0022] As a preferred embodiment of the present invention, the processing result based on the user's initial state graph to establish a fluctuation threshold frame includes:
[0023] Determine the activity range of the user's area of interest based on the behavior type of the image block of the user's area of interest and the user's physiological data;
[0024] Determine the scope of the user's dangerous behavior by drawing the edge curve of the activity range of the user's area of interest;
[0025] The fluctuation threshold box is determined based on the activity range of the user's area of interest and the range of the user's dangerous behavior.
[0026] As a preferred embodiment of the present invention, the periodic acquisition of the user's driving image is specifically to acquire the user's current image at a fixed interval after the user's initial state image is determined, and the interval is specifically any one of 15s, 30s or 45s;
[0027] The periodically collected driving images are processed in the same manner as the user images.
[0028] As a preferred embodiment of this embodiment, comparing the processed user driving image with the fluctuation threshold frame to determine whether the user has dangerous driving behavior includes:
[0029] Obtain edge images of the user's head, torso, and arm regions in the user's driving image, and compare the user's head, torso, and arms with the fluctuation threshold frame;
[0030] The behavior score of the user's behavior is determined based on the comparison results of the edge images of the user's head, torso and arms with the fluctuation threshold box.
[0031] As a preferred embodiment of this embodiment, the behavior score of the user behavior determined based on the comparison result of the edge image of the user's head, torso and arms with the fluctuation threshold frame includes:
[0032] according to , determining the behavior score;
[0033] in Score the behavior. is the adjustment coefficient, which is related to the proportion of the user's head, torso and arms in the behavior score, M is the edge image feature data of the user's head, and M max is the maximum edge image feature data of the user's head, N is the edge image feature data of the user's torso, and N max is the maximum edge image feature data of the user's torso, J is the edge image feature data of the user's arm, and J max It is the maximum edge image feature data of the user's arm.
[0034] As a preferred embodiment of this invention, a user driving behavior detection system is used to implement the above-mentioned user driving behavior detection method, including:
[0035] An acquisition unit, used for acquiring an image of a user in the cab through an electronic device;
[0036] The image processing unit is used to obtain the relevant contours and feature point positions of the user's appearance, obtain the binary mask of the ROI of the head, torso and arm areas of the user image to obtain the first image, obtain the second image based on the binary mask and edge curve of the user's head, torso and arm areas of interest in the first image, the second image is the user's head, torso and arm areas ROI image of the color channel image of the first image, the color channel is a color channel that can reflect the user's behavior, extract the connected domain of the user's behavior based on the binary mask of the second image, and obtain the circumscribed rectangular frame corresponding to each connected domain, and obtain the image blocks at the corresponding positions of the two based on the circumscribed rectangular frame and the user image; and is also used to determine the activity range of the user's area of interest based on the behavior type of the image block of the user's area of interest and the user's physiological data, determine the range of the user's dangerous behavior by drawing the edge curve of the activity range of the user's area of interest, and determine the fluctuation threshold frame based on the activity range of the user's area of interest and the range of the user's dangerous behavior.
[0037] A comparison and analysis unit, used to compare the processed user driving image with the fluctuation threshold frame to determine whether the user has dangerous driving behavior;
[0038] The memory is used to store the user image data and the processed user image data.
[0039] As a preference of this embodiment, the comparison and analysis process of the comparison and analysis unit includes:
[0040] Obtain edge images of the user's head, torso, and arm regions in the user's driving image, and compare the user's head, torso, and arms with the fluctuation threshold frame;
[0041] Determine a behavior score of the user's behavior based on the comparison result of the edge images of the user's head, torso and arms with the fluctuation threshold box;
[0042] Among them, according to , determining the behavior score;
[0043] in Score the behavior. is the adjustment coefficient, which is related to the proportion of the user's head, torso and arms in the behavior score, M is the edge image feature data of the user's head, and M max is the maximum edge image feature data of the user's head, N is the edge image feature data of the user's torso, and N max is the maximum edge image feature data of the user's torso, J is the edge image feature data of the user's arm, and J max It is the maximum edge image feature data of the user's arm.
[0044] As a preference of this embodiment, a computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the above-mentioned user driving behavior detection method.
[0045] (III) Beneficial effects
[0046] The present invention provides a user driving behavior detection method, system and storage medium, which have the following beneficial effects: by acquiring a user image in a cab, performing image processing on the user image, determining the user's current state based on the processed user image, judging whether the user's current state belongs to a dangerous behavior, if not, judging it as an initial state, if so, reminding the user to adjust to a normal driving state, and establishing an initial state based on the adjusted state, establishing a fluctuation threshold frame based on the processing result of the image of the user's initial state, periodically collecting the user's driving image, processing the periodically collected driving image, comparing the processed user driving image with the fluctuation threshold frame, judging whether the user has a dangerous driving behavior, realizing a customized detection method for the user's driving behavior, and realizing accurate detection of the user's driving behavior, and, by using a formula based on the comparison and classification results of the user's behavior, realizing scoring of the degree of danger of the user's behavior, each time it is determined that the user is in normal driving behavior, the processed picture of the immediate normal driving behavior will be used as the initial picture for subsequent comparison, and the user's driving behavior can be detected more accurately through continuous comparison. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] Figure 1 is a flow chart of the user driving behavior detection method of the present invention;
[0048] Figure 2 It is a framework diagram of the user driving behavior detection system of the present invention. DETAILED DESCRIPTION
[0049] The embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.
[0050] The disclosure below provides many different embodiments or examples to realize different structures of the present invention. In order to simplify the disclosure of the present invention, the parts and settings of specific examples are described below. Of course, they are only examples, and the purpose is not to limit the present invention. In addition, the present invention can repeat reference numbers and / or reference letters in different examples, and this repetition is for the purpose of simplicity and clarity, which itself does not indicate the relationship between the various embodiments and / or settings discussed. In addition, the present invention provides various specific examples of processes and materials, but those of ordinary skill in the art can appreciate the application of other processes and / or the use of other materials.
[0051] like Figure 1 As shown, an embodiment of the present invention provides a method for detecting a user's driving behavior, comprising:
[0052] Acquire a user image in the cab and perform image processing on the user image;
[0053] The image of the user in the cab can be captured by an electronic device to obtain an image of the user in a driving state.
[0054] Determine the user's current state based on the processed user image, judge whether the user's current state is a dangerous behavior, and if not, determine it as an initial state. If so, remind the user to adjust to a normal driving state, and establish an initial state based on the adjusted state;
[0055] Among them, when judging for the first time whether the user's current state is a dangerous behavior, it can also be determined by sending information to the user, collecting a picture of the user at a time after receiving the information, and determining whether the user's posture in the picture has obvious changes to determine whether the user is in a normal driving state.
[0056] It is understandable that after receiving the information, the user will judge his own driving status, and when he determines that he is in a dangerous driving state, he will adjust his posture.
[0057] In addition, each user has different driving habits, that is, each user has his or her own standard driving state. The user confirms the driving state by himself or herself, which can keep the user in an independent marked driving state in the early stage of driving.
[0058] Establishing a fluctuation threshold frame based on the processing result of the image of the user's initial state;
[0059] The user's driving images are collected periodically, processed, and compared with the fluctuation threshold frame based on the processed user driving images to determine whether the user has dangerous driving behavior.
[0060] Specifically, by collecting the user's initial driving status picture and analyzing it, it can be determined that the user maintains a normal driving state in the initial stage of driving. Then, a threshold frame is established based on the user's initial driving state. By periodically collecting the user's driving status pictures and using the same processing method, the offset of the user's posture during driving can be determined. When the user's offset exceeds the threshold frame, it can be determined whether the user is in a dangerous behavior. At the same time, the detection method can establish corresponding detection standards for different users, which can effectively improve the accuracy of detection.
[0061] It should be noted that after processing the user image, the background area in the image will be removed.
[0062] Furthermore, the acquiring of the user image in the cab and performing image processing on the user image includes:
[0063] Collect the user's current state image through the electronic device to obtain the relevant contour and feature point positions of the user's appearance;
[0064] Obtain a binary mask of the ROI of the head, torso and arm regions of the user image to obtain a first image;
[0065] Based on the binary mask and edge curve of the region of interest of the user's head, torso and arm in the first image, a second image is obtained, where the second image is a ROI image of the region of interest of the user's head, torso and arm of the color channel image of the first image, where the color channel is a color channel that can reflect the user's behavior;
[0066] Based on the binary mask of the second image, the connected domain of the user behavior is extracted, and the circumscribed rectangular frame corresponding to each connected domain is obtained;
[0067] Based on the circumscribed rectangular frame and the user image, image blocks at corresponding positions of the two are obtained.
[0068] Specifically, the binary masks of the ROIs of the user's head, torso, and arm regions are obtained. The features of the head, torso, and arm regions can well reflect the user's current state. Among them, the key area of the user's head image is the user's eye area. The image size of the eye area can be used to analyze whether the user is currently in a tired state.
[0069] Among them, the edge curves of the areas of interest of the user's head, torso and arms can detect the position of the user's hands and the steering wheel, whether the eyes are sleepy, and abnormal behaviors such as distraction while driving, effectively improving the effectiveness of detection.
[0070] For example, the user's behavior can be determined to have changed during the time interval from the initial image to the periodic image capture by observing the displacement of the user's head compared to the initial image or a significant change in the size of the user's eyes, changes in the position of the user's torso, and changes in the position of the user's arms. The user's behavior can be detected more accurately by the amount of change in the user's behavior before and after.
[0071] Further, determining the current state of the user based on the processed user image and judging whether the current state of the user belongs to dangerous behavior includes:
[0072] Detecting the image blocks through a convolutional neural network model to determine the behavior type of the image blocks in the user's area of interest, the behavior type including the eye state, the torso position, and the arm position, and determining the user's physiological state when determining the eye state;
[0073] Analyze the user's current behavior status based on the user's behavior type.
[0074] It should be noted that the user's area of interest is specifically an area related to dangerous driving behavior, such as eyes, eye sockets, hand joints, etc.
[0075] Specifically, after obtaining the image blocks of the user's area of interest, the image blocks are detected through the convolutional neural network model to determine whether the corresponding behavior of each image block corresponding to the user's appearance area is normal or dangerous. Dangerous behavior can be squinting, closing eyes, torso position deviation, arms not on the steering wheel, etc.
[0076] Among them, pictures (image blocks) of the driving status of existing personnel can be collected. These pictures (image blocks) can be of dangerous driving or normal driving. These pictures (image blocks) are subjected to deep learning training of neural networks to obtain a convolutional neural network model. The convolutional neural network model can be used to perform feature analysis and judgment on the image blocks to determine whether the user is engaging in dangerous driving behavior.
[0077] Furthermore, the process of establishing a fluctuation threshold frame based on the processing result of the graph of the user's initial state includes:
[0078] Determine the activity range of the user's area of interest based on the behavior type of the image block of the user's area of interest and the user's physiological data;
[0079] Determine the scope of the user's dangerous behavior by drawing the edge curve of the activity range of the user's area of interest;
[0080] The fluctuation threshold box is determined based on the activity range of the user's area of interest and the range of the user's dangerous behavior.
[0081] Specifically, the user's external movable range is fixed, the user's movable range when driving will be smaller than the user's movable range, wherein the user's body movable range when driving will be smaller than the user's movable range, and a certain range of the user's activities when driving is in a normal driving state, and the fluctuation threshold block specifically is the range of the user's body movable in a normal driving state.
[0082] Furthermore, the periodic acquisition of the user's driving image is specifically to acquire the user's current image at a fixed interval after the user's initial state image is determined, and the interval is specifically any one of 15s, 30s or 45s;
[0083] The periodically collected driving images are processed in the same manner as the user images.
[0084] It is understandable that the interval of the interval time may also be other time periods.
[0085] Specifically, by processing the initial user image and the periodically collected user image in the same manner, the results of the processing of the two images can be similar, thereby enabling a better comparison between the two images.
[0086] At the same time, the processed image blocks eliminate most of the image areas and only compare and classify the areas related to the user's appearance and driving behavior, which can greatly simplify the calculation level and realize rapid comparison of user images.
[0087] Furthermore, comparing the processed user driving image with the fluctuation threshold frame to determine whether the user has dangerous driving behavior includes:
[0088] Obtain edge images of the user's head, torso, and arm regions in the user's driving image, and compare the user's head, torso, and arms with the fluctuation threshold frame;
[0089] The behavior score of the user's behavior is determined based on the comparison results of the edge images of the user's head, torso and arms with the fluctuation threshold box.
[0090] Furthermore, the behavior score of the user behavior determined based on the comparison result of the edge image of the user's head, torso and arms with the fluctuation threshold frame includes:
[0091] according to , determining the behavior score;
[0092] in Score the behavior. is the adjustment coefficient, which is related to the proportion of the user's head, torso and arms in the behavior score, M is the edge image feature data of the user's head, and M max is the maximum edge image feature data of the user's head, N is the edge image feature data of the user's torso, and Nmax is the maximum edge image feature data of the user's torso, J is the edge image feature data of the user's arm, and J max It is the maximum edge image feature data of the user's arm.
[0093] Specifically, the dangerousness of user behavior is scored using a formula based on the comparison and classification results of user behavior. Each time the user is determined to be in normal driving behavior, the processed picture of the immediate normal driving behavior will be used as the initial picture for subsequent comparison. Through continuous comparison, the user's driving behavior can be detected more accurately.
[0094] The present application provides a method for detecting user driving behavior, which obtains a user image in the cab and processes the user image. The current state of the user is determined based on the processed user image, and it is judged whether the current state of the user belongs to dangerous behavior. If it is not dangerous behavior, it is judged as the initial state. If it is, the user is reminded to adjust to a normal driving state, and the initial state is established based on the adjusted state. A fluctuation threshold frame is established based on the processing result of the image of the user's initial state. The user's driving image is periodically collected, and the periodically collected driving image is processed. Based on the comparison between the processed user driving image and the fluctuation threshold frame, it is judged whether the user has dangerous driving behavior. A customized detection method for user driving behavior is realized, and accurate detection of user driving behavior is realized at the same time. Moreover, by comparing and classifying the user behavior, a formula is used to score the degree of danger of the user behavior. Each time it is determined that the user is in normal driving behavior, the processed picture of the immediate normal driving behavior will be used as the initial picture for subsequent comparison, and the user's driving behavior can be detected more accurately through continuous comparison.
[0095] The present invention also provides a user driving behavior detection system, which is used to implement the above-mentioned user driving behavior detection method, comprising:
[0096] An acquisition unit, used for acquiring an image of a user in the cab through an electronic device;
[0097] The image processing unit is used to obtain the relevant contours and feature point positions of the user's appearance, obtain the binary mask of the ROI of the head, torso and arm areas of the user image to obtain the first image, obtain the second image based on the binary mask and edge curve of the user's head, torso and arm areas of interest in the first image, the second image is the user's head, torso and arm areas ROI image of the color channel image of the first image, the color channel is a color channel that can reflect the user's behavior, extract the connected domain of the user's behavior based on the binary mask of the second image, and obtain the circumscribed rectangular frame corresponding to each connected domain, and obtain the image blocks at the corresponding positions of the two based on the circumscribed rectangular frame and the user image; and is also used to determine the activity range of the user's area of interest based on the behavior type of the image block of the user's area of interest and the user's physiological data, determine the range of the user's dangerous behavior by drawing the edge curve of the activity range of the user's area of interest, and determine the fluctuation threshold frame based on the activity range of the user's area of interest and the range of the user's dangerous behavior.
[0098] A comparison and analysis unit, used to compare the processed user driving image with the fluctuation threshold frame to determine whether the user has dangerous driving behavior;
[0099] The memory is used to store the user image data and the processed user image data.
[0100] Furthermore, the comparison and analysis process of the comparison and analysis unit includes:
[0101] Obtain edge images of the user's head, torso, and arm regions in the user's driving image, and compare the user's head, torso, and arms with the fluctuation threshold frame;
[0102] Determine a behavior score of the user's behavior based on the comparison result of the edge images of the user's head, torso and arms with the fluctuation threshold box;
[0103] Among them, according to , determining the behavior score;
[0104] in Score the behavior. is the adjustment coefficient, which is related to the proportion of the user's head, torso and arms in the behavior score, M is the edge image feature data of the user's head, and M max is the maximum edge image feature data of the user's head, N is the edge image feature data of the user's torso, and N max is the maximum edge image feature data of the user's torso, J is the edge image feature data of the user's arm, and J max It is the maximum edge image feature data of the user's arm.
[0105] The present invention also provides a computer-readable storage medium, wherein the storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the above-mentioned user driving behavior detection method.
[0106] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for detecting user driving behavior, characterized in that: include Acquire a user image in the cab and perform image processing on the user image; Determine the current state of the user based on the processed user image, judge whether the current state of the user belongs to dangerous behavior, if not, determine it as the initial state, if so, remind the user to adjust to the normal driving state, and establish the initial state based on the adjusted state; Establishing a fluctuation threshold frame based on the processing result of the image of the user's initial state; Periodically collect driving images of users, process the periodically collected driving images, and compare the processed driving images of users with the fluctuation threshold frame to determine whether the user has dangerous driving behavior; The acquiring of the user image in the cab and performing image processing on the user image comprises: Collect the user's current state image through the electronic device to obtain the relevant contour and feature point positions of the user's appearance; Obtain a binary mask of the ROI of the head, torso and arm regions of the user image to obtain a first image; Based on the binary mask and edge curve of the region of interest of the user's head, torso and arm in the first image, a second image is obtained, where the second image is a ROI image of the region of interest of the user's head, torso and arm of the color channel image of the first image, where the color channel is a color channel that can reflect the user's behavior; Based on the binary mask of the second image, the connected domain of the user behavior is extracted, and the circumscribed rectangular frame corresponding to each connected domain is obtained; Based on the circumscribed rectangular frame and the user image, obtaining image blocks at corresponding positions of the two; Determining the current state of the user based on the processed user image and judging whether the current state of the user belongs to dangerous behavior includes: Detecting the image blocks through a convolutional neural network model to determine the behavior type of the image blocks in the user's area of interest, the behavior type including the eye state, the torso position, and the arm position, and determining the user's physiological state when determining the eye state; Analyze the user's current behavior status based on the user's behavior type; The processing result of the graph based on the user's initial state to establish a fluctuation threshold frame includes: Determine the activity range of the user's area of interest based on the behavior type of the image block of the user's area of interest and the user's physiological data; Determine the scope of the user's dangerous behavior by drawing the edge curve of the activity range of the user's area of interest; The fluctuation threshold box is determined based on the activity range of the user's area of interest and the range of the user's dangerous behavior.
2. A user driving behavior detection method according to claim 1, characterized in that: The periodic acquisition of the user's driving image specifically includes acquiring the user's current image at a fixed interval after the user's initial state image is determined, and the interval time is specifically any one of 15s, 30s or 45s; The periodically collected driving images are processed in the same manner as the user images.
3. A user driving behavior detection method according to claim 2, characterized in that: The comparing the processed user driving image with the fluctuation threshold frame to determine whether the user has dangerous driving behavior includes: Obtain edge images of the user's head, torso, and arm regions in the user's driving image, and compare the user's head, torso, and arms with the fluctuation threshold frame; The behavior score of the user's behavior is determined based on the comparison results of the edge images of the user's head, torso and arms with the fluctuation threshold box.
4. A method for detecting user driving behavior according to claim 3, characterized in that: The behavior score of the user behavior determined based on the comparison result of the edge images of the user's head, torso and arms with the fluctuation threshold frame includes: according to , determining the behavior score; in Score the behavior. is the adjustment coefficient, which is related to the proportion of the user's head, torso and arms in the behavior score, M is the edge image feature data of the user's head, and M max is the maximum edge image feature data of the user's head, N is the edge image feature data of the user's torso, and N max is the maximum edge image feature data of the user's torso, J is the edge image feature data of the user's arm, and J max It is the maximum edge image feature data of the user's arm.
5. A user driving behavior detection system, used to implement the user driving behavior detection method according to any one of claims 1 to 4, characterized in that: include: An acquisition unit, used for acquiring an image of a user in the cab through an electronic device; An image processing unit is used to obtain relevant contours and feature point positions of the user's appearance, obtain a binary mask of the ROI of the head, torso and arm regions of the user image to obtain a first image, obtain a second image based on the binary mask and edge curve of the ROI of the user's head, torso and arm regions in the first image, wherein the second image is an ROI image of the user's head, torso and arm regions of the color channel image of the first image, wherein the color channel is a color channel that can reflect the user's behavior, extract a connected domain of the user's behavior based on the binary mask of the second image, and obtain an external rectangular frame corresponding to each connected domain, and obtain an image block at a corresponding position of the two based on the external rectangular frame and the user image; and is also used to determine an activity range of the user's area of interest based on the behavior type of the image block of the user's area of interest and the user's physiological data, determine the range of the user's dangerous behavior by drawing an edge curve of the activity range of the user's area of interest, and determine a fluctuation threshold frame based on the activity range of the user's area of interest and the range of the user's dangerous behavior; A comparison and analysis unit, used to compare the processed user driving image with the fluctuation threshold frame to determine whether the user has dangerous driving behavior; The memory is used to store the user image data and the processed user image data.
6. A user driving behavior detection system according to claim 5, characterized in that: The comparison and analysis process of the comparison and analysis unit includes: Obtain edge images of the user's head, torso, and arm regions in the user's driving image, and compare the user's head, torso, and arms with the fluctuation threshold frame; Determine a behavior score of the user's behavior based on the comparison result of the edge images of the user's head, torso and arms with the fluctuation threshold box; according to , determining the behavior score; in Score the behavior. is the adjustment coefficient, which is related to the proportion of the user's head, torso and arms in the behavior score, M is the edge image feature data of the user's head, and M max is the maximum edge image feature data of the user's head, N is the edge image feature data of the user's torso, and N max is the maximum edge image feature data of the user's torso, J is the edge image feature data of the user's arm, and J max It is the maximum edge image feature data of the user's arm.
7. A computer-readable storage medium, characterized in that: The storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the user driving behavior detection method described in any one of claims 1-4.
Citation Information
Patent Citations
Driver attention state monitoring method and device
CN101877051A
