Automobile starting method based on voiceprint and face unlocking

Abstract

The invention discloses an automobile starting method based on voiceprint and face unlocking, and relates to the field of automobile starting methods, and the technical scheme comprises the following steps: step 1, audio acquisition: detecting user voice data in real time, and processing the acquired audio data; 2, uploading the recorded voice data to a cloud server, performing semantic analysis on the voice data by the cloud server, obtaining an automobile starting instruction, and obtaining voiceprint feature information in the voice data after obtaining the starting instruction; and step 3, comparing the obtained voiceprint feature information with the voiceprint feature information registered by the user, obtaining the state of the user in two minutes during reminding through a rest reminding function so as to carry out deep learning, generating a fatigue driving judgment module, detecting the current time period of the user through the fatigue driving judgment module, and judging whether the user is in a fatigue driving state or not. Therefore, whether the user has fatigue driving such as sleepiness or not is judged, and reminding is given in time.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile starting methods, and in particular to a automobile starting method based on voiceprint and facial unlocking. Background Art

[0002] With the development of modern technology, smart cars have become the strategic development direction of the automotive industry, and the control requirements for vehicle posture are becoming increasingly higher.

[0003] A search revealed a utility model patent in China with the number CN206124986U, which discloses a vehicle starting device based on facial recognition and voiceprint cloud recognition. The device comprises: an identification component, including a microphone and camera, for capturing the vehicle owner's voice information and facial video image; a vehicle-mounted computer component, for acquiring the identification component's voice information and facial video image, and performing video signal conversion and decoding, as well as voice information compression and voiceprint information conversion; a cloud server, which communicates with the vehicle-mounted computer component, receives the decoded facial video image, compares it with registered facial features, and performs similarity recognition based on the received voiceprint information, returning the recognition result to the vehicle-mounted computer component; and an ECM, which communicates with the vehicle-mounted computer component via the CAN bus, obtains the recognition result, and performs the corresponding start / stop actions. Compared with existing technologies, this invention offers advantages such as dual recognition, accurate recognition, simple operation, low cost, fingerprint verification, and improved accuracy.

[0004] However, the above-mentioned device can only improve the accuracy of identity authentication through dual recognition. At present, the most commonly used smart car starting methods include voice start and facial start, but the recognition credibility of the two is low for starting. The speed and accuracy of facial voice recognition are low, and there is also a lack of combined use of sound, image and fingerprint. At the same time, it is also impossible to detect and remind the user's status in real time during driving. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the existing technology, such as low recognition credibility, low speed and accuracy of facial voice recognition, and inability to detect and remind the user's status in real time during driving. Therefore, a car starting method based on voiceprint and facial unlocking is proposed.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A car starting method based on voiceprint and facial unlocking, comprising the following steps:

[0008] Step 1: Audio collection: Real-time detection of user voice data and processing of the collected audio data;

[0009] Step 2: Upload the recorded voice data to the cloud server, which performs semantic analysis on the voice data to obtain the car start command, and then obtains the voiceprint feature information in the voice data after obtaining the start command;

[0010] Step 3: Compare the acquired voiceprint feature information with the user's registered voiceprint feature information. After verification, the result is sent to the vehicle control system, and the vehicle door is opened and closed through the vehicle control system;

[0011] Step 4: The vehicle control system extracts the recorded user facial image data and uploads it to the cloud server;

[0012] Step 5: The cloud server compares the received facial image data with the user's registered facial features, verifies any changes in eye features in the facial image data, and uses blink detection to identify whether the person is alive. If the verification is successful, the results are sent to the vehicle control system.

[0013] Step 6: The cloud server recognizes and judges the user's fingerprint information. The vehicle control system controls the remote ignition of the unmanned vehicle based on the received facial feature verification results, fingerprint comparison results, and voice recognition results. At the same time, a motion gesture recording unit is set up to give the user's motion characteristics specific functions;

[0014] Step 7: Utilize the image acquisition and processing unit to collect the driver's facial image data, especially the eye data, multiple times to realize the auxiliary function of reminding the driver to take a rest.

[0015] The above technical solution further includes:

[0016] Furthermore, during the audio detection process, when it is detected that the user starts to input sound, the voice data and the user's facial image data are recorded in real time. When it is detected that the user's voice input is completed, the recording of voice and image data is stopped. During the image recognition process, the user's dynamic image within 1.5 seconds is collected, and multiple groups of photos are fused to obtain the best image of the user's face.

[0017] Furthermore, in the audio detection, the cluttered audio data in the acquired audio data is first removed, and then the voice data of the car owner is repaired and enhanced.

[0018] Furthermore, the cloud server repairs the acquired facial image data during the process of comparing the facial image data with the facial features registered by the user, thereby reducing noise in the image and increasing texture in the image data.

[0019] Furthermore, the rest reminder function first obtains the user's status in 2 minutes when giving a reminder, thereby performing deep learning and generating a "fatigue driving judgment module".

[0020] Furthermore, every two minutes of the user is regarded as a time period, and the user's current time period is detected by the "fatigue driving determination module" to determine whether the user is driving fatigued such as drowsiness.

[0021] Furthermore, the speed of the unmanned vehicle can be limited after the ignition is remotely started.

[0022] Furthermore, the cloud storage is cleaned every 14 days to ensure the smooth operation of the system.

[0023] Furthermore, when performing blink detection, the face in the current frame image is detected, and the feature points used to calibrate the upper eyelid and the lower eyelid respectively are located, the fluctuation of the distance value between the feature points of the upper and lower eyelids on the face in the read frame image is counted, and based on the statistical fluctuations, it is determined whether the face in the face video is alive.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] 1. In the present invention, the cloud server performs semantic analysis on the voice data to obtain the car start-up command, and compares the obtained voiceprint feature information with the voiceprint feature information registered by the user. After verification, the result is sent to the vehicle control system, which improves the security of audio data verification.

[0026] 2. In the present invention, during the image recognition process, dynamic images of the user within 1.5 seconds are collected, and multiple groups of photos are fused to obtain the best image of the user's face. The cloud server compares the received facial image data with the facial features registered by the user, and verifies the eye change features in all facial image data. After verification, the results are sent to the vehicle control system, thereby improving the security of facial data verification.

[0027] 3. In the present invention, the rest reminder function first obtains the user's status in 2 minutes when giving a reminder, thereby performing deep learning and generating a "fatigue driving determination module". The "fatigue driving determination module" detects the user's current time period to determine whether the user is driving fatigued such as drowsiness, and gives a reminder in time. DETAILED DESCRIPTION

[0028] Example 1

[0029] The first step is to detect the user's voice data in real time and process the collected audio data. During the audio detection, the cluttered audio data in the acquired audio data is first removed, and then the driver's voice data is repaired and enhanced;

[0030] During the audio detection process, when the user starts to input voice, the voice data and the user's facial image data are recorded in real time. When the user's voice input is completed, the recording of voice and image data ends. During the image recognition process, the user's dynamic image within 1.5 seconds is collected, and multiple groups of photos are fused to obtain the best image of the user's face;

[0031] The second step is to upload the recorded voice data to the cloud server, which performs semantic analysis on the voice data to obtain the car start command. After obtaining the start command, the voiceprint feature information in the voice data is obtained. The cloud storage is cleaned every 14 days to ensure the smoothness of the system.

[0032] The third step is to compare the acquired voiceprint feature information with the user's registered voiceprint feature information. After verification, the result is sent to the vehicle control system, and the vehicle door is opened and closed through the vehicle control system;

[0033] In the fourth step, the vehicle control system extracts the recorded user facial image data and uploads it to the cloud server;

[0034] In the fifth step, the cloud server compares the received facial image data with the facial features registered by the user. During the comparison process, the cloud server repairs the acquired facial image data to reduce noise in the image and increase texture in the image data.

[0035] Next, blink detection is used to identify whether the person is alive. When performing blink detection, the face in the current frame image is detected, and the feature points used to calibrate the upper and lower eyelids are located. The fluctuation of the distance value between the feature points of the upper and lower eyelids on the face in the read frame image is counted. Based on the statistical fluctuation, it is determined whether the face in the face video is alive;

[0036] If the face is judged to be alive, the following operations are performed. If it is judged to be not alive, the face unlock function is disabled and then the detection is restarted.

[0037] The cloud server recognizes and judges the user's fingerprint information. The vehicle control system controls the remote ignition of the unmanned vehicle based on the received facial feature verification results, fingerprint comparison results, and voice recognition results. At the same time, a motion posture recording unit is set up to give the user's motion characteristics specific functions.

[0038] Step 6: After receiving the facial feature verification result, the vehicle control system controls the remote ignition of the unmanned vehicle, and then limits the speed of the vehicle;

[0039] Step 7: Using the image acquisition and processing unit, the driver's facial image data, especially eye data, is collected multiple times to implement an auxiliary function of reminding the driver to take a rest.

[0040] The rest reminder function first obtains the user's status for 2 minutes before issuing a reminder, performs deep learning, and generates a "fatigue driving judgment module";

[0041] Every two minutes of the user's time is regarded as a time period, and the user's current time period is detected through the "fatigue driving judgment module" to determine whether the user is driving fatigued such as drowsiness.

[0042] Example 2

[0043] The first step is to detect the user's voice data in real time and process the collected audio data. During the audio detection, the cluttered audio data in the acquired audio data is first removed, and then the driver's voice data is repaired and enhanced;

[0044] During the audio detection process, when it is detected that the user has started to input voice, the voice data and the user's face image data are recorded in real time, and the recording of voice and image data is stopped when it is detected that the user's voice input is completed;

[0045] The second step is to upload the recorded voice data to the cloud server, which performs semantic analysis on the voice data to obtain the car start command. After obtaining the start command, the voiceprint feature information in the voice data is obtained. The cloud storage is cleaned every 14 days to ensure the smoothness of the system.

[0046] The third step is to compare the acquired voiceprint feature information with the user's registered voiceprint feature information. After verification, the result will be sent to the vehicle control system;

[0047] In the fourth step, the vehicle control system extracts the recorded user facial image data and uploads it to the cloud server;

[0048] In the fifth step, the cloud server compares the received facial image data with the facial features registered by the user. During the comparison process, the cloud server repairs the acquired facial image data to reduce noise in the image and increase texture in the image data.

[0049] It also verifies changes in eye features in all facial image data, and sends the results to the vehicle control system after verification.

[0050] Step 6: After receiving the facial feature verification result, the vehicle control system controls the remote ignition of the unmanned vehicle, and then limits the speed of the vehicle;

[0051] The seventh step is to use the image acquisition and processing unit to collect the driver's facial image data, especially the eye data, multiple times to realize the auxiliary function of reminding the driver to rest.

[0052] Example 3

[0053] The first step is to detect the user's voice data in real time and process the collected audio data. During the audio detection process, when the user starts to input voice, the voice data and the user's face image data are recorded in real time. When the user's voice input is completed, the recording of voice and image data ends;

[0054] The second step is to upload the recorded voice data to the cloud server, which performs semantic analysis on the voice data to obtain the car start command, and then obtains the voiceprint feature information in the voice data after obtaining the start command;

[0055] The third step is to compare the acquired voiceprint feature information with the user's registered voiceprint feature information. After verification, the result will be sent to the vehicle control system;

[0056] In the fourth step, the vehicle control system extracts the recorded user facial image data and uploads it to the cloud server;

[0057] In the fifth step, the cloud server compares the received facial image data with the facial features registered by the user;

[0058] It also verifies changes in eye features in all facial image data, and sends the results to the vehicle control system after verification.

[0059] Step 6: After receiving the facial feature verification result, the vehicle control system controls the remote ignition of the unmanned vehicle, and then limits the speed of the vehicle;

[0060] The seventh step is to use the image acquisition and processing unit to collect the driver's facial image data, especially the eye data, multiple times to realize the auxiliary function of reminding the driver to rest.

[0061] 40 users were selected to experience the car starting methods in Example 1, Example 2, and Example 3. The results of the experience are shown in the following table:

[0062]

[0063]

[0064] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A car starting method based on voiceprint and facial unlocking, characterized in that: The following steps are involved: Step 1: Audio collection: Real-time detection of user voice data and processing of the collected audio data; Step 2: Upload the recorded voice data to the cloud server, which performs semantic analysis on the voice data to obtain the car start command, and then obtains the voiceprint feature information in the voice data after obtaining the start command; Step 3: Compare the acquired voiceprint feature information with the user's registered voiceprint feature information. After verification, the result is sent to the vehicle control system, and the vehicle door is opened and closed through the vehicle control system; Step 4: The vehicle control system extracts the recorded user facial image data and uploads it to the cloud server; Step 5: The cloud server compares the received facial image data with the user's registered facial features, verifies any changes in eye features in the facial image data, and uses blink detection to identify whether the person is alive. If the verification is successful, the results are sent to the vehicle control system. Step 6: The cloud server recognizes and judges the user's fingerprint information. The vehicle control system controls the remote ignition of the unmanned vehicle based on the received facial feature verification results, fingerprint comparison results, and voice recognition results. At the same time, a motion gesture recording unit is set up to give the user's motion characteristics specific functions; Step 7: Utilize the image acquisition and processing unit to collect the driver's facial image data, especially the eye data, multiple times to realize the auxiliary function of reminding the driver to take a rest.

2. The car starting method based on voiceprint and facial unlocking according to claim 1, characterized in that: During the audio detection process, when it is detected that the user starts to input sound, the voice data and the user's facial image data are recorded in real time. When it is detected that the user's voice input is completed, the recording of voice and image data is stopped. During the image recognition process, the user's dynamic image within 1.5 seconds is collected, and multiple groups of photos are fused to obtain the best image of the user's face.

3. The car starting method based on voiceprint and facial unlocking according to claim 2, characterized in that: In the audio detection, the cluttered audio data in the acquired audio data is first removed, and then the voice data of the car owner is repaired and enhanced.

4. The car starting method based on voiceprint and facial unlocking according to claim 3, characterized in that: The cloud server repairs the acquired facial image data during the process of comparing the facial image data with the facial features registered by the user, reduces the noise in the image, and increases the texture in the image data.

5. The car starting method based on voiceprint and facial unlocking according to claim 4, characterized in that: When giving a rest reminder, the function first obtains the user's status in the past 2 minutes, performs deep learning, and generates a "fatigue driving judgment module." 6. The car starting method based on voiceprint and facial unlocking according to claim 5, characterized in that: Every two minutes of the user's time is considered a time period, and the "Fatigue Driving Determination Module" detects the user's current time period to determine whether the user is driving while fatigued, such as drowsy.

7. The car starting method based on voiceprint and facial unlocking according to claim 6, characterized in that: The unmanned vehicle can limit its speed after being remotely ignited.

8. The car starting method based on voiceprint and facial unlocking according to claim 7, characterized in that: The cloud storage is cleaned every 14 days to ensure the smooth operation of the system.

9. The car starting method based on voiceprint and facial unlocking according to claim 7, characterized in that: When performing blink detection, the face in the current frame image is detected, and the feature points used to calibrate the upper eyelid and lower eyelid respectively are located. The fluctuation of the distance value between the feature points of the upper and lower eyelids on the face in the read frame image is counted, and based on the statistical fluctuation, it is determined whether the face in the face video is alive.

Images