Driver abnormal emotion recognition method and vehicle distance control method

Abstract

The invention discloses a driver abnormal emotion recognition method and a vehicle distance control method, and the method comprises the steps: firstly, collecting image monitoring data and various physiological monitoring data of a driver in real time; and then, whether the driver has an abnormal emotion or not is quickly judged through image monitoring data. When the driver has the abnormal emotion, monitoring data in the corresponding time period are intercepted from the collected physiological monitoring data according to the time when the driver has the abnormal emotion, and whether the driver has the abnormal emotion or not is further judged. And when the driver has the abnormal emotion, adjusting the distance between the target vehicle driven by the driver and the front vehicle until the abnormal emotion of the driver disappears. The abnormal emotion of the driver can be accurately and reliably recognized by integrating the multi-modal data, and only part of monitoring data is intercepted for emotion recognition, so that the data processing amount is reduced, the data processing time is shortened, the emotion recognition efficiency is improved, and the emotion recognition result can be quickly and accurately fed back.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and in particular to a method for recognizing abnormal emotions of a driver and a method for controlling vehicle distance. Background Art

[0002] During the process of autonomous driving, the driver may experience abnormal emotions such as anxiety and panic due to the shortened distance between the vehicle and the vehicle in front. These abnormal emotions can easily cause the driver to make mistakes and lead to vehicle accidents. Therefore, during the process of autonomous driving, it is particularly important to quickly and accurately determine whether the driver has abnormal emotions.

[0003] Multimodal data refers to data from different sources or different types, such as text, images, audio, video, etc., with different characteristics and attributes. By integrating multiple types of data, more comprehensive and rich information can be obtained, thereby improving the effect of data processing and analysis. The use of multimodal data can comprehensively consider information from different aspects and improve the accuracy and reliability of emotion recognition.

[0004] Heart rate variability refers to the changing pattern of the heartbeat intervals. Studies have shown that heart rate variability has a certain ability to identify anxiety. In an anxious state, the pattern of heart rate changes will show a special pattern, usually characterized by a higher frequency of heart rate changes and increased irregularity. Therefore, by analyzing heart rate variability, an individual's anxiety level can be identified to a certain extent.

[0005] Electrodermal activity refers to changes in the conductivity of the human skin surface, which is usually related to emotional state. In anxiety, changes in skin electrical activity will reflect the body's stress response. In an anxious state, skin electrical activity tends to increase, that is, the skin conductance value increases. Therefore, by monitoring skin electrical activity, an individual's anxiety state can be identified to a certain extent. The ability of skin electricity to identify anxiety has been widely used in the fields of psychophysiology and psychological assessment.

[0006] In facial image recognition, eyebrows and pupils have a certain ability to identify anxiety. In an anxious state, eyebrows often have slight changes, such as frowning or slightly pulling eyebrows, which may show nervousness or worry. At the same time, the size of the pupil may also change. In an anxious state, the pupil tends to dilate, reflecting the stress response of the human nervous system. Therefore, by analyzing the changes in eyebrows and pupils, the individual's anxiety level can be identified to a certain extent. This facial image recognition technology has important application prospects in the fields of psychology, neuroscience and emotion recognition.

[0007] In this way, these types of monitoring data can comprehensively consider information from different aspects, thereby improving the accuracy and reliability of emotion recognition. However, in the application process, due to the excessive amount of data from physiological characteristic monitoring data such as skin electricity and heart rate variability, the amount of data processing is too large when emotion recognition is performed based on these physiological monitoring data, which requires more computing power resources and a longer data processing time. This results in low emotion recognition efficiency and the inability to quickly provide feedback on emotion recognition results. Summary of the invention

[0008] In view of the shortcomings of the prior art, the present invention proposes a method for identifying abnormal driver emotions and a method for controlling vehicle distance, which can improve the efficiency of accurately identifying driver emotions. The specific technical solution is as follows:

[0009] In a first aspect, a method for identifying abnormal emotions of a driver is provided. In a first implementable manner of the first aspect, the method comprises:

[0010] Obtain the driver's image monitoring data and various physiological monitoring data in real time;

[0011] Determining whether the driver has abnormal emotions through the acquired image monitoring data, and in response to the driver having abnormal emotions, extracting corresponding monitoring data from each item of the physiological monitoring data;

[0012] The extracted monitoring data are integrated to further verify the abnormal emotions of the driver.

[0013] In combination with the first implementable manner of the first aspect, in a second implementable manner of the first aspect, determining whether the driver has abnormal emotions through the image monitoring data includes:

[0014] Extracting key points of the human body from the image monitoring data, and locating position information of different key points of the human body in the image monitoring data;

[0015] According to the position information of the corresponding key points of the human body, determine the position of different parts of the human body;

[0016] The abnormal emotion is determined by combining the position changes of all the human body parts.

[0017] In combination with the second implementable manner of the first aspect, in a third implementable manner of the first aspect, extracting human body key points from the monitoring image includes:

[0018] Using a human body part detection algorithm to extract human body feature areas in the monitoring image;

[0019] The human body key points in the human body feature area are detected by using the trained key point detection model, and the position information of different human body key points is determined.

[0020] In combination with the third implementable manner of the first aspect, in a fourth implementable manner of the first aspect, extracting a human body feature area in the monitoring image using a human body part detection algorithm includes:

[0021] The human body feature area in the monitoring image is extracted using a human body part detection algorithm, including:

[0022] Performing Gaussian difference filtering on the monitoring image, and performing Gaussian difference processing on the processed monitoring image, marking key points corresponding to human body feature areas in the monitoring image;

[0023] The marked monitoring image is subjected to illumination balancing processing, and the monitoring image subjected to illumination balancing processing is subjected to grayscale histogram processing, and the human body feature area is extracted through the grayscale histogram obtained by the processing.

[0024] In combination with the third implementable manner of the first aspect, in a fifth implementable manner of the first aspect, the human body feature area is a facial area.

[0025] In combination with the fifth implementable manner of the first aspect, in a sixth implementable manner of the first aspect, the abnormal emotion is determined in combination with position changes of eyebrows and pupils in the facial area.

[0026] In combination with the first implementable manner of the first aspect, in a seventh implementable manner of the first aspect, extracting corresponding monitoring data from the physiological monitoring data includes:

[0027] Drawing a time domain graph corresponding to the corresponding physiological characteristics based on the physiological monitoring data;

[0028] According to the time when the driver shows abnormal emotions, the characteristic graph of the corresponding time period is intercepted from the time domain graph according to the specified time range as the monitoring data.

[0029] In combination with the first implementable manner of the first aspect, in an eighth implementable manner of the first aspect, the extracted monitoring data are integrated to further verify the abnormal emotion of the driver, including:

[0030] Extracting features from each of the monitoring data to obtain feature values ​​corresponding to different physiological features;

[0031] The characteristic values ​​corresponding to each physiological feature are matched with the characteristic values ​​corresponding to the corresponding abnormal emotions respectively, and the matching results corresponding to all physiological features are integrated to determine whether the driver has abnormal emotions.

[0032] In combination with the eighth implementable manner of the first aspect, in a ninth implementable manner of the first aspect, a neural network classifier is used to extract features of the monitoring data.

[0033] In a second aspect, a vehicle distance control method based on abnormal driver emotions is provided, comprising:

[0034] Using the driver's abnormal emotion recognition method described in any one of the first to ninth possible implementations of the first aspect to recognize the driver's abnormal emotion in real time;

[0035] When the driver shows abnormal emotions, adjust the distance between the target vehicle and the vehicle in front until the driver's abnormal emotions disappear.

[0036] Beneficial effect: The method for identifying abnormal emotions of a driver and the method for controlling vehicle distance of the present invention are used to obtain image monitoring data and use an image recognition algorithm to determine whether the driver has abnormal emotions. When the driver has abnormal emotions, part of the monitoring data can be intercepted from each of the physiological monitoring data to further verify the driver's abnormal emotions. By integrating multiple types of monitoring data, the driver's abnormal emotions can be accurately and reliably identified, and only a special part of the monitoring data is intercepted from a large amount of physiological monitoring data for emotion recognition, which reduces a large amount of data processing, does not require a large amount of computing power resources, shortens data processing time, improves emotion recognition efficiency, and can quickly and accurately provide feedback on emotion recognition results. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the specific implementation of the present invention, the following will briefly introduce the drawings required for use in the specific implementation. In all the drawings, each element or part is not necessarily drawn according to the actual scale.

[0038] Figure 1 A flowchart of a method for identifying abnormal driver emotions provided by an embodiment of the present invention;

[0039] Figure 2 A flow chart of a vehicle distance control method based on abnormal driver emotions provided by an embodiment of the present invention;

[0040] Figure 3 A schematic diagram of capturing a characteristic graph corresponding to skin electricity;

[0041] Figure 4 Schematic diagram of the characteristic graph corresponding to the intercepted heart rate variability. DETAILED DESCRIPTION

[0042] The following embodiments of the technical solution of the present invention are described in detail in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore only used as examples, and cannot be used to limit the protection scope of the present invention.

[0043] like Figure 1 The flowchart of the method for identifying abnormal driver emotions shown in FIG. 1 includes:

[0044] Step 1: Acquire the driver's image monitoring data and various physiological monitoring data in real time;

[0045] Step 2: determining whether the driver has abnormal emotions through the acquired image monitoring data, and in response to the driver having abnormal emotions, extracting corresponding monitoring data from each item of the physiological monitoring data;

[0046] Step 3: Fusion the extracted monitoring data to further verify the driver's abnormal emotions.

[0047] Specifically, first, the image monitoring data and various physiological monitoring data of the driver can be collected in real time through image acquisition equipment and physiological characteristic monitoring equipment respectively. Then, through the collected image monitoring data, the existing image emotion recognition algorithm can be used to quickly determine whether the driver has abnormal emotions. If the driver does not have abnormal emotions, data collection can continue. If the driver has abnormal emotions, the monitoring data of the corresponding time period can be intercepted from the collected physiological monitoring data according to the time when the driver has abnormal emotions, and the intercepted monitoring data can be further used to determine whether the driver has abnormal emotions.

[0048] In this way, by integrating various types of monitoring data, the driver's abnormal emotions can be accurately and reliably identified. Moreover, since only a special part of the monitoring data is extracted from a large amount of physiological monitoring data for emotion recognition, a large amount of data processing is reduced, and a large amount of computing resources are not required, which reduces the computing power requirements of the data processing equipment. Emotion recognition can be completed through the edge computing unit deployed on the vehicle, and there is no need to upload a large amount of data to the cloud platform for processing, which shortens the feedback time of the emotion recognition results. Moreover, due to the reduction in the amount of data, the data processing time is shortened, the efficiency of emotion recognition is improved, and the feedback time of the emotion recognition results is further shortened, so that the emotion recognition results can be fed back quickly and accurately.

[0049] In this embodiment, optionally, in step 2, judging whether the driver has abnormal emotions through the image monitoring data includes:

[0050] Extracting key points of the human body from the image monitoring data, and locating position information of different key points of the human body in the image monitoring data;

[0051] According to the position information of the corresponding key points of the human body, determine the position of different parts of the human body;

[0052] The abnormal emotion is determined by combining the position changes of all the human body parts.

[0053] Specifically, first, feature point extraction can be performed on the collected monitoring images, and the key points of the human body required for emotion recognition can be extracted from the monitoring images, and the image positions of the key points of the human body in the monitoring images can be located. Then, the changes in the parts of the human body that can reflect emotions can be determined based on the image positions of the key points of the human body in the monitoring images. Finally, the changes in the parts of the human body can be integrated to finally determine whether the driver has abnormal emotions.

[0054] In this embodiment, optionally, extracting key points of a human body from the monitoring image includes:

[0055] Using a human body part detection algorithm to extract human body feature areas in the monitoring image;

[0056] The human body key points in the human body feature area are detected by using the trained key point detection model, and the position information of different human body key points is determined.

[0057] Specifically, it should be understood that there are a large number of key points of the human body. If all the key points of the human body are extracted, the amount of processing will inevitably increase. Therefore, when extracting the key points of the human body, first, the existing human body part detection algorithm can be used to extract the clustered parts of the key points of the human body that can be used for emotion recognition from the monitoring image, such as the human body feature area image of the face. Then, through the trained key point monitoring model, such as the regression tree model, the key points of the human body in the human body feature area image are identified and located. In this way, the amount of data processing can be reduced, the data processing time can be shortened, the efficiency of emotion recognition can be improved, and the feedback time of the emotion recognition results can be further shortened.

[0058] In this embodiment, optionally, a human body part detection algorithm is used to extract a human body feature area in the monitoring image, including:

[0059] The human body feature area in the monitoring image is extracted using a human body part detection algorithm, including:

[0060] Performing Gaussian difference filtering on the monitoring image, and performing Gaussian difference processing on the processed monitoring image, marking key points corresponding to human body feature areas in the monitoring image;

[0061] The marked monitoring image is subjected to illumination balancing processing, and the monitoring image subjected to illumination balancing processing is subjected to grayscale histogram processing, and the human body feature area is extracted through the grayscale histogram obtained by the processing.

[0062] Specifically, first, the monitoring image is processed by Gaussian difference filtering. Gaussian difference filtering can highlight the edge information in the monitoring image and suppress noise to obtain significant edge features in the monitoring image, providing a basis for subsequent human feature area extraction. Then, the previously processed image can be processed by Gaussian difference. Gaussian difference processing can extract key points in the monitoring image, which often correspond to the features of human feature areas such as faces. This helps to locate areas that may contain faces. Since the lighting conditions in the monitoring environment may be poor, the monitoring image can be processed by lighting equalization to reduce the impact of lighting and make the facial features more prominent. Finally, the pre-processed monitoring image is processed by grayscale histogram. By analyzing the grayscale histogram, the characteristics of the brightness distribution in the monitoring image are determined, and the human feature area that may contain faces is further determined.

[0063] In this embodiment, optionally, the human feature area is a facial area. Because the eyebrows and pupils of the human face have certain recognition capabilities in anxiety. In an anxious state, the eyebrows often have slight changes, such as frowning or slightly pulling the eyebrows, which may manifest as nervous or worried emotions. At the same time, the size of the pupil may also change, and the pupil tends to expand in an anxious state, reflecting the stress response of the human nervous system. By analyzing the changes in eyebrows and pupils, the individual's anxiety level can be identified to a certain extent. Therefore, the driver's facial area can be selected as the human feature area, and the eyebrows and pupils can be selected as the parts of the human body that reflect human emotions.

[0064] It should be understood that in addition to facial features, the shaking of fingers, sweating of palms, clenched fists, etc.; shoulders that are sometimes raised and sometimes relaxed, upper body tense or shivering, upper body shaking back and forth frequently; feet moving restlessly, the center of gravity of the soles of the feet frequently shifting, leg muscle tension; voice trembling, speaking faster, speaking lower, breathing faster; eyes flickering, blinking frequency increased, eyes avoiding other places; head nodding or shaking frequently, neck muscles tense, head turning up and down from time to time, etc. can all be used to identify the abnormal emotions of the safety officer in unmanned driving. Therefore, these parts and areas can also be selected as human feature areas for the recognition of abnormal emotions of drivers.

[0065] In this embodiment, optionally, the abnormal emotion is determined in combination with changes in the positions of eyebrows and pupils in the facial area.

[0066] Specifically, by monitoring the overall movement distance of the eyebrows in the image, the driver's tension can be evaluated. Therefore, the movement distance of the driver's left eyebrow can be determined by the position coordinates of the eyebrow key points. The specific calculation formula is as follows:

[0067]

[0068] Among them, L 1x , L 1y is the coordinate of the left eyebrow key point L1, L 5x , L 5y is the coordinate of the left eyebrow key point L5. The movement distance D of the right eyebrow RM The calculation method is the same as that of the moving distance of the left eyebrow, so it will not be repeated here.

[0069] The driver's nervousness can be assessed by monitoring the distance between the pupil and the eyebrow in the image. The specific calculation formula is as follows:

[0070]

[0071] Among them, D LP is the distance between the left pupil and the left eyebrow, P 1x , P 1y is the coordinate of the left pupil key point P1, L 3x , L 3y is the coordinate of the left eyebrow key point L3. The distance D between the right pupil and the right eyebrow RP The calculation method is the same as that of the distance between the left pupil and the left eyebrow, so it will not be repeated here.

[0072] The driver's nervousness can also be assessed by monitoring the changes in the angle of the eyebrows in the image. The specific calculation formula is as follows:

[0073] △θ LM =θ(L3, L1, L5)-θ reference ;

[0074] Among them, △ LM is the angle change of the left eyebrow, θ(L3, L1, L5) is the overall angle of the left eyebrow, θ reference is the reference angle for setting. The angle change value of the right eyebrow△ RM The calculation method is the same as that of the angle change value of the left eyebrow, and will not be repeated here.

[0075] After calculating the above-mentioned changes in eyebrow angles, the distance between pupils and eyebrows, and the overall movement distance of eyebrows, the driver's nervousness can be comprehensively evaluated by combining the weight coefficients corresponding to each parameter, thereby integrating data from a variety of different parameters to accurately and reliably identify the driver's abnormal emotions and further improve the accuracy of identifying the driver's abnormal emotions. The specific calculation formula is as follows:

[0076] A=k1D LM +k2D RM +k3D LP +k4D RP +k5θ LM +k6θ RM ;

[0077] k1, k2, k3, k4, k5, k6 are weight coefficients corresponding to various parameters respectively. The weight coefficients corresponding to various parameters can be determined by using existing weight coefficient analysis methods, such as hierarchical analysis method, regression analysis, etc.

[0078] In this embodiment, optionally, extracting corresponding monitoring data from the physiological monitoring data includes:

[0079] Drawing a time domain graph corresponding to the corresponding physiological characteristics based on the physiological monitoring data;

[0080] According to the time when the driver shows abnormal emotions, the characteristic graph of the corresponding time period is intercepted from the time domain graph according to the specified time range as the monitoring data.

[0081] Specifically, if Figure 3 , Figure 4 As shown, when intercepting monitoring data from physiological monitoring data, the corresponding time domain diagram can be drawn according to the collected physiological monitoring data. In this embodiment, the collected physiological monitoring data may include the driver's heart rate monitoring data and skin electricity data. Then, the time when the driver has abnormal emotions is used as the reference time point, that is, the image acquisition time corresponding to the monitoring image in which the driver has abnormal emotions is determined, and the image within 2 seconds before and after the reference time point is intercepted from the time domain diagram as a feature map for further verification of subsequent abnormal emotions. In this way, the subsequent data processing volume can be greatly reduced, without the need for a large amount of computing power resources, and the data processing time is shortened, the efficiency of emotion recognition is improved, and the emotion recognition results can be fed back quickly and accurately.

[0082] In this embodiment, optionally, the extracted monitoring data are integrated to further verify the abnormal emotions of the driver, including:

[0083] Extracting features from each of the monitoring data to obtain feature values ​​corresponding to different physiological features;

[0084] The characteristic values ​​corresponding to each physiological feature are matched with the characteristic values ​​corresponding to the corresponding abnormal emotions respectively, and the matching results corresponding to all physiological features are integrated to determine whether the driver has abnormal emotions.

[0085] Specifically, after intercepting the feature graph corresponding to each physiological feature, the feature graph can be extracted to obtain the feature value performance of each physiological feature of the driver before and after the abnormal emotion appears, such as the amplitude and change of skin electricity. Then compare the feature value of each physiological feature with the physiological feature value corresponding to the abnormal emotion. If the feature value of each physiological feature value matches the physiological feature value, it can be determined that the driver has abnormal emotion, otherwise the driver does not have abnormal emotion. In this way, the driver's abnormal emotion can be accurately and reliably identified by integrating various types of data such as the driver's image, heart rate and skin electricity.

[0086] In this embodiment, a neural network classifier is optionally used to extract features from the monitoring data. It should be understood that when the driver has abnormal emotions, there will be obvious trend changes in the characteristic graphs corresponding to the heart rate and skin conductivity. Therefore, it is possible to determine whether the driver has abnormal emotions by the changing trends of the characteristic graphs corresponding to these two physiological characteristics. However, the manual determination method is inefficient and easily affected by human factors, resulting in inaccurate determination results. Therefore, a trained neural network classifier can be used to automatically extract features from the intercepted characteristic graph to determine whether the driver has abnormal emotions.

[0087] Specifically, a convolutional neural network can be used to extract features from feature maps. The convolutional neural network includes an input layer, multiple convolutional layers, a pooling layer, a fully connected layer, and an output layer. The input layer is used to receive feature maps, and each convolutional layer can use different convolution kernels to extract local features of feature maps, such as edges, textures, and shapes. The pooling layer can average pool the extraction results of the convolution layer to reduce the feature dimension and enhance the robustness of the model. Its activation function can use the ReLU function. The fully connected layer can flatten the extracted features and classify them. The output layer can use the Softmax function to output the classification results according to task requirements.

[0088] In the training process of the convolutional neural network, the cross entropy loss function can be used to evaluate the training effect of the model. Specifically:

[0089]

[0090] Among them, N is the number of training samples, C is the number of categories, and y ij is the true label of the jth category of the i-th sample, p ijis the probability that the i-th sample belongs to the j-th category. The cross entropy loss function can effectively minimize the difference between the model's predicted probability and the true label, thereby improving the classification accuracy.

[0091] During the training process, you can choose optimization algorithms such as Adam, RMSProp or SGD to optimize the model, and use regularization such as Dropout and L1 / L2 to prevent overfitting in training.

[0092] like Figure 2 The flowchart of the vehicle distance control method based on the abnormal emotion of the driver is shown, and the control method includes:

[0093] Using the above-mentioned driver abnormal emotion recognition method, the driver's abnormal emotion is recognized in real time;

[0094] When the driver shows abnormal emotions, adjust the distance between the target vehicle and the vehicle in front until the driver's abnormal emotions disappear.

[0095] Specifically, firstly, the above-mentioned driver abnormal emotion recognition method can be used to determine in real time whether the driver has abnormal emotions. When the driver has abnormal emotions, the distance between the target vehicle driven by the driver and the vehicle in front can be adjusted, and the determination of whether the driver has abnormal emotions can be continued until the driver's abnormal emotions disappear.

[0096] The driver's abnormal emotion recognition method can quickly and accurately judge the driver's abnormal emotions, so as to timely adjust the distance between vehicles, alleviate the driver's abnormal emotions and avoid traffic accidents.

[0097] In this embodiment, the distance between the target vehicle and the vehicle in front can also be sensed in real time. When the driver has abnormal emotions, the distance between the current target vehicle and the vehicle in front can be compared with the distance between the target vehicle and the vehicle in front when the driver was in normal emotions before. If the distance changes significantly, it can be determined that the driver's abnormal emotions are caused by the change in distance, and the distance between the target vehicle and the vehicle in front can be adjusted to alleviate the driver's abnormal emotions. On the contrary, if the distance does not change significantly, it may be caused by other reasons, and the driver needs to be reminded.

[0098] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A method for identifying abnormal driver emotions, characterized in that: include: Obtain the driver's image monitoring data and various physiological monitoring data in real time; Determining whether the driver has abnormal emotions through the acquired image monitoring data, and in response to the driver having abnormal emotions, extracting corresponding monitoring data from each item of the physiological monitoring data; The extracted monitoring data are integrated to further verify the abnormal emotions of the driver.

2. The method for identifying abnormal driver emotions according to claim 1, characterized in that: Judging whether the driver has abnormal emotions through the image monitoring data includes: Extracting key points of the human body from the image monitoring data, and locating position information of different key points of the human body in the image monitoring data; According to the position information of the corresponding key points of the human body, determine the position of different parts of the human body; The abnormal emotion is determined by combining the position changes of all the human body parts.

3. The method for identifying abnormal driver emotions according to claim 2, characterized in that: Extracting key points of a human body from the monitoring image includes: Using a human body part detection algorithm to extract human body feature areas in the monitoring image; The human body key points in the human body feature area are detected by using the trained key point detection model, and the position information of different human body key points is determined.

4. The method for identifying abnormal driver emotions according to claim 3, characterized in that: The human body feature area in the monitoring image is extracted using a human body part detection algorithm, including: Performing Gaussian difference filtering on the monitoring image, and performing Gaussian difference processing on the processed monitoring image, marking key points corresponding to human body feature areas in the monitoring image; The marked monitoring image is subjected to illumination balancing processing, and the monitoring image subjected to illumination balancing processing is subjected to grayscale histogram processing, and the human body feature area is extracted through the grayscale histogram obtained by the processing.

5. The method for identifying abnormal driver emotions according to claim 3, characterized in that: The human body feature area is a facial area.

6. The method for identifying abnormal driver emotions according to claim 5, characterized in that: The abnormal emotion is determined by combining the position changes of the eyebrows and pupils in the facial area.

7. The method for identifying abnormal driver emotions according to claim 1, characterized in that: Extracting corresponding monitoring data from the physiological monitoring data includes: Drawing a time domain graph corresponding to the corresponding physiological characteristics based on the physiological monitoring data; According to the time when the driver shows abnormal emotions, the characteristic graph of the corresponding time period is intercepted from the time domain graph according to the specified time range as the monitoring data.

8. The method for identifying abnormal driver emotions according to claim 1, characterized in that: The extracted monitoring data are integrated to further verify the abnormal emotions of the driver, including: Extracting features from each of the monitoring data to obtain feature values ​​corresponding to different physiological features; The characteristic values ​​corresponding to each physiological feature are matched with the characteristic values ​​corresponding to the corresponding abnormal emotions respectively, and the matching results corresponding to all physiological features are integrated to determine whether the driver has abnormal emotions.

9. The method for identifying abnormal driver emotions according to claim 8, characterized in that: A neural network classifier is used to extract features from the monitoring data.

10. A vehicle distance control method based on abnormal driver emotions, characterized in that: include: Using the driver's abnormal emotion recognition method as described in any one of claims 1 to 9 to recognize the driver's abnormal emotions in real time; When the driver shows abnormal emotions, adjust the distance between the target vehicle and the vehicle in front until the driver's abnormal emotions disappear.

Images