Computer visual feature recognition method and system for intelligent driving

Abstract

The invention relates to the field of image processing, in particular to a computer visual feature recognition method and system for intelligent driving, and the method comprises the following steps: obtaining image information; a first image area is positioned and cut through a face detection algorithm, and the rest part is a second image area; for the first image area, positioning and cutting out a lip image area, and forming a first recognition result through image analysis of the lip image area; forming a second recognition result through image analysis of the second image area; performing fusion processing on the first identification result and the second identification result through a fusion algorithm to output a fusion result; and judging whether a dangerous driving risk exists or not according to a fusion result. According to the invention, the target scene of dispute between the driver and the passenger can be accurately identified, and timely and accurate risk judgment can be made; and moreover, the analysis efficiency is high, and the instantaneity is strong.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a computer vision feature recognition method and system for intelligent driving. Background Art

[0002] With the rapid development of intelligent driving technology, we have more and more technical means to ensure driving safety. Among the many safety measures, the application of image analysis technology is particularly common and critical. First, the system detects potential dangerous driving behaviors through accurate image recognition functions; then, the system will take corresponding intervention measures, which may include directly taking over the control of the car, or promptly warning the driver and guiding them to take necessary safety actions.

[0003] There are two key points in the above image analysis technology used to ensure safe driving: 1. How to accurately identify dangerous driving behaviors, that is, how to build a model of dangerous driving behaviors, or how to locate dangerous driving scenarios; 2. How to simplify the algorithm and reduce the amount of calculation to ensure a quick response.

[0004] Among all dangerous driving behaviors, there is one type of behavior that is extremely dangerous: disputes between passengers and drivers, which seriously affect the driver's emotions and even actions. Quarrels may cause the driver to lose focus on driving and make correct driving decisions in a timely manner, such as braking and steering, increasing the risk of traffic accidents. Pushing and fighting may cause the driver to lose control of the vehicle, seriously endangering the safety of people in the cab, and may also endanger public safety and cause serious traffic accidents because the driver loses control of the vehicle.

[0005] The existing technology lacks an identification solution for the above-mentioned dangerous driving behaviors. If the existing mature emotion management algorithm is used, it has the disadvantages of large computational complexity and poor recognition accuracy. Therefore, it is necessary to design a new visual feature recognition method. Summary of the invention

[0006] In response to the above technical problems, the present application provides a computer vision feature recognition method and system that can accurately and efficiently identify target scenes.

[0007] The present application provides a computer vision feature recognition method for intelligent driving, comprising the following steps:

[0008] Step S1: Acquire image information including the driver and the driver's environment;

[0009] Step S2: using a face detection algorithm, locating and cropping a face region in the image information as a first image region, and cropping the remaining portion as a second image region;

[0010] Step S3: for the first image area, further locate and crop the lip image area, and form a first recognition result for determining whether there is a risk of dangerous driving by analyzing the image of the lip image area;

[0011] Step S4: for the second image area, forming a second recognition result for determining whether there is a risk of dangerous driving by analyzing the image of the second image area;

[0012] The image analysis of the second image area includes: determining whether external interference material appears in the area;

[0013] Step S5: fusing the first recognition result and the second recognition result through a fusion algorithm to output a fusion result;

[0014] Step S6: Determine whether there is a risk of dangerous driving based on the fusion result.

[0015] In the above technical solution, further, the step S3 specifically includes:

[0016] Step S31, roughly locating the lips using skin color;

[0017] Step S32, using the shape to accurately locate the lips;

[0018] Step S33, picking up a number of feature points on the lip image;

[0019] Step S34, judging the amplitude and / or frequency of lip changes according to a number of feature points, and scoring the driver's negative emotions;

[0020] Step S35: Generate a first recognition result according to the score.

[0021] In the above technical solution, further, the step S33 specifically includes: picking up the upper vertex of the lip center axis, the lower base point of the lip center axis, the leftmost lip corner point and the rightmost lip corner point respectively, and scoring the driver's negative emotions according to the positional relationship between the four feature points.

[0022] In the above technical solution, further, the step S3 specifically includes:

[0023] By analyzing the image of the lip image area, the language information is identified and the driver's negative emotions are scored based on the voice information.

[0024] In the above technical solution, further, in step S4, the image analysis of the second image area includes:

[0025] Step S41, by comparing two adjacent image frames, determining whether other materials except the driver appear in the second image area; if yes, proceeding to step S42;

[0026] Step S42, counting the time when other materials exist as a first interference determination result;

[0027] Step S43, calculating the movement frequency and amplitude of other materials in the second image area as a second interference determination result;

[0028] Step S44: judging whether other materials are external interference materials according to the first interference judgment result and the second interference judgment result.

[0029] In the above technical solution, further, after determining that other materials are external interference materials, it also includes:

[0030] Step S45, extracting the driver's profile information;

[0031] Step S46, calculating the change range of the driver's posture through the profile information;

[0032] Step S47: forming a second recognition result according to the change range.

[0033] In the above technical solution, further, step S2 also includes: through an edge detection algorithm, locating and cropping the driver's posture image in the second image area as a third image area, and intercepting the image within a certain range outside the third area as a fourth image area; in step S41, by comparing two adjacent image frames, it is determined whether other materials besides the driver appear in the fourth image area; in step S45, the driver's contour information is extracted in the third image area.

[0034] In the above technical solution, further, step S3 also includes: performing facial emotion recognition on the first image area, and forming a first recognition result according to the facial emotion recognition result and the image analysis result of the lip image area.

[0035] In the above technical solution, further, facial emotion recognition is performed on the first image area, including: analyzing facial emotions based on the position and shape of eyebrows, the position of the corners of the eyes, the degree of eye openness, the facial contour and facial landmarks; the facial landmarks refer to key positions of the face, including the corners of the eyes, the tip of the nose and the corners of the mouth.

[0036] In addition, the present application also provides a computer vision feature recognition system for intelligent driving, including:

[0037] Acquisition module: used to acquire image information including the driver and the environment in which the driver is located;

[0038] Image segmentation module: through the face detection algorithm, locate and crop the face area in the image information as the first image area, and the remaining cropped part as the second image area;

[0039] First image region analysis module: for the first image region, further locates and crops the lip image region, and forms a first recognition result for determining whether there is a risk of dangerous driving by analyzing the image of the lip image region;

[0040] Second image area analysis module: for the second image area, forming a second recognition result for judging whether there is a risk of dangerous driving by analyzing the image of the second image area; the image analysis of the second image area includes: judging whether there is external interference material in the area;

[0041] Fusion calculation module: fusing the first recognition result and the second recognition result through a fusion algorithm to output a fusion result;

[0042] Risk judgment module: Determine whether there is a risk of dangerous driving based on the fusion results.

[0043] The computer vision feature recognition method and system for intelligent driving disclosed in the embodiment of the present application utilize a face detection algorithm to divide image information into a first image area and a second image area, perform lip image analysis on the first image area, and analyze whether external interference materials appear in the second image area, and then comprehensively judge whether there is a risk of dangerous driving through the image analysis results of the two parts. By comprehensively judging the two conditions of the driver's own state and external interference, the target scenario of a dispute between the driver and the passenger can be accurately identified, and a timely and accurate risk judgment can be made; moreover, since relatively independent image analysis is adopted for different areas, the two-step image analysis can be carried out simultaneously, which is conducive to improving analysis efficiency, which is crucial for intelligent driving; in addition, compared with the facial recognition emotion solution in the prior art, the present application only analyzes the lip image, which can reduce the amount of calculation and further improve the analysis efficiency, and the accuracy lost can be compensated by the analysis result of the second image area. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0045] Figure 1 A flowchart of a computer vision feature recognition method for intelligent driving provided in one embodiment of the present application.

[0046] Figure 2 A structural block diagram of a computer vision feature recognition system for intelligent driving provided in one embodiment of the present application.

[0047] Reference numerals

[0048] 100 - acquisition module; 200 - image segmentation module; 300 - first image region analysis module; 400 - second image region analysis module; 500 - fusion calculation module; 600 - risk judgment module. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0050] It should be noted that the terms "including" and "having" in the embodiments of the present application and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0051] It is understood that the terms "first", "second", etc. used in this application may be used herein to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, a first displacement vector may be referred to as a second displacement vector, and similarly, a second displacement vector may be referred to as a first displacement vector without departing from the scope of this application. Both the first displacement vector and the second displacement vector are displacement vectors, but they are not the same displacement vector.

[0052] The embodiments of the present application disclose a computer vision feature recognition method, device, electronic device and storage medium for intelligent driving, which can improve the accuracy of determining movement information.

[0053] The following is a detailed description with reference to the accompanying drawings.

[0054] Figure 1 It is a flowchart of a computer vision feature recognition method for intelligent driving disclosed in an embodiment of the present application.

[0055] like Figure 1 As shown, the computer vision feature recognition method for intelligent driving includes the following steps:

[0056] Step S1: Acquire image information including the driver and the driver's environment.

[0057] The image of the target area can be obtained by using the in-vehicle camera facing the driver's seat, and the image contains at least the image information of the driver and the image information of a certain range around the driver. The target area can be adaptively adjusted according to the driver's body shape and driving habits to include all the driver's information as much as possible.

[0058] The in-vehicle camera can reuse the vehicle-mounted camera in the existing technology, such as the camera in the driver monitoring system (DMS, Driver Monitor System) and the built-in camera in the advanced driver assistance system (ADAS).

[0059] Step S2: using a face detection algorithm, locate and crop the face area in the image information as the first image area, and crop the remaining part as the second image area.

[0060] The cropping of the first image area specifically comprises the following steps:

[0061] S21, image preprocessing: grayscale, noise removal, contrast enhancement and other operations are performed on the collected images to improve the accuracy of subsequent processing.

[0062] S22, face detection: Use face detection algorithms, such as Haar feature + AdaBoost algorithm, deep learning methods (such as MTCNN, RetinaFace, CenterFace, BlazeFace, YOLO, SSD, CascadeCNN, etc.) to detect the face area in the image. These algorithms can detect the position and size of the face from the image.

[0063] S23, candidate region generation: The face detection algorithm generates candidate regions (bounding boxes) containing faces, which are determined through classification and regression analysis.

[0064] S24, face region cropping: Once a face is detected, the algorithm will crop the face region from the original image to form a first image region.

[0065] Step S3: for the first image area, further locate and crop the lip image area, and form a first recognition result for determining whether there is a risk of dangerous driving through image analysis of the lip image area.

[0066] In some embodiments, step S3 specifically includes:

[0067] Step S31, roughly locate the lips using skin color. Preliminary and rapid location can be performed by skin color detection and the chromaticity difference between lip color and skin color.

[0068] Step S32, using the shape to accurately locate the lips. For the roughly located lip image area, further identification and positioning are achieved based on a deep learning method. For example, a convolutional neural network (CNN) is used to identify lip features.

[0069] Step S33, picking up several feature points on the lip image.

[0070] The upper vertex of the lip center axis, the lower bottom point of the lip center axis, the leftmost lip corner point and the rightmost lip corner point are picked up respectively, and the driver's negative emotions are scored according to the positional relationship between the four feature points.

[0071] Step S34, judging the amplitude and / or frequency of lip changes according to a number of feature points, and scoring the driver's negative emotions;

[0072] Only the above four feature points are needed to reflect a large amount of information. For example, when the distance between the upper vertex and the lower base exceeds a certain threshold, the driver is likely to be shouting. Another example is when the change rate of the distance between the upper vertex and the lower base, or the change rate of the distance between the leftmost lip corner and the rightmost lip corner exceeds a certain threshold, the driver may be speaking in an anxious and angry manner.

[0073] Therefore, the extracted four feature points can be used to train the negative emotion scoring model. This model can be trained by supervised learning using a dataset labeled with emotion labels (such as FER2013, AffectNet, etc.).

[0074] Since fewer feature points are used, the amount of calculation is small and the efficiency is high, which is conducive to ensuring the timeliness of the analysis results.

[0075] Step S35: Generate a first recognition result according to the score.

[0076] In other embodiments, when the hardware conditions are high enough to meet the computing requirements and ensure the computing efficiency, lip reading recognition can also be performed on the lip image. For the lip image, the lip reading recognition algorithm is used to obtain the language output by the driver, and after semantic recognition, it is combined with the above expression analysis results to make a more accurate negative emotion score.

[0077] In other embodiments, when the hardware conditions are high enough to meet the computing requirements and ensure computing efficiency, facial emotion recognition is performed on the first image area, and a first recognition result is formed based on the facial emotion recognition result and the image analysis result of the lip image area.

[0078] In addition to the lip image, emotion recognition is also performed on the entire face to score the driver's negative emotions.

[0079] For example, facial emotion recognition is performed on the first image area. The analysis of facial emotions usually involves multiple aspects of facial expressions, including the position and shape of eyebrows, the position of eye corners, the degree of eye openness, facial contours, and facial landmarks.

[0080] Eyebrow position and shape: Raising, lowering or frowning eyebrows can convey different emotions, such as surprise, sadness or anger. For example, a raised eyebrow may indicate surprise or happiness, while a frown may indicate anger or confusion.

[0081] Eye position: The elevation or drooping of the corners of the eyes can help identify emotions. For example, an upward angle of the eye is often associated with happy or excited emotions, while a drooping eye may indicate sadness or tiredness.

[0082] Eye openness: The degree to which the eyes are open can reflect the intensity of the emotion. For example, wide-open eyes are often associated with surprise or fear, while half-closed eyes can be associated with relaxation or sleepiness.

[0083] Facial profile: The overall outline of the face, including the tension in the cheeks and the position of the corners of the mouth, can provide clues to emotion. For example, upturned corners of the mouth often indicate happiness, while downturned corners may indicate sadness or dissatisfaction.

[0084] Facial landmarks: Facial expressions can be analyzed more accurately by using facial landmarks, such as the 68 facial feature points in the dlib library. These landmarks can be used to measure the movement and position changes of different areas of the face, such as the degree of opening of the mouth, the degree of closing of the eyes, etc.

[0085] Machine learning or deep learning models can be used to analyze these features and predict emotional states based on their changes. These models can be trained to recognize patterns of different emotions, thereby achieving automated emotion recognition.

[0086] Step S4: for the second image area, forming a second recognition result for judging whether there is a risk of dangerous driving by analyzing the image of the second image area. The image analysis of the second image area includes: judging whether there is external interference material in the area.

[0087] Image analysis is performed on the second image area to determine whether there is any material that interferes with or affects the driver. For example, when a dispute occurs, passengers in the front passenger seat or the back seat attack the driver with their body parts or objects; or a pet in the car enters the driver's driving space.

[0088] In some embodiments, the material that enters the second image area and meets certain motion frequency and amplitude may be regarded as interference material. The specific determination method includes the following steps:

[0089] Step S41, by comparing two adjacent image frames, determining whether other materials except the driver appear in the second image area; if yes, proceeding to step S42;

[0090] Step S42, counting the time when other materials exist as a first interference determination result;

[0091] Step S43, calculating the movement frequency and amplitude of other materials in the second image area as a second interference determination result;

[0092] Step S44: judging whether other materials are external interference materials according to the first interference judgment result and the second interference judgment result.

[0093] In other embodiments, in order to ensure the recognition accuracy, it is determined whether external interference occurs from both the outside world and the driver's own perspectives. When other materials are determined to be external interference materials, it also includes:

[0094] Step S45, extracting the driver's profile information;

[0095] Step S46, calculating the change range of the driver's posture through the profile information;

[0096] Step S47: forming a second recognition result according to the change range.

[0097] In other embodiments, through a face detection algorithm, the face area is located and cropped in the image information as the first image area, and the remaining cropped part is used as the second image area; through an edge detection algorithm, the driver posture image is located and cropped in the second image area as the third image area, and the image within a certain range outside the third area is captured as the fourth image area.

[0098] In the above step S41, by comparing two adjacent image frames, it is determined whether other materials except the driver appear in the fourth image area;

[0099] In the above step S45, the driver's profile information is extracted in the third image area.

[0100] Step S5: Fusing the first recognition result and the second recognition result through a fusion algorithm to output a fusion result.

[0101] Step S6: Determine whether there is a risk of dangerous driving based on the fusion result.

[0102] The computer vision feature recognition method for intelligent driving disclosed in the embodiment of the present application utilizes a face detection algorithm to divide image information into a first image area and a second image area, performs lip image analysis on the first image area, and analyzes whether external interference materials appear in the second image area, and then comprehensively judges whether there is a risk of dangerous driving through the image analysis results of the two parts. By comprehensively judging the two conditions of the driver's own state and external interference, the target scenario of a dispute between the driver and the passenger can be accurately identified, and a timely and accurate risk judgment can be made; moreover, since relatively independent image analysis is adopted for different areas, the two-step image analysis can be carried out simultaneously, which is conducive to improving analysis efficiency, which is crucial for intelligent driving; in addition, compared with the facial recognition emotion solution in the prior art, the present application only analyzes the lip image, which can reduce the amount of calculation and further improve the analysis efficiency, and the accuracy lost can be compensated by the analysis result of the second image area.

[0103] Figure 2 A structural block diagram of a computer vision feature recognition system for intelligent driving provided in one embodiment of the present application.

[0104] like Figure 2 As shown, a computer vision feature recognition system for intelligent driving provided by an embodiment of the present application includes:

[0105] Acquisition module 100: used to acquire image information including the driver and the driver's environment;

[0106] Image segmentation module 200: using a face detection algorithm, locates and cuts out a face area in the image information as a first image area, and cuts out the rest as a second image area;

[0107] The first image region analysis module 300 further locates and crops the lip image region in the first image region, and forms a first recognition result for determining whether there is a risk of dangerous driving by analyzing the image of the lip image region;

[0108] The second image region analysis module 400: for the second image region, forming a second recognition result for determining whether there is a risk of dangerous driving by analyzing the image of the second image region; the image analysis of the second image region includes: determining whether there is external interference material in the region;

[0109] Fusion calculation module 500: fusing the first recognition result and the second recognition result through a fusion algorithm to output a fusion result;

[0110] Risk judgment module 600: judges whether there is a risk of dangerous driving based on the fusion result.

[0111] The computer vision feature recognition system for intelligent driving disclosed in the embodiment of the present application can be integrated into the intelligent driving system or used as a separate system. Its principles and advantages are consistent with the above-mentioned computer vision feature recognition method for intelligent driving, and will not be repeated here.

[0112] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0113] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0114] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0115] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0116] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0117] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0118] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device that can carry the computer program code to an electronic device, a recording medium, a computer memory, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), an electric carrier signal, a telecommunication signal, and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0119] Finally, it should be noted that the above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A computer vision feature recognition method for intelligent driving, characterized in that: The following steps are involved: Step S1: Acquire image information including the driver and the driver's environment; Step S2: using a face detection algorithm, locating and cropping a face region in the image information as a first image region, and the rest as a second image region; Step S3: for the first image area, further locate and crop the lip image area, and form a first recognition result for determining whether there is a risk of dangerous driving by analyzing the image of the lip image area; Step S4: for the second image area, forming a second recognition result for determining whether there is a risk of dangerous driving by analyzing the image of the second image area; The image analysis of the second image area includes: determining whether external interference material appears in the area; Step S5: fusing the first recognition result and the second recognition result through a fusion algorithm to output a fusion result; Step S6: Determine whether there is a risk of dangerous driving based on the fusion result.

2. The computer vision feature recognition method for intelligent driving according to claim 1, characterized in that: The step S3 specifically includes: Step S31, roughly locating the lips using skin color; Step S32, using the shape to accurately locate the lips; Step S33, picking up a number of feature points on the lip image; Step S34, judging the amplitude and / or frequency of lip changes according to a number of feature points, and scoring the driver's negative emotions; Step S35: generating a first recognition result according to the score.

3. The computer vision feature recognition method for intelligent driving according to claim 2, characterized in that: The step S33 specifically includes: picking up the upper vertex of the lip center axis, the lower base point of the lip center axis, the leftmost lip corner point and the rightmost lip corner point respectively, and scoring the driver's negative emotions according to the positional relationship between the four feature points.

4. The computer vision feature recognition method for intelligent driving according to any one of claims 1 to 3, characterized in that: The step S3 specifically includes: By analyzing the image of the lip image area, the language information is identified and the driver's negative emotions are scored based on the voice information.

5. The computer vision feature recognition method for intelligent driving according to claim 1, characterized in that: In step S4, the image analysis of the second image area includes: Step S41, by comparing two adjacent image frames, determining whether other materials except the driver appear in the second image area; if yes, proceeding to step S42; Step S42, counting the time when other materials exist as a first interference determination result; Step S43, calculating the movement frequency and amplitude of other materials in the second image area as a second interference determination result; Step S44: judging whether other materials are external interference materials according to the first interference judgment result and the second interference judgment result.

6. The computer vision feature recognition method for intelligent driving according to claim 5, characterized in that: When other materials are judged to be external interference materials, it also includes: Step S45, extracting the driver's profile information; Step S46, calculating the change range of the driver's posture through the profile information; Step S47: forming a second recognition result according to the change range.

7. The computer vision feature recognition method for intelligent driving according to claim 6, characterized in that: Step S2 also includes: using an edge detection algorithm, locating and cropping the driver's posture image in the second image area as a third image area, and intercepting the image within a certain range outside the third area as a fourth image area; in step S41, by comparing two adjacent image frames, determining whether other materials besides the driver appear in the fourth image area; and in step S45, extracting the driver's contour information in the third image area.

8. The computer vision feature recognition method for intelligent driving according to claim 1, characterized in that: The step S3 also includes: performing facial emotion recognition on the first image area, and forming a first recognition result according to the facial emotion recognition result and the image analysis result of the lip image area.

9. The computer vision feature recognition method for intelligent driving according to claim 8, characterized in that: Performing facial emotion recognition on the first image area includes: analyzing facial emotions based on eyebrow position and shape, eye corner position, eye openness, facial contour and facial landmarks; the facial landmarks refer to key positions of the face including eye corners, nose tip and mouth corners.

10. A computer vision feature recognition system for intelligent driving, characterized in that: include: Acquisition module: used to acquire image information including the driver and the environment in which the driver is located; Image segmentation module: through the face detection algorithm, locate and crop the face area in the image information as the first image area, and the remaining cropped part as the second image area; First image region analysis module: for the first image region, further locates and crops the lip image region, and forms a first recognition result for determining whether there is a risk of dangerous driving by analyzing the image of the lip image region; A second image region analysis module: for the second image region, forming a second recognition result for determining whether there is a risk of dangerous driving by analyzing the image of the second image region; The image analysis of the second image area includes: determining whether external interference material appears in the area; Fusion calculation module: fusing the first recognition result and the second recognition result through a fusion algorithm to output a fusion result; Risk judgment module: Determine whether there is a risk of dangerous driving based on the fusion results.

Images