Automobile anti-theft method and system based on automobile monitoring image recognition

Abstract

The invention discloses an automobile anti-theft method and system based on automobile monitoring image recognition, and the method comprises the following steps: generating a monitoring region, and carrying out the image collection of the region, so as to obtain a plurality of groups of monitoring images on the same timestamp; processing the monitoring image to judge a target action; if the action of the target is judged to be the car stealing action, enabling the car end to enter an anti-theft mode, and sending the car information to a communication end of a car owner; through careful monitoring of a monitoring area, image acquisition of a target action is realized, a monitoring image is used as a data source to realize processing and judgment, when a target is suspected to steal a vehicle, a vehicle end enters an anti-theft mode to prevent the vehicle from being lost, and a user can be informed at the first time, so that the user experience is improved, and the user experience is improved. The monitoring system and the anti-theft system cooperate to realize precise theft prevention.

Description

Technical Field

[0001] The present invention relates to the field of vehicle anti-theft, and particularly to a vehicle anti-theft method and system based on vehicle monitoring image recognition. Background Art

[0002] In modern society, vehicles have become an indispensable means of transportation in people's daily lives. Vehicle theft poses a great threat to people's property safety and social stability. Conventional anti-theft solutions usually use mechanical locks or electronic anti-theft devices. However, in actual use, it is found that mechanical locks are easily damaged by violence, and electronic anti-theft devices may be interfered with or cracked by signals. Therefore, traditional anti-theft means are difficult to provide sufficient security guarantees. On the other hand, existing vehicles are all equipped with monitoring systems. In theory, the monitoring system can most intuitively determine the actions of the target, but the monitoring system and the anti-theft system perform their respective functions and there is no good cooperation plan. Based on this, a vehicle anti-theft method that is linked to the monitoring images presented by the monitoring system is needed in society. Summary of the Invention

[0003] Aiming at the technical problem that the anti-theft system of vehicles in the prior art is relatively rigid and performs its own functions separately from the monitoring system, resulting in a relatively single anti-theft system, the present invention provides a solution.

[0004] To achieve the above object, the present invention provides a vehicle anti-theft method based on vehicle monitoring image recognition, including the following steps: Generate a monitoring area and collect images of the area to obtain multiple groups of monitoring images at the same time stamp; Process the monitoring images to identify the target action; if the action of the target is determined to be a vehicle theft action, the vehicle terminal enters the anti-theft mode, and the vehicle information is sent to the communication terminal of the vehicle owner.

[0005] As an improved solution of the present invention, before processing the monitoring images to identify the target action, the following steps are further included: Analyze the content of multiple groups of the monitoring images to determine the target pattern; Calculate the screen proportion of any one of the target patterns in its corresponding monitoring image, and summarize to generate a proportion parameter. If the proportion parameter is greater than a preset standard proportion parameter, then: Calculate the proportion time of the proportion parameter at the time stamp. If the proportion time is greater than a preset standard proportion time, enter the sensitive mode to identify the action of the target.

[0006] As an improved solution of the present invention, generate a database, input the original data of multiple groups of the monitoring images into a deep learning model, and enable the model to perform data matching in association with the database, so as to determine the action of the target.

[0007] As an improved solution of the present invention, before inputting the original data into the deep learning model, it is also necessary to perform pre-annotation processing on the data, and the steps are as follows: Selectively blur the background area of any one of the monitoring images to selectively separate the main body; Save the action animations formed by multiple groups of image frames of the separated main bodies to form annotation data; During the processing of the deep learning model, it also includes: When outputting the result of the action of the target as a determination, perform a confidence evaluation on the result output to obtain a first confidence value; and, Input the label data into the deep learning model for processing, and at the same time generate and output an auxiliary result, and perform a confidence evaluation on the auxiliary result to obtain a second confidence value.

[0008] As an improved solution of the present invention, before executing the anti-theft method, if a pre-detected authorized driving account is logged in, the anti-theft mode is not executed.

[0009] As an improved solution of the present invention, in the anti-theft mode, always collect the live condition of the target to generate media information, upload the media information to the cloud and send it to the communication terminal of the vehicle owner.

[0010] As an improved solution of the present invention, the media information is one of video, voice or electronic photo.

[0011] As an improved solution of the present invention, in the anti-theft mode stage, the vehicle terminal plays a warning message to warn the target.

[0012] As an improved solution of the present invention, the steps of the vehicle terminal playing a warning message: The communication terminal collects the warning audio of the vehicle owner and uploads it to the server; Download the warning audio to the vehicle terminal for playing.

[0013] This application also provides an automobile anti-theft system based on automobile monitoring image recognition for executing the method described above. The system includes a collection module, a processing module, an anti-theft module and a communication module; The collection module generates a monitoring area and collects images of the area to obtain multiple groups of monitoring images at the same time stamp; The processing module processes through the monitoring images to determine the target action; if the action of the target is determined to be a car theft action, it controls the anti-theft module to make the vehicle end enter the anti-theft mode, and sends the vehicle information to the communication terminal of the vehicle owner through the communication module.

[0014] The beneficial effects of the present invention are as follows: Compared with the prior art, the present invention provides an automobile anti-theft method and system based on automobile monitoring image recognition. The method includes the following steps: generating a monitoring area and collecting images of the area to obtain multiple groups of monitoring images at the same timestamp; processing through the monitoring images to determine the target action; if the action of the target is determined to be a car theft action, making the vehicle end enter the anti-theft mode, and sending the vehicle information to the communication terminal of the vehicle owner; realizing image collection of the target action by closely monitoring the monitoring area, and based on the monitoring images as the data source, processing and judging them. When the target has a suspicion of stealing a car, the vehicle end is made to enter the anti-theft mode to prevent the vehicle from being lost, and the user can be informed in the first time. It can be seen that the monitoring system and the anti-theft system cooperate with each other to achieve precise anti-theft. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Attached Figure 1 is a flowchart of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0017] In order to describe the present invention more clearly, the present invention will be further described below with reference to the accompanying drawings.

[0018] In the following description, specific details are given to provide a deeper understanding of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. It should be understood that the specific embodiments are only used to explain the present invention, and are not used to limit the present invention.

[0019] It should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of the described features, wholes, steps, operations, elements or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components or combinations thereof.

[0020] Aiming at the technical problem that the anti-theft system of automobiles in the prior art is relatively rigid, and the anti-theft system is relatively single because it performs its own functions separately from the monitoring system, the present application provides an automobile anti-theft method based on automobile monitoring image recognition. Please refer to Attached Figure 1, The method includes the following steps: generating a monitoring area and collecting images of the area to obtain multiple groups of monitoring images at the same timestamp; processing the monitoring images to identify the target action; if the target's action is determined to be a car theft action, then causing the vehicle terminal to enter the anti-theft mode and sending the vehicle information to the communication terminal of the vehicle owner; through the careful monitoring of the monitoring area, the image collection of the target action is realized, and based on the monitoring images as the data source, the processing and judgment are carried out. When the target has the suspicion of stealing a car, the vehicle terminal enters the anti-theft mode to prevent the vehicle from being lost, and the user can be informed in the first time. It can be seen that the monitoring system and the anti-theft system cooperate with each other to achieve accurate anti-theft.

[0021] In order to improve the accuracy of identifying the target action, in this embodiment, before processing the monitoring images to identify the target action, the following steps are further included: Analyzing the content of multiple groups of monitoring images to determine the target pattern; Calculating the screen ratio of any target pattern in its corresponding monitoring image, summarizing to generate a ratio parameter. If the ratio parameter is greater than the preset standard ratio parameter, then: calculating the ratio time of the ratio parameter at the timestamp. If the ratio time is greater than the preset standard ratio time, then enter the sensitive mode to identify the target's action; In this embodiment, since pedestrians will inevitably pass by when the vehicle is parked at the vehicle terminal, in order to prevent misjudgment caused by the stay or passing of pedestrians, based on the situation that the target will stay for a long time when performing the car theft action, because it is necessary to keep the relative position with the door to be opened fixed. In this case, the preset standard ratio parameter and the standard ratio time are used as the analysis parameter quantities to initially judge the target's action, effectively preventing directly entering the processing stage to prevent misjudgment and starting the anti-theft mode, which causes trouble to the vehicle owner.

[0022] In the process of discriminating multiple surveillance images, traditional machine learning can be adopted. As an important embodiment, the process of determining multiple groups of surveillance images is as follows: The foregoing method includes: generating a database, inputting the original data of multiple groups of surveillance images into a deep learning model, enabling the deep learning model to associate with the database for data matching to determine the actions of the target; It is not difficult to understand that the database is mainly divided into two parts: the car theft action dataset part and the normal action recognition dataset part. The action recognition dataset part can use datasets such as UCF101 or HMDB51. The combination of the two parts can provide a good data foundation for the deep learning model and then output accurate answers; The deep learning model can adopt a 2DCNN-based model, an RNN-based model, or a Transformer-based model. Under the database of the above datasets, it can quickly survey and determine people, quickly obtain whether the target has the suspicion of car theft, and once it appears, enter the anti-theft mode to lock the vehicle end to prevent the car owner from suffering the loss of losing the car.

[0023] In a further solution, before inputting the original data into the deep learning model, it is also necessary to perform pre-annotation processing on the data. The steps are as follows: Selectively blur the background area of any surveillance image to selectively separate the main body; Save the action animations formed by multiple groups of images of the separated main bodies to form annotation data; During the processing of the deep learning model, it also includes: When outputting the result of the target's action as a determination, perform a confidence evaluation on the result output to obtain a first trust value; and input the label data into the deep learning model for processing, and at the same time generate and output an auxiliary result, and perform a confidence evaluation on the auxiliary result to obtain a second trust value.

[0024] It is not difficult to understand that selectively blurring the surveillance image can obtain purer action data, which is convenient for the deep learning model to analyze. In the subsequent process, although the second trust value is used as the discrimination criterion, technicians can use the first trust value to judge the second trust value as a data reference to judge whether the accuracy of the deep learning model used can reach the expected value. If it is lower than the expected value, the model type of the deep learning model can be replaced to ensure the accuracy of the output result.

[0025] In this embodiment, before executing the anti-theft method, if a pre-detected authorized driving account is logged in, the anti-theft mode is not executed. In this case, it is understood that the car owner performs an operation so that the vehicle end will not be too sensitive, resulting in a poor car-using experience for the car owner.

[0026] In this embodiment, in the anti-theft mode, the live condition of the target is collected at all times to generate media information, and the media information is uploaded to the cloud and sent to the communication terminal of the vehicle owner, which is convenient to be used as important video evidence in subsequent processing. In a specific solution, the media information is one of video, voice or electronic photo.

[0027] In this embodiment, during the anti-theft mode stage, the vehicle terminal plays a warning message to warn the target, which can immediately warn the target and prevent the subsequent events from deteriorating. In a specific solution, the steps for the vehicle terminal to play the warning message are as follows: The communication terminal collects the warning audio of the vehicle owner and uploads it to the server; the warning audio is downloaded to the vehicle terminal for playback; It is not difficult to understand that since the existing vehicle terminal obtains network traffic based on the Internet of Things protocol and cannot directly make calls between mobile phone terminals, therefore, in order to be able to play the warning audio in time to warn the target, through the above solution, timely admonition of the target is realized, and further deterioration of subsequent theft events is avoided.

[0028] This application also provides an automotive anti-theft system based on automotive monitoring image recognition. The system includes a collection module, a processing module, an anti-theft module, and a communication module; The collection module generates a monitoring area and performs image collection on the area to obtain multiple groups of monitoring images at the same time stamp; The processing module processes the monitoring images to determine the target action; if the action of the target is determined to be a car theft action, it controls the anti-theft module to make the vehicle terminal enter the anti-theft mode, and sends the vehicle information to the communication terminal of the vehicle owner through the communication module; By the cooperation of the collection module, the processing module, the anti-theft module and the communication module, the foregoing method is implemented. This embodiment has the same step processing method as the foregoing embodiment, so it has the same beneficial effects and functions as the foregoing embodiment, and thus will not be elaborated here.

[0029] The advantages of the present invention are as follows: By closely monitoring the monitoring area, image collection of the target action is realized, and based on the monitoring images as the data source, processing and judgment are performed on them. When the target has a suspicion of car theft, the vehicle terminal is made to enter the anti-theft mode to prevent the vehicle from being lost, and the user can be informed in the first time. It can be seen that the monitoring system and the anti-theft system are coordinated to achieve precise anti-theft.

[0030] The above discloses only several specific embodiments of the present invention, but the present invention is not limited thereto. Any changes that can be thought of by those skilled in the art should fall within the protection scope of the present invention.

Claims

1. A car anti-theft method based on car monitoring image recognition, characterized in that: The following steps are involved: Generate a monitoring area and collect images of the area to obtain multiple groups of monitoring images at the same time stamp; Processing the monitoring image to identify the target action; If the target's action is determined to be a car theft action, the vehicle end is made to enter an anti-theft mode, and the vehicle information is sent to the communication end of the car owner.

2. The method for preventing automobile theft based on automobile monitoring image recognition according to claim 1, characterized in that: Before the monitoring image is processed to identify the target action, the following steps are also included: Analyzing the contents of the plurality of groups of monitoring images to determine a target pattern; Calculate the screen proportion of any target pattern in the surveillance image in which it is located, and summarize to generate a proportion parameter. If the proportion parameter is greater than a preset standard proportion parameter, then: The proportion time of the proportion parameter on the timestamp is calculated. If the proportion time is greater than a preset standard proportion time, a sensitive mode is entered to identify the action of the target.

3. The method for preventing automobile theft based on automobile monitoring image recognition according to claim 2, characterized in that: The steps of processing the monitoring image to identify the target action are as follows: Generate a database, input multiple groups of raw data of the monitoring images into the deep learning model, and associate the model with the database for data matching, so as to determine the action of the target.

4. The method for preventing automobile theft based on automobile monitoring image recognition according to claim 3 is characterized in that: Before the raw data is input into the deep learning model, the data needs to be annotated in advance, and the steps are as follows: Selectively blurring the background area of ​​any of the surveillance images to selectively separate the subject; Saving the action animation formed by the plurality of groups of image frames of the separated subjects to form annotation data; The deep learning model processing process also includes: While outputting a result of the action of the target as a judgment, a confidence evaluation is performed on the result output to obtain a first trust value; as well as, The label data is input into the deep learning model for processing, and an auxiliary result is generated for output at the same time, and a confidence evaluation is performed on the auxiliary result to obtain a second trust value.

5. A car anti-theft method based on car monitoring image recognition according to any one of claims 1 to 4, characterized in that: Before executing the anti-theft method, if an authorized driving account is pre-detected for login, the anti-theft mode will not be executed.

6. The method for preventing automobile theft based on automobile monitoring image recognition according to claim 1, characterized in that: In the anti-theft mode, the real-time situation of the target is collected at all times to generate media information, which is uploaded to the cloud and sent to the communication terminal of the car owner.

7. The method for preventing automobile theft based on automobile monitoring image recognition according to claim 6, characterized in that: The media information is one of video, voice or electronic photo.

8. The method for preventing automobile theft based on automobile monitoring image recognition according to claim 1, characterized in that: In the anti-theft mode stage, the vehicle end plays a warning message to warn the target.

9. The method for preventing automobile theft based on automobile monitoring image recognition according to claim 8, characterized in that: The steps of playing the warning message on the vehicle end: The communication terminal collects the warning audio of the vehicle owner and uploads it to the server; The warning audio is downloaded to the vehicle end for playing.

10. An automobile anti-theft system based on automobile monitoring image recognition, characterized in that: Used to execute the method described in any one of claims 1 to 9, the system includes a collection module, a processing module, an anti-theft module and a communication module; The acquisition module generates a monitoring area and performs image acquisition on the area to obtain multiple groups of monitoring images at the same time stamp; The processing module processes the monitoring image to identify the target action; if the target action is determined to be a car theft action, the anti-theft module is controlled to make the vehicle end enter the anti-theft mode, and the vehicle information is sent to the communication end of the vehicle owner through the communication module.

Images