WO2019109412A1

WO2019109412A1 - Active rescue calling method for accident, and vehicle-mounted automatic help-seeking system

Info

Publication number: WO2019109412A1
Application number: PCT/CN2017/118473
Authority: WO
Inventors: 段拥政; 张猛; 黄锦昌; 赵振峰
Original assignee: 惠州市德赛西威汽车电子股份有限公司
Priority date: 2017-12-04
Filing date: 2017-12-26
Publication date: 2019-06-13
Also published as: CN107933471A; CN107933471B

Abstract

An active rescue calling method for an accident, and a vehicle-mounted automatic help-seeking system. The system comprises a GPS module, a sound collection module, and a processing unit. The method comprises: collecting real-time video data in a vehicle (S10); intercepting frame images in the video data, and analyzing a real-time sitting posture of a person at a driving seat (S20); and determining whether a difference between the real-time sitting posture and a calibrated sitting posture reaches a deviation threshold, and if a duration of continuously reaching the threshold exceeds a first preset time period, invoking a wireless communication module to send a help-seeking signal (S30). The present invention has the beneficial advantages: the actual situation of a driver after a vehicle accident is determined based on an intelligent image recognition capability, and rescue information can be more accurately sent. By means of a scheme of establishing a panoramic survey system in the vehicle and by means of recognition and voice recognition, rescue information can be actively sent after it can be determined a vehicle is hijacked.

Description

Emergency call rescue method and onboard automatic help system

Technical field

The invention relates to an automobile intelligent alarm system, in particular to an accident active call rescue method and an onboard automatic rescue system and device.

Background technique

In China, the number of deaths due to road traffic safety accidents is between 50,000 and 60,000, and the number of people who die without timely assistance within one hour after the accident is as high as 60%. When a serious traffic accident occurs in a vehicle, especially when the location of the accident is relatively remote, timely call assistance is of great significance to the life of the personnel in the vehicle. In the event that the collision of the accident vehicle is serious and the personnel in the vehicle cannot call for help, a reliable vehicle active calling technology is needed to provide the calling service. In addition, the criminal activities for vehicles and passengers are also increasing. When the passengers encounter the incident of looting vehicles and belongings, they may not be able to send a rescue signal safely.

Most of the automatic SOS systems in the prior art determine whether they are in a traffic accident through signals such as airbag damage, such as airbags and vehicle attitudes, but most of the situations can be rescued by the driver after an accident, so In the case of over-reporting in a case, the automatic help-seeking system only makes sense when the driver cannot or cannot ask for help.

Summary of the invention

In order to solve the above technical problems, the present invention provides a method for accidental active call rescue and an onboard automatic rescue system and device.

A method for active call rescue of an accident, comprising the following steps:

S10, collecting real-time video data in the car;

S20, intercepting a frame image in the video data, and analyzing a real-time sitting posture of the driver's seat;

S30, determining whether the difference between the real-time sitting posture and the calibration sitting posture reaches a deviation threshold value, if the duration of the threshold value is more than the first preset time period, step S40 is performed;

S40. Calling the wireless communication module to send a distress signal.

Further, the step S20 includes the following sub-steps:

S21: Painting a driver's real-time sitting posture by corner detection, and determining a key identification point therefrom;

S22. Calculate a real-time position proportional relationship between key identification points;

S23. Comparing the real-time position proportional relationship with the position proportional relationship of the calibration sitting posture;

S24. Calculate the variation amount.

Further, the step key identification point includes at least one of a facial features, a shoulder contour, or a head contour.

Further, the step S10 further includes an audio collection step:

S101. Acquire an in-car audio signal and perform noise reduction processing;

S102. Analyze the voice and decompose the keyword;

S103. When the decomposed keyword matches the preset keyword, step S40 is directly performed.

Further, the method further includes establishing a real-time video communication with the outside.

Further, performing step S30 further includes acquiring the real-time situation of the vehicle body image:

S301: Obtain at least one image in the airbag area, the A-pillar area of the automobile, and the front area of the vehicle;

S302. Analyze the key area of the image in step S301 and compare it with the calibration image.

S303. If the difference exceeds the deviation threshold, step S40 is performed.

Further, the distress signal in the step S40 includes at least one of personal information preset by the driver, satellite positioning, and real-time video data.

In addition, based on the above-mentioned accident active call rescue method, the present invention further provides an onboard automatic help system, including:

a GPS module for obtaining vehicle location information;

a camera module for acquiring image information in and around the vehicle;

a sound collection module for acquiring sound information in the vehicle;

a processing unit, configured to receive and analyze information output by the camera module and the sound collection module, and determine whether an accident occurs;

The wireless communication module is configured to forward the distress signal sent by the processing unit.

Further, the camera module is mounted on the interior rear view mirror of the automobile.

Further, the camera module is a look-around camera module, and can obtain image information of a person in the vehicle, a car A main body, and a front cover direction.

The method for the accident active call rescue of the present invention and the beneficial effects of the onboard automatic help system and system include:

1. Based on the intelligent image recognition ability to judge the actual situation of the driver after the vehicle accident, the rescue information can be more accurately transmitted;

2. By implementing a scheme for constructing a panoramic viewing system inside the vehicle, as well as identification and voice recognition, it can be determined that the vehicle actively sends rescue information after the vehicle is hijacked.

DRAWINGS

FIG. 1 is a flowchart of a method in Embodiment 1 of the present invention.

2 is a schematic diagram of a method in Embodiment 2 of the present invention.

FIG. 3 is a system architecture diagram of Embodiment 3 of the present invention.

Detailed ways

The preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings, in which the advantages and features of the invention are more readily understood by those skilled in the art.

Example 1:

As shown in FIG. 1 , the embodiment discloses a method for emergency call rescue of an accident, which is mainly used for determining a traffic accident and automatically calling a rescue, and specifically includes the following steps:

S10. Collect real-time video data in the car, and collect image information in the car in real time through the camera module. In addition, in order to reduce the audible sound of the misjudgment, the video data is collected after the car is started, and when the seat belt of the main driving position is detected.

S20: intercepting a frame image in the video data, and analyzing a real-time sitting posture of the driver's seat. Usually in the normal driving process, due to the limitation of the car seat, the driver's sitting posture is relatively fixed. Only when an accident causes the driver to lose consciousness, the driver's sitting posture will undergo a relatively long time change.

Specifically, the steps of analyzing the driver's real-time sitting posture are as follows:

S21: The driver's real-time sitting posture is drawn through the corner point detection, and the key identification point is determined therefrom. First, the contour of the driver is drawn by the method of corner detection, and the image is extracted according to the contour. The determination of the key identification point is performed after the driver profile is extracted. In this embodiment, the key identification point may be, but is not limited to, one of a facial features, a shoulder contour, or a head wheel.

The sitting posture can be, but is not limited to, the hand movement, the head movement, the body position, etc. Since the frequency of the hand movement is variable during driving, the present embodiment focuses on the face, the head movement, and the body position. Elements.

By depicting the shoulder profile to identify whether the driver is in a normal driving state, in a normal driving state, the shoulder profile should be substantially symmetrical and the head will look forward. If an accident causes loss of consciousness, the head and body will lose support and will continue to fall to one side.

In addition, face recognition can be judged based on the face recognition, and the facial features can be removed from the facial features. When an accident occurs, the facial features will change greatly, such as overall deflection, closed eyes or injured bleeding.

S22. Calculating a real-time position proportional relationship between key identification points,

In the above two steps, the key identification points separated in step S21 are calculated, such as shoulder contour information, and the positional relationship will change after tilting. Data such as tilt angle. At the same time, the person's side head, head down, etc. will produce a relative position deviation. After the calculation by the relevant algorithm, the positional relationship with the position of the calibration sitting position is compared.

For example, under normal circumstances, the driver's shoulder connection should be close to the horizontal level, so the angle between the connection and the horizontal line is greater than 10 °, if the driver has an accident and is unconscious, his body The position will be tilted, even reaching 30° or more.

S24. Calculate the variation amount. The actual obtained data is calculated with the calibrated data, and finally the amount of deviation is obtained.

S30. If the deviation amount indicates that the difference between the real-time sitting posture and the calibration sitting posture is too large, and the deviation threshold is reached, it is determined that an accident is suspected. The deviation threshold is set according to different judgment objects. It can be understood that if the judgment object is a shoulder contour, the deviation threshold is more than 10°. If it is a facial feature, it can be tilted from the whole angle, and the facial features in the image The distance and other aspects to set.

In addition, if the duration of the threshold value is continuously exceeded for the first predetermined time period, such as 1 minute, etc., it may be determined that the driver has an abnormality, and step S40 is performed.

S40. Calling the wireless communication module to send a distress signal. The distress signal includes at least one of personal information preset by the driver, satellite positioning, and real-time video data.

In addition, when the passengers encounter the incident of looting vehicles and property, they may not be able to safely send out rescue signals. Vehicles are required to provide alarm calls and video materials to help combat theft, robbery of motor vehicle criminal activities, and to maintain the safety and interests of the owners. Therefore step S10 may also include an audio collection step:

S101. Acquire an in-car audio signal and perform noise reduction processing;

S102. Analyze the voice and decompose the keyword;

S103: Match the decomposed keyword with a preset keyword, and the preset keyword may be a vocabulary, a tone, and the like commonly used by a robber to implement robbery.

And through the image algorithm to identify the facial expressions and relative movements of the insiders, and recognize the horror expressions and hijacking actions of the insiders, the basic analysis is combined with the semantics, and it is possible to determine whether there is a crime at this time and determine the crime situation. After the occurrence, the recording or video is sent to the server of the service provider or the police through the wireless communication module, and the relevant semantics are finally confirmed.

When the service provider or the police finally determines, it can also control the system inside the vehicle to establish real-time video communication with the outside, leaving evidence of crime and monitoring the situation inside the car.

Example 2:

The present embodiment is further optimized on the basis of the first embodiment. As shown in FIG. 2, in order to improve the accuracy of the accident determination, in the step S30 of the embodiment, when it is determined that the accident is an accident, the real-time situation of the vehicle body image can be performed. step:

S301: Acquire at least one image in an airbag area, a car A-pillar area, and a front area. In the event of a collision, these three locations are usually damaged or changed more seriously, and can be assisted by obtaining image information of the above three regions.

S303. When the difference exceeds the deviation threshold, the suspected accident may be determined as a traffic accident, and step 40 is directly executed.

Example 3:

In addition, based on the accident active call rescue method of the above embodiment, the embodiment further provides an onboard automatic help system, as shown in FIG. 3, comprising: a GPS module, a GPS module, a sound collection module, and a processing unit.

The GPS module and the processing unit are connected through a communication interface, and are used to obtain current location information of the automobile.

The camera module and the processing unit are connected through a communication interface for acquiring image information in and around the vehicle. In this embodiment, the camera is mounted on the rearview mirror of the automobile, and includes a front wide-angle camera and a rear wide-angle camera. The two are combined to form an in-vehicle viewing camera, which can acquire image information of the interior personnel, the automobile A main body, and the front cover direction. .

The sound collection module is connected to the processor unit through an audio signal interface for acquiring in-vehicle sound information.

The wireless communication module and the processing unit are connected through a data transmission interface and a communication interface, and are used for forwarding a distress signal sent by the processing unit.

The processing unit is configured to receive and analyze information output by the camera module and the sound collection module, and determine whether an accident occurs.

The embodiments of the present invention have been described in detail above with reference to the drawings, but the present invention is not limited to the above-described embodiments, and various modifications may be made without departing from the spirit of the invention. Kind of change.

Claims

A method for emergency call rescue of an accident, comprising: the following steps:

S10, collecting real-time video data in the car;

S20, intercepting a frame image in the video data, and analyzing a real-time sitting posture of the driver's seat;

S30, determining whether the difference between the real-time sitting posture and the calibration sitting posture reaches a deviation threshold value, if the duration of the threshold value is more than the first preset time period, step S40 is performed;

S40. Calling the wireless communication module to send a distress signal.
The method for emergency call assistance according to claim 1, wherein the step S20 comprises the following substeps:

S21: Painting a driver's real-time sitting posture by corner detection, and determining a key identification point therefrom;

S22. Calculate a real-time position proportional relationship between key identification points;

S23. Comparing the real-time position proportional relationship with the position proportional relationship of the calibration sitting posture;

S24. Calculate the variation amount.
The method according to claim 2, wherein the step key identification point comprises at least one of a facial features, a shoulder contour or a head contour.
The method for emergency call assistance according to claim 1, wherein the step S10 further comprises an audio collection step:

S101. Acquire an in-car audio signal and perform noise reduction processing;

S102. Analyze the voice and decompose the keyword;

S103. When the decomposed keyword matches the preset keyword, step S40 is directly performed.
The method of emergency call assistance according to claim 4, further comprising the step of establishing real-time video communication with the outside.
The method for emergency call assistance according to claim 1, wherein the performing step S30 further comprises acquiring the real-time situation of the vehicle body image:

S301: Obtain at least one image in the airbag area, the A-pillar area of the automobile, and the front area of the vehicle;

S302. Analyze the key area of the image in step S301 and compare it with the calibration image.

S303. If the difference exceeds the deviation threshold, step S40 is performed.
The method for emergency call assistance according to claim 1, wherein the distress signal in step S40 comprises at least one of personal information preset by the driver, satellite positioning, and real-time video data.
An in-vehicle automatic help-seeking system based on the accident active call rescue method according to any one of claims 1 to 7, characterized in that it comprises:

a GPS module for obtaining vehicle location information;

a camera module for acquiring image information in and around the vehicle;

a sound collection module for acquiring sound information in the vehicle;

a processing unit, configured to receive and analyze information output by the camera module and the sound collection module, and determine whether an accident occurs;

The wireless communication module is configured to forward the distress signal sent by the processing unit.
The in-vehicle automated help-seeking system according to claim 8, wherein said camera module is mounted on said interior mirror of the automobile.
The on-vehicle automatic help-seeking system according to claim 8, wherein the camera module is a look-around camera module, and can acquire image information of a person in the vehicle, a car A main body, and a front cover direction.